Index of sw-kernelinferencelibrary-public-generic-local/5c190289e830f906b9bb3b9e7ce8feb53f1c96b6/fmha/trtllm-gen
Name Last modified Size
../
include/ 05-Jun-2026 22:35 -
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 101.32 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvCausalP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 105.01 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvCausalPagedMultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 124.66 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvCausalPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 128.85 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvCausalPagedMultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 140.72 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvCausalPagedMultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 115.45 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 131.42 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 118.73 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 94.84 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvCausalPagedMultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 104.48 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 118.40 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 97.68 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvCausalPagedVarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 108.71 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvCausalPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.09 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvCausalPagedVarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 138.50 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvCausalPagedVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 123.00 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvCausalPagedVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 88.20 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 117.98 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 104.08 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 101.88 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 101.26 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 85.36 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 115.05 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 79.52 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 142.03 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 126.88 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 145.02 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 123.70 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 114.01 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 115.45 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 103.90 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 118.43 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 106.98 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvDensePagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 115.16 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvDensePagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 103.88 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 118.27 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 106.96 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvDenseStaticTokenSparseP1MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 164.70 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvDenseStaticTokenSparseP1MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 118.42 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvDenseStaticTokenSparseP1MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 127.40 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvDenseStaticTokenSparseP1MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 135.52 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvDenseStaticTokenSparseP1MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 114.22 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvDenseStaticTokenSparseP1MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 152.23 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvDenseStaticTokenSparseP1MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 115.88 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvDenseStaticTokenSparseP1MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 124.48 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvDenseStaticTokenSparseP1MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 124.75 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvDenseStaticTokenSparseP1VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 140.95 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvDenseStaticTokenSparseP1VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 116.57 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvDenseStaticTokenSparseP1VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 125.50 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvDenseStaticTokenSparseP1VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 111.63 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvDenseStaticTokenSparseP1VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 124.70 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvDenseStaticTokenSparseP1VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 110.82 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvDenseStaticTokenSparseP1VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 111.60 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvDenseStaticTokenSparseP1VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 100.17 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 153.99 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 154.52 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 167.75 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 142.62 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 147.45 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 156.24 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 112.59 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 123.11 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 144.08 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 107.18 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 140.88 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 151.99 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 162.84 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 131.40 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 146.04 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 142.99 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 110.26 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.46 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 153.11 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 140.29 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 133.39 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 122.78 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 149.34 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 133.88 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 165.80 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 148.76 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 115.47 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 140.34 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 125.19 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 155.84 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 139.65 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 136.20 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 126.84 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 114.48 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 99.05 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 111.29 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 129.64 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 117.63 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 106.73 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 92.76 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 138.91 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 140.46 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 151.74 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 127.42 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 133.51 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 141.47 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 104.66 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 115.18 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 129.41 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 99.12 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 125.67 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 137.48 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 146.78 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 116.33 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 128.23 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 102.33 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 111.33 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 118.44 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 134.46 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 117.63 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 107.66 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 136.97 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 121.15 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 150.73 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 135.51 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 98.77 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 128.14 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 110.66 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 137.31 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 126.55 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 120.45 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 111.33 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 108.33 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 90.95 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 123.49 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 103.19 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 116.48 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 101.60 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 100.41 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.66 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 131.84 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 132.95 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 144.90 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 120.34 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 125.87 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 135.43 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 101.90 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 112.41 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 123.37 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 96.35 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 118.59 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 130.36 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 140.01 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 109.25 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 124.41 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 122.19 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 99.56 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 108.56 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 112.40 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.76 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 126.41 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 116.14 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 111.69 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 101.72 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 129.44 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 113.80 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 142.95 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 128.04 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 106.99 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 92.70 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 120.43 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 103.79 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 129.25 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 119.80 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 114.62 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 105.73 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 104.12 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 88.18 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 118.94 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 100.41 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 110.01 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 96.13 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 96.38 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 81.89 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 132.77 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 133.67 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 145.68 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 121.41 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 126.59 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 136.62 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 104.88 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 115.39 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 124.43 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 99.33 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 119.66 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 131.09 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 141.46 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 110.44 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 125.13 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 123.25 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 102.41 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 111.41 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 113.45 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 97.73 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 126.96 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 116.77 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 113.02 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 103.15 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 129.62 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 114.39 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 143.80 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 128.22 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 107.94 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 93.99 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 118.01 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 106.24 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 129.80 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 120.52 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 115.95 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 106.98 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 106.07 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.10 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 120.64 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 102.09 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 111.16 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 97.28 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 96.23 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 83.81 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256SeparateQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 100.93 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256SeparateQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 86.91 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256SeparateQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 104.02 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256SeparateQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 89.84 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256SeparateQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 101.00 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256SeparateQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 89.77 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 103.99 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H256SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 92.85 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta128PagedKvDenseDynamicTokenSparseP1MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 160.07 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta128PagedKvDenseDynamicTokenSparseP1MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 167.20 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta128PagedKvDenseDynamicTokenSparseP1MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 154.99 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta128PagedKvDenseDynamicTokenSparseP1MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 154.63 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta128PagedKvDenseDynamicTokenSparseP1MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 158.06 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta128PagedKvDenseDynamicTokenSparseP1MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 164.47 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta128PagedKvDenseDynamicTokenSparseP1MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 153.87 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta128PagedKvDenseDynamicTokenSparseP1VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 134.27 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta128PagedKvDenseDynamicTokenSparseP1VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 120.46 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta128PagedKvDenseDynamicTokenSparseP1VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 141.36 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta128PagedKvDenseDynamicTokenSparseP1VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 127.38 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta128PagedKvDenseDynamicTokenSparseP1VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 137.20 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta128PagedKvDenseDynamicTokenSparseP1VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 124.79 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta128PagedKvDenseDynamicTokenSparseP1VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 129.75 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta128PagedKvDenseDynamicTokenSparseP1VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 118.42 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 109.07 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 97.46 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 111.78 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 100.60 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 108.76 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 100.62 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 111.75 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 103.45 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP16MultiCtasKvGmemSepVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 153.45 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP16MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 154.27 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP16MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 154.56 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP16MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 152.59 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 159.38 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 150.50 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 150.13 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 144.63 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 153.45 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 145.79 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 162.38 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 153.45 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 152.94 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 144.11 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 151.07 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 140.20 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP32MultiCtasKvGmemSepVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 128.65 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP32MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 125.38 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP32MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 133.66 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP32MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 127.71 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 136.78 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 131.23 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 123.67 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 117.93 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 130.17 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 117.16 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 139.65 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 135.34 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 126.48 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 126.54 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 131.45 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 118.56 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP64MultiCtasKvGmemSepVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 115.78 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP64MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 112.55 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP64MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 122.67 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP64MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 115.18 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 123.13 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 117.75 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 112.40 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 106.27 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 116.77 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 103.55 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 126.00 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 124.52 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 115.34 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 113.37 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 119.66 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.63 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvDenseDynamicTokenSparseP1MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 184.50 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvDenseDynamicTokenSparseP1MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 193.42 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvDenseDynamicTokenSparseP1MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 179.75 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvDenseDynamicTokenSparseP1MultiCtasKvGmemSepVarSeqQ64Kv128Static2CtaKeepsAbForGen.cubin 05-Jun-2026 22:36 170.79 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvDenseDynamicTokenSparseP1MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 182.36 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvDenseDynamicTokenSparseP1MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 182.45 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvDenseDynamicTokenSparseP1MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 190.78 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvDenseDynamicTokenSparseP1MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 178.31 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvDenseDynamicTokenSparseP1VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 161.10 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvDenseDynamicTokenSparseP1VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 146.92 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvDenseDynamicTokenSparseP1VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 170.01 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvDenseDynamicTokenSparseP1VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 155.65 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvDenseDynamicTokenSparseP1VarSeqQ64Kv128Persistent2CtaKeepsAbForGen.cubin 05-Jun-2026 22:35 173.85 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvDenseDynamicTokenSparseP1VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 167.54 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvDenseDynamicTokenSparseP1VarSeqQ64Kv128Static2CtaKeepsAbForGen.cubin 05-Jun-2026 22:35 149.54 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvDenseDynamicTokenSparseP1VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 154.29 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvDenseDynamicTokenSparseP1VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 155.68 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvDenseDynamicTokenSparseP1VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 145.41 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 158.41 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 149.22 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 161.27 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 152.18 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 138.30 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 128.65 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 141.17 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 131.73 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 124.12 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 115.74 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 126.99 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 118.71 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP16MultiCtasKvGmemSepVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 161.01 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP16MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 159.66 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP16MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 159.07 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP16MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 157.98 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 159.16 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 151.99 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 158.59 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 150.67 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 156.13 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 148.59 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 156.21 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 144.35 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP32MultiCtasKvGmemSepVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 136.20 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP32MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 130.77 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP32MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 138.05 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP32MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 133.10 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 139.27 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 125.29 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 135.31 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 121.84 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 139.83 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 131.18 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 136.59 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 123.14 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP64MultiCtasKvGmemSepVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 123.21 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP64MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 117.94 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP64MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 127.05 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP64MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 120.57 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 125.11 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 113.73 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 121.77 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 108.74 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 129.02 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 117.88 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 124.79 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 111.70 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 231.33 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 222.58 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 141.95 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 135.24 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 226.73 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 240.55 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 219.57 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 139.53 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 149.67 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 134.20 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 201.79 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 191.79 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 223.32 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 216.49 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 190.55 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 184.59 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 140.68 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 125.38 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 159.09 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 139.78 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 129.73 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 117.84 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 192.61 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 183.98 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 128.74 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 121.96 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 188.27 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 203.77 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 181.12 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 125.00 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 135.27 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 119.70 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 176.23 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 166.67 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 200.88 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 188.15 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 163.98 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 155.32 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 127.01 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 112.37 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 147.98 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 126.66 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 116.30 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 104.85 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 177.52 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 168.89 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 121.98 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 115.34 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 173.19 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 188.13 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 166.03 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 118.45 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 128.72 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 113.15 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 162.58 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 149.82 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 184.70 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 171.34 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 149.19 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 138.55 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 120.39 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 105.73 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 140.73 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 120.03 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 109.68 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvCausalP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 98.22 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvDenseDynamicTokenSparseP1MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 210.52 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvDenseDynamicTokenSparseP1MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 203.65 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvDenseDynamicTokenSparseP1MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 211.88 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvDenseDynamicTokenSparseP1MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 316.70 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvDenseDynamicTokenSparseP1MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 327.18 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvDenseDynamicTokenSparseP1MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 311.36 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvDenseDynamicTokenSparseP1VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 184.05 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvDenseDynamicTokenSparseP1VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 170.99 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvDenseDynamicTokenSparseP1VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 195.23 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvDenseDynamicTokenSparseP1VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 182.02 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvDenseDynamicTokenSparseP1VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 191.66 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvDenseDynamicTokenSparseP1VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 182.91 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvDenseDynamicTokenSparseP1VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 173.10 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvDenseDynamicTokenSparseP1VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 165.88 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 235.73 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 226.59 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 146.03 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 139.07 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 231.12 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 245.17 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 223.84 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 143.61 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 154.38 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 138.03 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 206.02 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 196.05 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 228.94 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 220.48 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 195.05 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 188.04 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 143.15 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 129.70 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 164.50 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 144.87 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 133.99 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 121.67 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 197.12 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 187.99 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 132.93 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 125.79 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 192.67 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 208.16 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 185.26 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 129.21 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 140.11 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 123.55 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 181.30 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 170.95 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 205.00 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 192.11 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 170.34 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 158.88 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 132.09 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 116.56 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 153.41 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 131.88 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 122.91 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 108.65 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 182.04 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 172.91 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 126.19 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 119.18 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 177.45 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 192.76 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 170.17 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 122.65 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 133.43 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 116.99 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 166.38 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 154.09 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 188.95 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 174.98 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 155.42 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 142.11 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 124.98 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 109.93 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 146.17 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 125.23 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 115.67 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H512PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.02 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 100.80 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 88.71 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 106.77 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 94.86 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 99.95 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 88.45 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 106.12 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 94.55 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 111.34 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 99.52 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 117.22 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 105.84 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 121.64 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 107.62 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 117.50 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP16MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 111.20 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.73 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 132.00 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 84.29 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 92.82 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 113.82 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 81.30 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 113.36 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 106.75 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 115.69 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP16MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 103.23 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 102.74 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 122.40 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 83.42 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 90.91 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 106.09 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 80.31 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 116.57 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 112.92 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 104.53 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 99.61 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 106.95 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 94.52 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 119.09 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.90 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 101.32 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 86.65 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 101.70 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 89.61 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 123.17 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 120.63 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 111.20 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 109.17 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 85.09 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 72.70 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 97.27 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 81.47 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 105.28 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 90.48 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 80.72 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 68.60 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 118.83 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 103.27 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 113.16 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP32MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 108.27 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 99.52 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 129.31 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.00 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 91.41 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 111.38 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.90 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 110.55 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 102.41 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 111.47 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP32MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 100.41 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 98.41 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 119.59 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 82.01 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 89.50 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 103.66 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 78.91 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 114.22 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 110.06 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 102.25 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 96.77 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 102.41 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.23 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 115.36 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 102.80 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 98.99 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 83.94 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 97.16 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 85.30 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 120.06 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 118.01 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 108.41 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 106.34 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 83.66 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 71.30 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 95.58 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 80.19 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 102.83 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 87.65 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 79.42 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 67.20 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 117.29 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 101.10 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 110.98 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP64MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 106.73 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 97.23 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 127.78 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 82.12 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.65 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 109.98 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 79.14 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 109.02 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 100.23 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 109.30 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP64MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 98.88 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 96.23 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 118.17 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.26 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 88.74 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 102.12 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 78.15 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 112.67 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 108.52 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 100.73 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 95.24 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 100.10 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 88.05 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 113.17 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 100.51 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 97.45 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 82.41 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 94.98 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 83.13 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 118.52 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 116.48 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 107.01 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 104.80 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 82.91 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 70.52 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 94.48 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 79.41 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 101.29 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 86.11 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 78.65 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 66.42 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalPagedMultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 120.81 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 104.20 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalPagedMultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 113.95 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalPagedMultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 110.52 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalPagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.20 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 130.77 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 84.76 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.16 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 113.24 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 81.78 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalPagedMultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 112.41 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalPagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 103.09 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalPagedMultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 112.27 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalPagedMultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 102.66 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalPagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 99.38 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 121.04 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 83.77 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 91.38 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 105.44 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 80.79 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 114.10 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalPagedVarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 110.52 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 101.88 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalPagedVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 97.14 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalPagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 102.16 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 90.46 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalPagedVarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 115.52 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalPagedVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 103.15 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalPagedVarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 99.70 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalPagedVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 84.66 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalPagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 97.16 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalPagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 85.41 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 119.95 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 118.59 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 108.04 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 106.56 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 88.04 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 72.73 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 97.44 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 81.49 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 103.40 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 88.37 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 83.09 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 68.62 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 124.90 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 105.97 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 115.87 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 102.48 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 132.72 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 116.45 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 105.10 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 114.32 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 101.37 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 123.11 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 111.62 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 98.70 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 104.24 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.24 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 114.43 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 100.30 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 100.17 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 88.36 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 116.97 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 105.83 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 122.08 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 102.51 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 112.28 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 99.02 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 130.03 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 113.63 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 101.52 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 110.86 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 97.91 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 120.30 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 108.63 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 95.89 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 99.71 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.88 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 110.86 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 96.05 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 95.64 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 83.99 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 114.23 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 102.98 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 120.55 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 100.34 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 110.11 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 96.73 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 128.49 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 112.10 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 99.34 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 108.70 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 95.73 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 118.77 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 107.09 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 94.48 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 97.40 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 85.70 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 108.55 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 93.76 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 93.33 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.82 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 112.69 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 101.48 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomPagedMultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 123.82 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 103.45 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomPagedMultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 113.34 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomPagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 99.84 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 131.37 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomPagedMultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 115.50 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomPagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 102.33 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomPagedMultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 111.67 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomPagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 98.84 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomPagedMultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 121.62 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomPagedVarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 109.45 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomPagedVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 96.22 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomPagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 99.45 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 88.05 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomPagedVarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 110.78 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomPagedVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 96.10 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomPagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 95.51 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomPagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.04 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomPagedVarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 114.77 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvCustomPagedVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 103.19 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 114.88 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 104.41 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 122.59 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 110.95 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 112.68 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 102.12 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 119.48 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 108.16 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 111.11 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 100.60 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 117.93 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 106.77 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvDensePagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 112.42 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvDensePagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 101.66 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 119.38 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 107.95 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvDenseStaticTokenSparseP1MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 238.23 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvDenseStaticTokenSparseP1MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.72 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvDenseStaticTokenSparseP1MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 92.16 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvDenseStaticTokenSparseP1MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 96.88 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvDenseStaticTokenSparseP1MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 83.17 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvDenseStaticTokenSparseP1MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 230.21 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvDenseStaticTokenSparseP1MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 83.39 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvDenseStaticTokenSparseP1MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 90.66 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvDenseStaticTokenSparseP1MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 89.08 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvDenseStaticTokenSparseP1MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 81.62 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvDenseStaticTokenSparseP1VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 229.73 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvDenseStaticTokenSparseP1VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 215.96 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvDenseStaticTokenSparseP1VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 84.98 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvDenseStaticTokenSparseP1VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 72.59 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvDenseStaticTokenSparseP1VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 91.09 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvDenseStaticTokenSparseP1VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 78.39 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvDenseStaticTokenSparseP1VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 87.02 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvDenseStaticTokenSparseP1VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 75.74 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvDenseStaticTokenSparseP1VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 82.66 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvDenseStaticTokenSparseP1VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 69.74 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 129.16 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 112.31 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 122.86 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 117.65 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 108.07 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 137.04 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.37 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.40 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 120.05 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.02 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 120.76 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 111.45 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 121.05 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 109.92 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.08 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 127.31 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 87.50 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 95.61 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 112.32 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 84.03 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 128.37 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 120.07 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 116.12 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 106.65 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 110.62 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 98.27 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 123.98 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 111.97 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 106.81 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 91.96 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 105.20 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 93.11 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 134.73 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 125.43 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 122.51 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 113.62 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 92.27 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 76.83 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 102.55 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 86.38 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 110.65 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 95.66 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 86.95 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 72.40 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 126.34 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 107.98 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 118.52 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 114.70 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 103.85 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 134.26 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 86.95 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 95.98 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 117.73 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 83.61 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 117.94 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 107.11 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 116.84 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 106.98 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 102.86 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 124.53 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 86.09 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 94.20 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 110.01 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 82.62 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 125.88 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 117.21 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 113.84 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 103.81 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 106.12 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.15 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 120.24 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 107.72 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 104.36 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 89.12 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 100.69 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 88.99 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 131.62 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 122.80 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 119.73 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 110.91 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 90.70 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 75.41 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 100.86 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 85.09 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 108.20 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 92.95 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 85.38 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 70.98 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 124.80 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 105.80 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 116.35 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 113.17 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.55 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 132.85 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 86.20 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 95.23 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 116.20 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.85 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 116.41 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 104.94 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 114.66 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 105.45 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 100.57 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 122.99 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 85.33 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 93.45 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 108.48 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 81.86 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 124.34 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 115.66 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 112.45 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 102.41 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 103.77 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 91.85 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 117.93 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 105.42 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 102.94 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 87.71 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 98.25 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 86.70 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 130.07 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 121.28 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 118.33 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 109.38 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 89.61 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 74.65 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 99.77 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 84.33 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 106.65 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 91.41 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 84.27 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 70.22 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 128.66 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 109.73 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 120.62 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 117.22 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 105.61 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 136.01 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 89.48 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 98.52 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 119.59 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 86.02 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 120.26 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 108.73 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 118.81 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 109.36 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 104.74 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 126.40 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 88.49 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 96.61 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 111.87 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 85.15 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 126.52 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 118.74 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 114.58 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 104.98 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 107.22 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.41 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 121.66 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 108.88 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 106.15 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 93.56 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 101.81 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.27 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 132.25 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 124.13 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 120.33 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 112.06 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 92.95 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 77.49 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:43 103.37 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 86.92 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 109.16 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 97.33 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 87.76 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 72.94 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H80PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 104.11 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H80PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 88.39 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H80PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 109.85 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H80PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 94.80 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H80PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 104.39 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H80PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 88.13 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H80PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 110.16 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H80PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 94.50 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H80PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 114.77 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H80PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 99.32 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H80PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 120.43 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H80PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 105.84 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H80PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 125.79 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H80PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 110.05 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H80PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 131.42 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H80PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 116.53 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H80PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 119.99 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H80PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 104.67 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H80PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 125.62 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H80PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 111.27 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H80PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 116.76 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H80PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 101.16 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H80PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 122.38 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H80PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 107.66 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H80PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 126.10 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H80PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 109.78 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H80PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 132.09 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H80PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 116.13 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H80PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 119.59 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H80PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 103.99 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H80PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 125.46 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H80PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 110.37 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H80PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 116.48 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H80PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 101.04 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H80PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 122.48 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H80PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 107.41 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H80PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 137.33 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H80PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 121.79 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H80PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 142.85 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H80PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 127.80 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H80PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 131.54 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H80PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 116.41 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H80PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 137.17 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H80PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 122.42 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H80PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 128.30 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H80PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 112.78 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H80PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 133.80 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16H80PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 118.80 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk128HV64SeparateQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 107.73 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk128HV64SeparateQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 96.38 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk128HV64SeparateQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 113.46 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk128HV64SeparateQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 102.14 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk128HV64SeparateQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 107.53 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk128HV64SeparateQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 96.27 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk128HV64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 113.17 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk128HV64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 101.84 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk192HV128SeparateQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 111.20 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk192HV128SeparateQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 98.27 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk192HV128SeparateQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 116.20 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk192HV128SeparateQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 103.07 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk192HV128SeparateQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 110.37 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk192HV128SeparateQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 99.84 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk192HV128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 115.40 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk192HV128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 104.74 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 160.45 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 161.43 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 174.46 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP16MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 147.53 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 155.24 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 160.01 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 116.82 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 126.14 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 146.58 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 112.00 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 147.37 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 159.11 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 171.47 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP16MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 136.78 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 153.64 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 146.93 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 114.60 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 122.46 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP16MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 135.70 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 110.55 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 145.73 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 134.28 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 159.92 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 147.09 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 171.17 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 159.18 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 137.40 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 124.11 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 151.58 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 141.26 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 141.89 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 134.12 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 114.34 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 101.87 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 125.28 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 110.20 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 132.41 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 122.80 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 107.26 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 97.50 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 141.16 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 145.08 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 155.67 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP32MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 128.38 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 138.89 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 143.30 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.67 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 116.12 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 129.83 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 101.98 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 128.09 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 142.50 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 152.08 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP32MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 117.51 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 137.03 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 130.22 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 104.26 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 112.00 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP32MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 118.95 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 100.09 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 125.95 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 115.23 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 143.26 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 127.55 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 154.04 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 139.77 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 116.38 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 104.79 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 132.79 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 121.84 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 128.17 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 117.24 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 106.17 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 91.44 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 117.23 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 99.77 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 117.48 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 105.89 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 98.43 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 87.06 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 132.14 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 135.34 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 146.70 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP64MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 119.36 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 129.16 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 135.73 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 102.05 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 111.49 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 122.27 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 97.35 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 119.07 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 133.18 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 143.12 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP64MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 108.48 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 127.71 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 122.66 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 99.95 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 107.69 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP64MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 111.39 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 95.78 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 116.33 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 105.93 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 133.56 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 118.80 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 144.38 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 130.60 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 107.14 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 95.41 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 123.72 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 113.10 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 119.99 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 109.55 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 100.85 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 86.81 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 111.91 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 95.02 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 109.86 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 98.33 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 93.35 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvDenseP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.45 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 171.38 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 166.12 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 178.08 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 155.13 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 159.55 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 171.07 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 120.38 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 130.08 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 154.19 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 115.44 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 158.25 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 163.69 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 175.09 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 144.26 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 158.08 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 157.94 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 118.28 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 126.27 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 143.31 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 113.86 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 155.77 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 145.45 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 162.33 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 148.73 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 174.82 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 160.66 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 144.64 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 130.73 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 154.23 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 142.67 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 152.74 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 147.43 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 118.44 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 102.89 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 129.91 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 111.46 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 139.80 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 133.04 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 112.70 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 98.20 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 152.23 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 148.70 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 159.28 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 135.98 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 142.11 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 154.36 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 110.24 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 120.07 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 137.45 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 105.43 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 139.09 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 145.77 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 155.56 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 125.10 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 140.16 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 141.09 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 107.84 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 115.83 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 126.57 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.54 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 136.77 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 124.16 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 145.77 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 128.77 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 157.30 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 140.70 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 123.61 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 110.43 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 135.45 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 122.71 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 138.73 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 126.66 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 109.70 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 93.84 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 121.16 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 102.41 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 124.86 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 112.24 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 104.09 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 89.16 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 143.20 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 138.70 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 150.30 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 126.95 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 132.25 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 146.66 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 105.61 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 115.30 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 129.88 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 100.66 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 130.07 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 136.42 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 146.74 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 116.08 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 130.93 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 133.53 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 103.52 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 111.50 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 119.01 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 99.09 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 126.88 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 115.21 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 136.29 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 119.91 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 148.00 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 131.41 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 114.24 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 101.30 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 126.48 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 113.84 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 130.55 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 118.84 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 104.52 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 89.22 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 115.98 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 97.79 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 117.27 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 104.27 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 98.90 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 84.53 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 185.34 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvCgaVarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 152.55 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 196.98 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvCgaVarSeqQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 163.63 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 177.98 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvCgaVarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 145.59 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 125.54 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 112.15 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 133.84 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 119.75 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 119.80 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 106.94 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 146.38 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 146.51 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 184.12 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvVarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 151.20 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 194.11 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvVarSeqQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 160.62 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 177.13 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvVarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 144.87 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 124.23 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 110.26 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 131.08 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 116.42 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 119.23 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 105.67 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 173.62 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 170.66 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqQ16Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 151.18 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 138.19 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 187.49 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 181.84 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqQ32Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 163.55 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 148.52 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 142.23 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 139.12 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 167.81 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 164.66 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqQ8Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 146.02 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 132.65 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 122.23 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 110.12 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqSkipsSoftmaxQ16Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 109.59 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 97.18 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 130.54 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 117.55 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqSkipsSoftmaxQ32Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 117.38 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqSkipsSoftmaxQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.91 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 142.38 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 139.35 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 117.69 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 106.27 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqSkipsSoftmaxQ8Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 105.75 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 93.61 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 158.52 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvCgaVarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 140.71 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 169.89 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvCgaVarSeqQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 151.66 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 150.91 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvCgaVarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 133.74 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 114.48 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 105.65 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 122.78 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 113.26 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 108.74 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 100.44 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 118.16 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 119.74 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 157.05 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvVarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 139.24 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 167.16 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvVarSeqQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 148.66 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 150.18 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvVarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 132.90 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 112.59 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 104.17 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 119.32 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.33 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 107.60 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 99.59 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 154.24 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 143.99 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqQ16Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 139.12 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 125.71 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 167.56 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 155.16 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqQ32Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 150.85 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 136.17 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 122.45 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 112.55 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 148.44 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 137.75 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqQ8Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 133.96 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 120.30 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 111.61 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 99.52 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqSkipsSoftmaxQ16Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 103.26 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.69 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 120.05 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 106.81 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqSkipsSoftmaxQ32Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 111.05 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqSkipsSoftmaxQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 97.41 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 121.80 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.27 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 107.10 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 95.64 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqSkipsSoftmaxQ8Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 99.41 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 87.13 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 146.52 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvCgaVarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 135.76 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 158.02 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvCgaVarSeqQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 146.59 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 139.03 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvCgaVarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 128.66 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 108.11 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 103.09 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 116.41 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 110.70 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 102.38 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 97.88 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.38 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 108.34 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 145.05 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvVarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 134.28 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 155.16 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvVarSeqQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 143.58 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 138.19 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvVarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 127.95 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 106.63 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 101.61 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 113.36 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 107.77 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.64 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 97.02 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 142.18 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 131.99 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqQ16Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 134.02 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 120.89 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 154.86 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 142.99 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqQ32Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 145.77 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 131.34 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 108.95 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 100.66 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 136.50 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 125.88 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqQ8Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 128.98 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 115.48 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 105.27 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 93.03 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqSkipsSoftmaxQ16Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 100.70 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 88.12 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 113.72 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 100.45 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqSkipsSoftmaxQ32Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 108.60 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqSkipsSoftmaxQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 94.85 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 110.81 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 102.73 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 100.88 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.29 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqSkipsSoftmaxQ8Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 96.85 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 84.57 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseStaticTokenSparseP1MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 137.82 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseStaticTokenSparseP1MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 145.09 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseStaticTokenSparseP1MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 133.13 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseStaticTokenSparseP1MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 132.91 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseStaticTokenSparseP1MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 135.84 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseStaticTokenSparseP1MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 142.52 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseStaticTokenSparseP1MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 131.65 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseStaticTokenSparseP1VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 135.91 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseStaticTokenSparseP1VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 123.21 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseStaticTokenSparseP1VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 144.40 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseStaticTokenSparseP1VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 130.00 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseStaticTokenSparseP1VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 139.67 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseStaticTokenSparseP1VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 127.44 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseStaticTokenSparseP1VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 131.21 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta128PagedKvDenseStaticTokenSparseP1VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 121.41 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 206.89 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvCgaVarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 165.91 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 225.62 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvCgaVarSeqQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 185.65 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 200.08 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvCgaVarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 159.87 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 133.70 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 119.02 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 143.14 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 127.77 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 128.88 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 114.72 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvGmemSepVarSeqQ64Kv128Static2CtaKeepsAbForGen.cubin 05-Jun-2026 22:35 149.75 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 164.80 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128Static2CtaKeepsAbForGen.cubin 05-Jun-2026 22:35 153.77 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 164.00 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 204.66 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvVarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 163.79 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 220.29 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvVarSeqQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 178.61 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 198.80 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvVarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 158.34 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 131.91 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 116.70 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 139.65 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 123.75 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 127.73 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 113.05 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 194.73 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 191.58 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqQ16Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 163.80 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 150.44 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 211.32 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 208.12 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqQ32Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 179.80 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 169.31 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqQ64Kv128Persistent2CtaKeepsAbForGen.cubin 05-Jun-2026 22:35 159.95 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 162.53 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqQ64Kv128Static2CtaKeepsAbForGen.cubin 05-Jun-2026 22:35 143.62 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 157.69 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 187.82 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 184.83 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqQ8Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 157.41 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 144.38 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 130.40 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 117.80 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ16Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 116.05 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 103.58 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 143.49 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 126.12 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ32Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 128.34 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 111.20 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ64Kv128Persistent2CtaKeepsAbForGen.cubin 05-Jun-2026 22:35 164.02 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 158.51 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ64Kv128Static2CtaKeepsAbForGen.cubin 05-Jun-2026 22:35 147.62 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 158.20 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 125.42 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 113.40 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ8Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 111.77 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 99.56 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 174.95 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvCgaVarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 151.03 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 195.12 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvCgaVarSeqQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 170.78 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 168.27 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvCgaVarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 144.99 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 121.26 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 111.48 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 130.58 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 120.35 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 116.31 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 107.18 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvGmemSepVarSeqQ64Kv128Static2CtaKeepsAbForGen.cubin 05-Jun-2026 22:35 132.88 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 131.61 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128Static2CtaKeepsAbForGen.cubin 05-Jun-2026 22:35 136.91 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 134.41 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 172.84 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvVarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 148.91 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 188.23 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvVarSeqQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 163.73 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 166.86 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvVarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 143.46 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 118.80 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 109.54 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 126.41 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 116.59 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 114.63 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 105.89 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 169.77 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 159.62 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqQ16Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 149.02 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 135.62 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 186.70 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 176.03 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqQ32Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 164.47 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 154.51 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqQ64Kv128Persistent2CtaKeepsAbForGen.cubin 05-Jun-2026 22:35 142.20 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 137.61 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqQ64Kv128Static2CtaKeepsAbForGen.cubin 05-Jun-2026 22:35 127.03 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 125.93 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 162.98 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 152.87 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqQ8Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 142.68 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 129.57 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 117.82 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 105.88 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ16Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 108.91 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 96.29 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 131.05 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 114.07 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ32Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 121.21 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 103.91 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ64Kv128Persistent2CtaKeepsAbForGen.cubin 05-Jun-2026 22:35 146.12 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 138.54 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ64Kv128Static2CtaKeepsAbForGen.cubin 05-Jun-2026 22:35 130.89 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 128.44 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 113.13 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 101.30 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ8Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 104.65 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 92.27 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 159.84 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvCgaVarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 145.30 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 180.02 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvCgaVarSeqQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 165.05 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 153.16 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvCgaVarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 139.27 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 114.10 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 108.60 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 123.42 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 117.35 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 109.16 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.31 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvGmemSepVarSeqQ64Kv128Static2CtaKeepsAbForGen.cubin 05-Jun-2026 22:35 125.22 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 117.05 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128Static2CtaKeepsAbForGen.cubin 05-Jun-2026 22:36 129.24 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 119.80 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 157.61 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvVarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 143.20 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 173.25 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvVarSeqQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 157.89 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 151.75 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvVarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 137.73 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 111.91 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 106.54 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 119.66 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 113.59 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 107.87 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 102.90 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 154.99 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 144.67 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqQ16Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 143.12 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 129.91 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 171.48 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 164.22 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqQ32Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 158.62 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 148.79 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqQ64Kv128Persistent2CtaKeepsAbForGen.cubin 05-Jun-2026 22:35 134.10 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 121.73 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqQ64Kv128Static2CtaKeepsAbForGen.cubin 05-Jun-2026 22:35 119.37 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 111.23 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 148.25 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 138.05 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqQ8Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 136.93 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 123.85 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 110.71 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 98.62 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ16Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 106.03 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 93.41 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 123.83 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 106.94 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ32Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 118.33 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 101.03 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ64Kv128Persistent2CtaKeepsAbForGen.cubin 05-Jun-2026 22:35 138.14 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 124.59 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ64Kv128Static2CtaKeepsAbForGen.cubin 05-Jun-2026 22:35 123.36 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 113.99 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 106.02 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.16 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ8Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 101.76 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 89.27 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseStaticTokenSparseP1MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 154.51 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseStaticTokenSparseP1MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 163.42 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseStaticTokenSparseP1MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 150.30 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseStaticTokenSparseP1MultiCtasKvGmemSepVarSeqQ64Kv128Static2CtaKeepsAbForGen.cubin 05-Jun-2026 22:35 145.65 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseStaticTokenSparseP1MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 152.08 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseStaticTokenSparseP1MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 152.54 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseStaticTokenSparseP1MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 160.78 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseStaticTokenSparseP1MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 149.09 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseStaticTokenSparseP1VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 150.59 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseStaticTokenSparseP1VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 138.98 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseStaticTokenSparseP1VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 162.78 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseStaticTokenSparseP1VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 147.41 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseStaticTokenSparseP1VarSeqQ64Kv128Persistent2CtaKeepsAbForGen.cubin 05-Jun-2026 22:35 169.57 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseStaticTokenSparseP1VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 160.55 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseStaticTokenSparseP1VarSeqQ64Kv128Static2CtaKeepsAbForGen.cubin 05-Jun-2026 22:35 139.91 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseStaticTokenSparseP1VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 146.31 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseStaticTokenSparseP1VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 146.24 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512HVPerCta256PagedKvDenseStaticTokenSparseP1VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 137.53 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 246.99 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16MultiCtasKvCgaVarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 188.91 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 238.36 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16MultiCtasKvCgaVarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 180.84 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 149.88 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 131.86 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 143.56 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 125.82 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 196.96 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 197.34 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 243.63 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16MultiCtasKvVarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 185.43 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 266.02 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16MultiCtasKvVarSeqQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 207.38 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 236.06 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16MultiCtasKvVarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 178.54 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 146.84 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 128.34 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 156.48 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 137.41 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 141.91 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 123.94 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 229.91 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 230.63 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16VarSeqQ16Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 187.54 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16VarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 173.00 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 257.55 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 253.69 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16VarSeqQ32Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 210.89 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16VarSeqQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 195.62 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 195.05 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 191.80 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 220.30 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 222.62 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16VarSeqQ8Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 178.62 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16VarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 165.67 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 146.04 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 132.84 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16VarSeqSkipsSoftmaxQ16Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 129.21 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16VarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 115.19 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 159.94 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 142.95 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16VarSeqSkipsSoftmaxQ32Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 142.44 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16VarSeqSkipsSoftmaxQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 124.86 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 183.83 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 191.40 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 138.76 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 127.59 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16VarSeqSkipsSoftmaxQ8Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 122.62 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP16VarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 110.45 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 203.82 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32MultiCtasKvCgaVarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 171.03 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 195.06 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32MultiCtasKvCgaVarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 162.96 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 134.52 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 123.51 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 128.20 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 117.60 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 155.33 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 158.97 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 200.21 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32MultiCtasKvVarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 167.55 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 222.60 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32MultiCtasKvVarSeqQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 189.49 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 192.88 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32MultiCtasKvVarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 160.66 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 130.94 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 120.09 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 140.58 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 129.29 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 125.88 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 115.80 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 198.46 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 187.59 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32VarSeqQ16Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 169.00 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32VarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 154.21 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 222.40 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 210.88 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32VarSeqQ32Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 191.52 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32VarSeqQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 176.73 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 160.98 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 149.73 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 188.85 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 179.33 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32VarSeqQ8Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 160.20 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32VarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 147.02 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 131.17 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 118.14 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32VarSeqSkipsSoftmaxQ16Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 120.80 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32VarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.96 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 145.05 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 128.38 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32VarSeqSkipsSoftmaxQ32Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 133.91 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32VarSeqSkipsSoftmaxQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 116.34 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 158.34 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 152.88 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 123.86 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 112.85 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32VarSeqSkipsSoftmaxQ8Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 114.09 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP32VarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 102.36 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 185.88 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64MultiCtasKvCgaVarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 163.39 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 177.12 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64MultiCtasKvCgaVarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 155.31 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 126.30 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 119.84 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 119.98 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 113.93 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 137.27 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 141.25 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 182.27 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64MultiCtasKvVarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 160.03 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 205.02 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64MultiCtasKvVarSeqQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 181.74 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 174.95 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64MultiCtasKvVarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 153.14 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 123.00 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 116.41 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 132.77 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 125.62 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 118.06 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 112.13 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 179.84 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 169.53 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64VarSeqQ16Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 161.45 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64VarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 146.70 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 203.04 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 192.95 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64VarSeqQ32Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 183.87 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64VarSeqQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 169.22 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 141.84 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 131.79 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 170.30 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 161.39 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64VarSeqQ8Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 152.53 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64VarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 139.51 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 122.95 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 109.91 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64VarSeqSkipsSoftmaxQ16Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 117.14 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64VarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 102.87 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 136.84 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 120.03 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64VarSeqSkipsSoftmaxQ32Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 130.26 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64VarSeqSkipsSoftmaxQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 112.54 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 144.94 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 135.39 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 115.81 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 104.62 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64VarSeqSkipsSoftmaxQ8Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 110.44 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseP64VarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 98.58 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseStaticTokenSparseP1MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 175.13 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseStaticTokenSparseP1MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 168.54 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseStaticTokenSparseP1MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 176.49 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseStaticTokenSparseP1MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 171.28 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseStaticTokenSparseP1MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 181.99 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseStaticTokenSparseP1MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 166.06 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseStaticTokenSparseP1VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 175.49 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseStaticTokenSparseP1VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 158.76 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseStaticTokenSparseP1VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 188.12 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseStaticTokenSparseP1VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 169.70 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseStaticTokenSparseP1VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 183.62 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseStaticTokenSparseP1VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 170.46 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseStaticTokenSparseP1VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 164.14 KB
fmhaSm100fKernel_QkvBfloat16OBfloat16HQk576HV512PagedKvDenseStaticTokenSparseP1VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 156.37 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 100.21 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 88.30 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 104.19 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 92.30 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 99.70 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PackedQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:35 107.20 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PackedQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:35 94.48 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PackedQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:35 113.26 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PackedQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:35 100.45 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PackedQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:35 109.15 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PackedQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:35 96.29 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 88.44 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 103.61 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:35 110.96 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:35 98.30 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:35 116.97 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:35 104.23 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:36 112.91 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:35 100.24 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 92.30 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 110.62 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 99.16 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 114.63 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 103.14 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 130.37 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 108.38 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 118.70 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP16MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 114.70 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 101.55 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 133.80 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 85.74 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 94.77 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 117.16 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 80.27 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 121.31 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 106.83 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 115.35 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP16MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 104.73 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 100.68 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 121.02 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 84.09 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 91.56 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 107.33 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 79.27 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 116.66 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 119.48 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 104.88 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 105.34 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 108.88 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 93.36 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 120.53 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 106.98 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 103.27 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 90.77 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 99.73 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 87.40 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 120.64 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 119.85 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 108.71 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 108.34 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 88.59 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 72.32 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 99.16 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 82.31 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 106.39 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 93.61 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 80.18 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 67.55 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 127.55 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.16 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 114.49 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP32MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 111.88 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 97.33 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 130.98 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 84.46 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 93.35 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 114.34 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 78.85 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 118.50 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 102.48 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 111.13 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP32MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 101.92 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 96.34 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 118.34 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 82.68 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 90.16 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 104.52 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 77.99 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 113.67 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 116.62 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 102.09 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 102.50 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 104.35 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 89.18 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 116.00 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 102.88 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 100.31 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 87.83 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 95.20 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 83.10 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 117.53 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 117.24 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 106.05 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 105.63 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 86.89 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 71.03 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 97.47 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 80.90 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 103.55 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 90.80 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 78.88 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 66.14 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 126.02 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 101.87 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 112.32 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP64MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 110.23 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.03 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 129.45 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.58 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 92.58 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 112.94 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 78.10 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 116.97 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 100.32 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 108.97 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP64MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 100.39 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 94.16 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 116.80 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 81.92 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 89.38 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 102.98 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 77.11 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 112.13 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 115.07 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 100.69 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 101.09 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 102.16 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 87.01 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:43 113.69 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 100.59 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 98.77 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 86.70 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 92.88 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 80.93 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 115.99 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 115.70 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 104.53 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 104.09 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 85.80 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 70.26 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 96.38 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 80.13 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 102.02 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 89.39 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 78.12 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 65.37 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalPagedMultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 129.54 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 104.84 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalPagedMultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 115.29 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalPagedMultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 114.13 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalPagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 98.00 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 132.28 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 86.22 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.25 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 116.20 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 80.61 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalPagedMultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 120.36 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalPagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 103.29 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalPagedMultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 112.06 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalPagedMultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 104.09 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalPagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 97.31 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 119.64 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 84.44 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 91.93 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 106.41 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 79.75 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 113.56 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalPagedVarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 117.08 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 101.56 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalPagedVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 102.75 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalPagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 104.39 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 89.41 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalPagedVarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 115.90 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalPagedVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 103.11 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalPagedVarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 100.89 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalPagedVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 88.81 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalPagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 95.06 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalPagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 83.46 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 117.41 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 117.81 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 105.41 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 105.48 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 88.63 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 72.46 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 99.33 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 82.11 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 104.26 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 91.66 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 82.41 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 67.57 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 134.85 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.63 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 116.66 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 100.91 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 137.58 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 125.80 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 106.25 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 114.48 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 99.91 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 125.64 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 118.09 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 104.36 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 105.96 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 90.96 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 114.74 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 102.57 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 98.07 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 86.17 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 118.71 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 107.23 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 132.04 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 103.41 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 112.44 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.56 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 134.77 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 122.98 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 102.04 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 110.27 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 95.57 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 122.83 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 115.11 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 101.54 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 101.43 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 86.72 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 110.21 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 98.07 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 93.54 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 81.80 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 115.85 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 104.52 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 130.51 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 101.12 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 110.15 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 94.39 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 133.23 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 121.45 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 99.74 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 108.09 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 93.40 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 121.42 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 113.57 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 100.01 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 99.12 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 84.42 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 107.90 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 95.90 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 91.23 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 79.63 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 114.31 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 103.02 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomPagedMultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 133.91 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 104.35 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomPagedMultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 113.38 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomPagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 97.38 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 136.22 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomPagedMultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 124.84 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomPagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 102.85 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomPagedMultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 111.08 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomPagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 96.63 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomPagedMultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 124.29 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomPagedVarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 115.67 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomPagedVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 101.74 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomPagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 101.34 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.77 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomPagedVarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 110.09 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomPagedVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 98.27 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomPagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 93.41 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomPagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 81.85 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomPagedVarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 116.41 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvCustomPagedVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 104.87 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 115.60 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 104.91 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 119.50 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 108.77 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 112.64 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 102.24 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 116.55 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 105.98 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 111.07 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 100.72 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 114.98 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 104.58 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvDensePagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 112.51 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvDensePagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 101.63 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 116.41 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 105.49 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvDenseStaticTokenSparseP1MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 215.09 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvDenseStaticTokenSparseP1MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 86.48 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvDenseStaticTokenSparseP1MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 94.14 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvDenseStaticTokenSparseP1MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 98.76 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvDenseStaticTokenSparseP1MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 82.15 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvDenseStaticTokenSparseP1MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 204.12 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvDenseStaticTokenSparseP1MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 84.59 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvDenseStaticTokenSparseP1MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 91.54 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvDenseStaticTokenSparseP1MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 89.80 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvDenseStaticTokenSparseP1MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 80.59 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvDenseStaticTokenSparseP1VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 204.32 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvDenseStaticTokenSparseP1VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 190.09 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvDenseStaticTokenSparseP1VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 84.90 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvDenseStaticTokenSparseP1VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 71.77 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvDenseStaticTokenSparseP1VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 92.56 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvDenseStaticTokenSparseP1VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 79.00 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvDenseStaticTokenSparseP1VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 89.26 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvDenseStaticTokenSparseP1VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 76.65 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvDenseStaticTokenSparseP1VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 82.40 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvDenseStaticTokenSparseP1VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 68.87 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 137.88 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 113.07 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 123.35 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 121.27 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 105.88 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 141.16 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 89.84 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 99.47 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 123.44 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 83.96 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 128.70 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 111.52 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 120.12 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 111.13 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 105.01 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 128.39 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 88.18 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 96.27 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 113.44 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 83.10 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 130.59 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 126.80 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 118.49 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 112.46 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 112.67 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 97.25 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 124.98 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 111.77 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 108.47 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 96.54 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 103.14 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 90.91 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 132.32 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 127.25 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 120.05 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 115.11 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 93.24 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 76.51 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 104.32 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 87.16 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 111.88 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 98.99 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 86.31 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 71.25 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 134.95 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 108.73 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 119.14 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 118.45 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 101.66 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 138.34 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.43 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 98.18 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 120.62 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 82.55 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 125.89 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 107.18 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 115.91 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 108.32 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 100.67 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 125.58 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 86.77 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 94.86 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 110.62 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 81.69 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 127.61 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 124.07 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 115.71 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 109.62 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 108.14 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 93.01 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 120.45 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 107.64 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 105.52 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 93.60 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 98.64 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 86.67 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 129.22 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 124.52 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 117.40 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 112.40 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 91.55 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 75.09 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 102.75 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 85.75 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 109.05 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 96.18 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 84.74 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 69.96 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 133.54 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 106.56 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 116.97 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 116.91 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 99.37 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 136.81 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 87.67 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 97.29 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 119.21 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 81.80 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 124.36 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 105.02 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 113.74 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 106.66 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 98.50 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 124.16 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 86.02 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 94.09 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 109.21 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 80.93 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 126.07 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 122.52 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 114.31 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 108.22 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 105.95 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 90.71 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 118.13 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 105.34 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 103.98 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 92.19 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 96.30 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 84.50 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 127.67 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 122.97 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 115.87 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 110.87 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 90.32 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 74.45 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 101.66 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 84.98 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 107.52 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 94.65 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 83.63 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 69.20 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 137.39 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 110.62 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 121.27 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 120.71 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 103.42 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 140.11 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 90.96 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 100.59 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 122.60 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 84.96 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 128.21 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 109.06 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 118.05 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 110.71 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 102.68 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 127.46 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 89.18 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 97.27 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 112.48 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 84.09 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 128.24 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 125.27 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 116.45 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 110.91 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 109.55 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 94.41 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 121.80 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.95 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 107.18 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 94.89 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 99.87 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 87.95 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 129.85 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 125.69 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 118.00 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 113.56 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 93.80 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 77.17 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 105.26 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 87.83 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 109.81 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 97.48 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 87.12 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 71.79 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128SeparateQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 100.09 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128SeparateQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 88.43 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128SeparateQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 104.05 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128SeparateQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 92.42 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128SeparateQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 99.82 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:35 107.33 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:35 94.48 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:35 113.26 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:35 100.58 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:35 109.15 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:35 96.41 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128SeparateQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 88.56 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 103.73 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:35 110.96 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:36 98.30 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:35 117.09 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:35 104.35 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:36 113.04 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:35 100.37 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 92.43 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 96.23 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 84.79 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 99.11 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 87.50 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 97.30 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 85.37 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 100.16 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 88.33 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 107.95 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 95.46 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 110.63 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 98.31 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 127.41 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 124.77 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 135.49 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP16MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 116.95 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 118.22 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 129.87 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 97.42 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 107.45 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 118.89 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 92.38 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.93 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 121.89 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 130.75 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP16MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 105.86 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 116.45 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 117.64 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 94.96 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 103.59 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 107.80 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 90.91 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 118.43 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 114.24 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 106.32 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 100.37 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 124.84 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 110.60 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 138.17 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 124.67 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 104.22 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 91.31 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 113.05 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 100.55 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 121.27 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 116.78 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 109.16 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 103.75 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 100.84 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 83.70 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 115.45 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 95.03 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 106.02 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 94.09 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 90.86 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 77.96 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 120.39 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 117.41 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 129.10 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP32MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 109.94 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 110.98 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 124.11 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 94.34 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.37 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 113.10 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 89.17 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 108.04 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 114.95 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 124.37 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP32MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 98.84 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 109.40 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 111.76 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 91.88 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 100.39 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 102.01 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 87.70 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 111.41 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 106.60 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 100.16 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 94.34 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 117.69 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 103.16 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 131.11 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 118.05 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 97.37 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 85.28 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 106.07 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 93.84 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 114.25 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 110.13 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 103.00 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 97.99 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 97.09 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 80.60 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 111.01 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 91.95 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 100.38 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 88.36 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 87.66 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 74.87 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 117.45 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 114.34 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 126.04 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP64MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 107.00 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 107.79 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 121.17 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 92.82 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 102.83 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 110.16 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 87.77 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 104.98 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 111.77 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 121.30 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP64MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 95.91 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 106.33 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 108.95 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 90.36 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 98.85 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 99.07 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.18 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 108.34 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 103.39 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 97.22 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 91.27 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 114.61 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 99.97 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 127.91 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 114.98 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 94.12 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 82.21 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 102.98 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 90.91 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 111.16 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 107.17 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 99.94 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 95.17 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 95.42 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 79.06 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 109.34 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 90.41 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 97.41 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 85.41 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 86.09 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 73.33 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalPagedMultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 119.47 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 116.49 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalPagedMultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 128.02 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalPagedMultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 108.88 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalPagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 110.06 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 123.44 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 95.62 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 105.63 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 112.20 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 90.59 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalPagedMultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 107.12 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalPagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 114.03 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalPagedMultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 123.28 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalPagedMultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 97.91 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalPagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 108.47 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 110.96 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 93.17 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 101.78 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 101.23 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 88.86 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 111.36 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalPagedVarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 104.78 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 98.96 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalPagedVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 92.98 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalPagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 116.09 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 102.10 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalPagedVarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 129.27 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalPagedVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 114.79 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalPagedVarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 96.02 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalPagedVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 83.89 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalPagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 104.38 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalPagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 92.41 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 114.30 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 108.55 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 101.68 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 96.97 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 97.59 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 80.96 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 111.51 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 92.43 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 99.37 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 87.23 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 90.48 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 75.23 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 121.20 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 107.38 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 124.20 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 110.21 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 112.93 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 100.96 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 116.05 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 103.80 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 109.85 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 98.02 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 112.84 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 100.85 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvDensePagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 111.49 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvDensePagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 99.65 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 114.36 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 102.37 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvDenseStaticTokenSparseP1MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 147.64 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvDenseStaticTokenSparseP1MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 102.23 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvDenseStaticTokenSparseP1MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 111.31 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvDenseStaticTokenSparseP1MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 119.00 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvDenseStaticTokenSparseP1MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.17 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvDenseStaticTokenSparseP1MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 135.23 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvDenseStaticTokenSparseP1MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 99.52 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvDenseStaticTokenSparseP1MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 108.51 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvDenseStaticTokenSparseP1MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 108.06 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvDenseStaticTokenSparseP1MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 96.78 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvDenseStaticTokenSparseP1VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 134.65 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvDenseStaticTokenSparseP1VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 121.48 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvDenseStaticTokenSparseP1VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 100.41 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvDenseStaticTokenSparseP1VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 86.89 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvDenseStaticTokenSparseP1VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 109.45 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvDenseStaticTokenSparseP1VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.28 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvDenseStaticTokenSparseP1VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 108.55 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvDenseStaticTokenSparseP1VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 94.70 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvDenseStaticTokenSparseP1VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 95.73 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvDenseStaticTokenSparseP1VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 83.55 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 134.99 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 129.30 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 140.14 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 122.08 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 122.48 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 137.45 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 101.52 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 112.16 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 124.25 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 96.08 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 122.52 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 126.44 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 135.25 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 111.10 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 120.62 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 125.23 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 99.05 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 108.30 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 113.15 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 94.61 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 134.36 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 121.71 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 118.34 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.45 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 129.24 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 114.84 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 143.12 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 127.48 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 109.23 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 97.12 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 120.52 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 103.98 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 135.45 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 123.67 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 121.14 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 111.51 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 105.60 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 87.95 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 120.76 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 100.16 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 111.14 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 99.71 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 98.08 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 81.66 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 127.98 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 122.23 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 133.76 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 115.06 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 115.42 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 131.70 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.33 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 109.06 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 118.46 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 93.02 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 115.62 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 119.65 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 128.73 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 104.22 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 113.84 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 119.35 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 95.87 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 105.09 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 107.37 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 91.42 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 126.44 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 113.77 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 111.69 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 101.80 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 121.77 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 107.52 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 135.53 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 121.32 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 102.35 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 90.48 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 113.20 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 97.40 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 127.38 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 117.43 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 114.38 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 105.45 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 101.49 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 84.85 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 116.33 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 97.09 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 105.38 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 93.68 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 94.03 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 78.57 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 125.04 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 119.04 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 130.69 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 112.12 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 112.23 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 128.76 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 96.79 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 107.53 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 115.52 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 91.48 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 112.69 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 116.58 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 125.67 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 101.15 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 110.77 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 116.53 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 94.45 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 103.68 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 104.42 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 89.88 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 123.23 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 110.56 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 108.75 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 98.87 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:43 118.68 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 104.46 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 132.34 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 118.26 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 99.13 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 87.53 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 110.00 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 94.46 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 124.30 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 114.34 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 111.44 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 102.60 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 99.83 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 83.30 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 114.79 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 95.55 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 102.54 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 90.74 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 92.49 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 77.02 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 127.65 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 121.30 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 133.13 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 114.72 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 114.62 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 131.52 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 100.25 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 111.01 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 118.02 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.94 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 115.42 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 118.97 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 129.04 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 103.74 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 113.03 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 119.17 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 97.91 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 107.16 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 107.05 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 93.34 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 125.57 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 112.73 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 111.05 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 101.03 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 120.66 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 106.16 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 134.78 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 120.09 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 101.76 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 89.57 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 108.86 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 98.31 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 126.65 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 116.62 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 113.62 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 104.86 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 102.39 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 85.73 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 117.37 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 97.95 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 105.09 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 93.00 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 93.06 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 79.58 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256SeparateQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 97.35 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256SeparateQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 84.91 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256SeparateQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 100.18 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256SeparateQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 87.75 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256SeparateQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 97.30 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256SeparateQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 85.49 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 100.29 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H256SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 88.33 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta128PagedKvDenseDynamicTokenSparseP1MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 124.05 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta128PagedKvDenseDynamicTokenSparseP1MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 131.55 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta128PagedKvDenseDynamicTokenSparseP1MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 119.38 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta128PagedKvDenseDynamicTokenSparseP1MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 119.09 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta128PagedKvDenseDynamicTokenSparseP1MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 121.94 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta128PagedKvDenseDynamicTokenSparseP1MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 129.06 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta128PagedKvDenseDynamicTokenSparseP1MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 117.99 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta128PagedKvDenseDynamicTokenSparseP1VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 110.65 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta128PagedKvDenseDynamicTokenSparseP1VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 96.64 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta128PagedKvDenseDynamicTokenSparseP1VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 117.84 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta128PagedKvDenseDynamicTokenSparseP1VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 103.45 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta128PagedKvDenseDynamicTokenSparseP1VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 114.29 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta128PagedKvDenseDynamicTokenSparseP1VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 100.99 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta128PagedKvDenseDynamicTokenSparseP1VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 106.55 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta128PagedKvDenseDynamicTokenSparseP1VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 93.91 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 105.15 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 92.80 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 108.15 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 95.50 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 106.34 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 93.41 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 109.20 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 96.38 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP16MultiCtasKvGmemSepVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 123.52 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP16MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 114.22 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP16MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 124.88 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP16MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 115.05 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 134.94 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 129.99 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 121.64 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 116.52 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 121.19 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 107.43 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 137.63 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 130.79 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 124.34 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 117.59 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 121.25 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 107.68 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP32MultiCtasKvGmemSepVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 111.77 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP32MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 102.45 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP32MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 115.10 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP32MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 105.37 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 122.85 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 118.03 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 110.88 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 105.01 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 109.45 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 95.91 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 125.56 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 121.30 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 113.58 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 108.64 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 112.21 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 98.85 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP64MultiCtasKvGmemSepVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 107.05 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP64MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 97.86 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP64MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 110.88 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP64MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 101.15 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 117.98 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 113.16 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 106.16 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 100.29 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 104.55 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 91.20 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 120.69 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 116.94 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 108.86 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 104.30 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 107.82 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 94.55 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvDenseDynamicTokenSparseP1MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 138.77 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvDenseDynamicTokenSparseP1MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 147.96 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvDenseDynamicTokenSparseP1MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 134.59 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvDenseDynamicTokenSparseP1MultiCtasKvGmemSepVarSeqQ64Kv128Static2CtaKeepsAbForGen.cubin 05-Jun-2026 22:35 118.05 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvDenseDynamicTokenSparseP1MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 121.82 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvDenseDynamicTokenSparseP1MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 136.70 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvDenseDynamicTokenSparseP1MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 145.20 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvDenseDynamicTokenSparseP1MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 133.25 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvDenseDynamicTokenSparseP1VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 125.20 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvDenseDynamicTokenSparseP1VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 110.43 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvDenseDynamicTokenSparseP1VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 133.86 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvDenseDynamicTokenSparseP1VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 119.10 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvDenseDynamicTokenSparseP1VarSeqQ64Kv128Persistent2CtaKeepsAbForGen.cubin 05-Jun-2026 22:36 119.89 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvDenseDynamicTokenSparseP1VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 118.00 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvDenseDynamicTokenSparseP1VarSeqQ64Kv128Static2CtaKeepsAbForGen.cubin 05-Jun-2026 22:35 104.99 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvDenseDynamicTokenSparseP1VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 104.19 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvDenseDynamicTokenSparseP1VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 120.30 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvDenseDynamicTokenSparseP1VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 107.62 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 136.56 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 122.70 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 139.55 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 125.66 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 124.53 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 111.72 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 127.52 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 114.68 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 119.66 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 107.00 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 122.65 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 109.96 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP16MultiCtasKvGmemSepVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 131.11 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP16MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 119.60 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP16MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 132.47 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP16MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 120.55 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 137.35 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 124.72 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 126.19 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 113.35 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 138.53 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 124.92 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 126.38 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 113.04 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP32MultiCtasKvGmemSepVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 119.22 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP32MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 107.84 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP32MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 122.70 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP32MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 110.75 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 125.53 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 112.48 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 114.43 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 101.11 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 128.69 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 115.97 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 117.32 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 104.17 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP64MultiCtasKvGmemSepVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 114.63 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP64MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 103.25 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP64MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 118.48 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP64MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 106.54 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 120.66 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 107.88 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 109.55 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 96.52 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 124.31 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 111.63 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 112.83 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 99.74 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 170.52 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 161.91 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 119.88 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 113.47 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 166.55 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 179.12 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 159.66 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 116.47 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 126.87 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 111.30 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 170.69 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 155.17 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 190.16 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 176.61 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 155.69 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 143.66 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 120.92 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 103.85 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 141.15 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 118.02 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 110.34 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.80 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 154.62 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 146.15 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 113.47 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.93 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 150.59 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 164.16 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 143.71 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.06 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 120.33 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 104.90 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 155.79 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 139.30 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 174.77 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 160.62 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 140.20 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 126.85 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 113.82 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.73 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 134.37 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 111.91 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 103.35 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.70 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 148.12 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 139.51 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.67 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.26 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 144.09 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 157.77 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 137.32 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 107.26 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 117.55 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 102.09 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 149.25 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 133.05 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 168.35 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 153.58 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 133.77 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 120.47 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 111.04 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.94 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 131.45 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 109.11 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 100.68 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvCausalP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.89 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvDenseDynamicTokenSparseP1MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 156.40 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvDenseDynamicTokenSparseP1MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 150.03 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvDenseDynamicTokenSparseP1MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 140.97 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvDenseDynamicTokenSparseP1MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 153.09 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvDenseDynamicTokenSparseP1MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 163.83 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvDenseDynamicTokenSparseP1MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 147.77 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvDenseDynamicTokenSparseP1VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 145.78 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvDenseDynamicTokenSparseP1VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 127.24 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvDenseDynamicTokenSparseP1VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 155.68 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvDenseDynamicTokenSparseP1VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 138.66 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvDenseDynamicTokenSparseP1VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 137.09 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvDenseDynamicTokenSparseP1VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 122.44 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvDenseDynamicTokenSparseP1VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 133.93 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvDenseDynamicTokenSparseP1VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 124.20 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 174.85 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 165.98 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 124.11 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 117.17 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 170.82 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 183.77 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 163.80 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 120.57 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 131.58 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 115.13 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 175.06 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 159.44 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 194.57 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 179.63 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 164.35 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 147.09 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 125.67 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 108.05 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 146.72 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 123.08 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 116.76 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 100.51 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 159.34 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 150.46 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 117.57 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 110.76 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 155.30 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 168.81 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 148.15 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.03 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 125.29 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.73 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 159.87 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 143.70 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 179.55 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 162.41 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 149.12 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 130.41 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 118.64 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.94 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 139.69 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 116.96 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 109.73 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 94.39 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 152.95 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 143.95 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.77 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.84 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 148.92 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 162.42 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 141.77 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 111.23 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 122.38 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.92 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 152.77 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 137.31 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 173.13 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 156.16 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 142.71 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 124.02 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 115.84 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.14 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 136.76 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.16 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 106.93 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H512PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 91.72 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 94.88 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 82.69 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 98.86 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 86.67 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 94.37 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PackedQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:36 110.36 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PackedQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:36 87.46 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PackedQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:35 112.94 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PackedQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:35 93.66 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PackedQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:36 112.23 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PackedQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:36 89.52 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 82.72 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 98.41 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:36 110.90 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:36 91.30 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:35 116.64 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:36 97.45 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:36 113.01 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:36 93.36 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 86.70 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 105.42 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 93.51 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 109.44 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 97.49 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 117.18 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.65 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.10 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP16MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 107.95 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.17 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 123.10 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.45 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.98 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 110.57 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 79.48 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 107.69 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 102.79 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.41 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP16MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 100.09 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.18 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 113.48 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 81.37 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 89.20 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 102.97 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 78.62 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 111.43 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 108.26 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 98.57 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 94.55 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 102.98 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.56 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 116.23 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 102.87 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 98.71 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 83.54 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 98.13 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.91 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 115.27 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 111.73 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 102.30 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 99.31 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 83.57 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 70.75 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 95.75 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.62 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 102.29 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 86.86 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 79.21 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 66.90 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.37 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.44 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 109.76 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP32MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 105.13 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.83 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 120.29 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 81.05 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.70 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 107.88 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 78.07 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 104.87 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.45 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.07 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP32MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 97.16 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 94.84 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 110.66 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 80.09 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.79 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 100.16 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 77.20 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 108.32 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 105.52 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 95.79 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 91.84 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 98.58 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 86.38 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 111.82 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.77 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 95.75 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 80.70 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 93.59 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.73 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 112.28 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 109.02 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 99.51 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 96.60 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 82.18 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 69.34 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 94.17 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 78.21 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 99.45 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 84.15 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 77.90 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 65.48 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.83 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.14 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.59 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP64MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 103.60 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.66 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 118.88 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 80.29 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.93 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.23 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 77.32 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 103.34 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.27 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.90 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP64MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 95.75 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.67 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 109.13 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.20 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.02 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 98.75 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 76.45 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 106.77 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 103.98 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 94.39 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 90.30 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 96.27 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.21 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 109.51 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.73 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 94.21 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 79.17 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 91.27 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.44 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 110.73 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 107.48 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 98.11 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 95.07 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 81.40 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 68.57 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 93.08 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 77.45 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 97.91 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 82.61 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 77.13 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 64.71 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalPagedMultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 116.23 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 100.23 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalPagedMultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.56 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalPagedMultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 107.27 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalPagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 96.51 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 121.72 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.93 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 91.47 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 109.62 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.95 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalPagedMultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 106.85 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalPagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.24 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalPagedMultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 108.74 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalPagedMultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 99.54 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalPagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.81 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 111.84 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.84 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.56 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 101.81 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.09 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 108.20 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalPagedVarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:35 106.11 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 95.79 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalPagedVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 92.56 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalPagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:35 98.32 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.62 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalPagedVarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 111.73 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalPagedVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.24 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalPagedVarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 96.45 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalPagedVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 81.30 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalPagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 93.45 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalPagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.85 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 112.16 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 109.43 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 99.63 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 97.33 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 86.39 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 70.77 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 96.04 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.64 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 100.03 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 84.87 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 81.70 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 66.79 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 121.52 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 102.76 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.70 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.41 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 126.88 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.16 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.77 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 113.37 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.29 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 117.13 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 106.80 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 93.53 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 100.30 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.12 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 111.10 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.56 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 96.48 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.67 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 110.34 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 98.24 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 118.71 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.55 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.36 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 95.06 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 124.07 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 109.22 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.55 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 109.03 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.07 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.31 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 103.94 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 90.72 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 95.77 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.88 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 106.56 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.19 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 91.95 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 80.42 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 107.48 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 95.52 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 117.18 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.25 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 106.20 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.90 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 122.53 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 107.80 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.38 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.86 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 91.78 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.91 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 102.40 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 89.30 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 93.58 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 81.70 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 104.26 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.02 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 89.63 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 78.12 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 105.94 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 94.02 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomPagedMultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 120.58 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.48 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomPagedMultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:35 109.42 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomPagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.00 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 125.52 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomPagedMultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 111.20 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomPagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.37 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomPagedMultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 109.84 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomPagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.01 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomPagedMultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 115.66 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomPagedVarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 104.63 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomPagedVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:35 91.54 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomPagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 95.63 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.92 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomPagedVarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 106.48 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomPagedVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.38 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomPagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 91.81 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomPagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 80.34 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomPagedVarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 108.16 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvCustomPagedVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 96.23 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 109.27 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 98.64 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 113.26 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 102.51 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 106.18 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 95.86 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 110.30 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 99.85 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 104.73 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 94.34 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 108.73 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 98.32 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvDensePagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 106.17 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvDensePagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 95.88 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 110.16 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 99.87 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvDenseStaticTokenSparseP1MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 209.44 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvDenseStaticTokenSparseP1MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.89 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvDenseStaticTokenSparseP1MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.63 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvDenseStaticTokenSparseP1MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 95.28 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvDenseStaticTokenSparseP1MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.62 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvDenseStaticTokenSparseP1MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 201.41 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvDenseStaticTokenSparseP1MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.59 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvDenseStaticTokenSparseP1MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.88 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvDenseStaticTokenSparseP1MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 87.60 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvDenseStaticTokenSparseP1MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.95 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvDenseStaticTokenSparseP1VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 200.03 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvDenseStaticTokenSparseP1VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 185.40 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvDenseStaticTokenSparseP1VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 83.44 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvDenseStaticTokenSparseP1VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 70.30 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvDenseStaticTokenSparseP1VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 89.86 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvDenseStaticTokenSparseP1VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 76.65 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvDenseStaticTokenSparseP1VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 85.91 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvDenseStaticTokenSparseP1VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 74.27 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvDenseStaticTokenSparseP1VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 81.76 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvDenseStaticTokenSparseP1VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 68.05 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 124.41 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.47 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 119.30 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.39 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.50 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 130.47 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.55 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.81 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 116.95 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.18 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 115.05 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.61 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 117.61 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.66 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.51 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 120.72 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.46 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.78 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 109.23 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.31 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 125.24 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 115.66 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 112.53 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 101.94 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 106.77 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.42 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 120.90 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.45 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 104.08 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 88.72 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 101.52 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.54 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 126.82 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 118.97 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 113.97 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.84 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 90.76 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 75.00 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 101.37 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.68 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 107.62 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 92.07 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 85.58 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 70.47 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 121.60 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.13 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.96 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 111.58 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.16 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 127.66 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.14 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.41 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.14 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.77 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.23 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.27 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 113.40 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 103.73 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.17 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 118.02 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.05 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.38 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.41 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 80.91 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 122.13 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 112.80 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 109.74 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 99.10 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 102.27 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.18 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 116.36 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.20 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 101.12 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 85.88 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 97.00 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.30 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 123.84 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 116.26 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 111.31 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 103.99 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 89.05 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 73.59 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 99.79 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.27 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 104.80 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 89.36 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 83.99 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 69.19 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 120.07 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.96 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.80 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 110.05 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.99 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 126.24 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.38 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.63 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.73 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.02 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 110.70 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.09 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 111.23 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 102.20 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.00 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 116.49 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.30 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 91.60 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 104.88 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 80.15 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 120.59 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 111.26 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 108.34 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 97.70 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 99.92 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.01 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 114.05 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.90 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 99.70 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 84.48 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 94.69 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.00 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 122.30 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 114.72 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 109.78 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 102.59 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 87.96 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 72.82 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 98.70 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.49 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 103.26 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 87.82 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 82.90 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 68.41 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 123.92 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.02 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 116.97 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.09 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.05 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 129.54 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.55 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.81 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 115.99 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.30 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.67 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.02 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 115.16 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.23 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.18 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 119.66 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.59 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.91 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.27 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.31 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 122.73 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 114.09 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 110.35 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 100.39 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 103.37 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 91.59 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 117.78 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.62 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 102.91 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 90.30 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 98.24 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.58 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 124.31 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 117.42 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 111.91 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 105.41 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 91.45 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 75.54 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 102.05 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.22 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 105.78 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 93.17 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 86.26 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 71.13 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:36 110.48 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:36 87.59 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:36 112.94 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:36 93.66 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:36 112.36 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:36 89.52 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:36 111.02 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:36 91.42 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:36 116.77 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:36 97.45 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:36 113.13 KB
fmhaSm100fKernel_QkvE4m3OBfloat16H64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:36 93.36 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk128HV64SeparateQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 98.28 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk128HV64SeparateQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 86.62 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk128HV64SeparateQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 102.24 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk128HV64SeparateQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 90.46 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk128HV64SeparateQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 98.04 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk128HV64SeparateQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 86.77 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk128HV64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 102.07 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk128HV64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 90.61 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk192HV128SeparateQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 103.29 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk192HV128SeparateQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 91.11 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk192HV128SeparateQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 107.45 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk192HV128SeparateQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 94.59 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk192HV128SeparateQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 102.82 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk192HV128SeparateQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 92.38 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk192HV128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 107.02 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk192HV128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 95.99 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 131.86 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 132.19 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 142.47 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP16MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 117.21 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 126.12 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 137.55 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.84 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 111.30 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 122.14 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.01 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 120.65 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 129.87 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 139.03 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP16MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 105.73 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 124.39 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 126.34 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.34 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.24 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP16MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 110.91 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.41 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 118.55 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.85 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 131.23 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.12 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 141.17 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 126.35 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 104.77 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 92.06 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 120.29 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.72 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 122.35 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.80 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 101.45 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.35 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 112.28 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.69 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 112.02 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 97.47 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 93.76 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.99 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 121.48 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 123.12 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 133.69 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP32MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.84 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 117.17 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 124.41 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.84 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.45 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.96 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.14 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 110.27 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 121.09 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 130.25 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP32MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 95.35 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 115.60 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 113.20 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.62 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.52 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP32MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 97.73 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.69 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 108.59 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 97.20 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 121.20 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.58 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 131.23 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 117.68 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 94.34 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 82.42 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 111.00 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.05 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 111.47 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 100.13 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 96.66 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.48 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 107.49 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.82 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 96.42 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 84.67 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 89.40 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 77.12 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 116.55 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 119.29 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 129.86 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP64MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 101.91 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 113.22 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 119.79 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.99 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.60 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 104.47 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.16 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 105.34 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 117.25 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 126.30 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP64MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 90.55 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 111.77 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.58 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.64 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.68 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP64MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 93.11 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.72 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 103.09 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 92.28 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 117.22 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.62 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 127.38 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 113.85 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 89.20 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 77.37 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 107.02 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.21 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 106.39 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 95.38 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 94.67 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.62 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 105.51 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.83 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 91.80 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 79.92 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 87.53 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvDenseP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 75.26 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 142.86 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 135.44 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 145.94 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 124.70 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 129.38 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 148.31 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.41 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 115.10 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 129.89 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.46 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 131.40 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 133.16 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 142.38 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 113.47 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 127.57 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 137.09 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.80 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.91 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 118.66 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.87 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 129.59 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 116.80 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 133.80 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 115.99 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 143.61 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 127.95 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 112.02 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 98.84 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 123.05 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.46 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 132.77 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 121.95 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 105.35 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.81 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 116.78 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.63 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 119.41 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 103.32 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 99.84 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.37 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 132.48 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 126.32 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 137.03 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.33 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 120.27 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 135.29 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.45 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.23 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 116.72 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.62 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 121.02 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 124.28 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 133.59 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 103.09 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 118.70 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 123.95 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.10 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.18 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 105.48 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.30 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 119.10 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.51 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 124.18 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.63 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 134.99 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 118.59 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 101.57 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 88.55 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 113.78 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.09 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 121.91 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 109.09 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 100.20 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.94 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 111.63 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.88 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 103.72 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 90.45 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 94.81 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.62 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 127.56 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 122.49 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 133.20 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 109.53 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 116.31 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 130.66 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.56 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.26 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.09 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.62 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 116.23 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 120.45 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 129.64 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 98.17 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.87 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 119.33 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.22 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.33 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 100.87 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.29 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 113.64 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 101.51 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 120.20 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.80 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 131.02 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.88 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 96.45 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 83.55 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 109.78 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.39 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 116.83 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 104.62 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 98.21 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.08 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 109.64 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.90 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 99.20 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 85.98 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 92.82 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk320HV256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 77.63 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 139.84 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvCgaVarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 126.42 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 151.51 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvCgaVarSeqQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 137.24 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 132.47 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvCgaVarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 119.45 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.07 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.45 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.60 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.02 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.58 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.25 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 107.79 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 109.09 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 138.50 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvVarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 125.08 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 148.51 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvVarSeqQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 134.38 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 131.74 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvVarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 118.60 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.56 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.98 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 111.67 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.26 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.69 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.39 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 138.06 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 124.36 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqQ16Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 124.54 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 111.57 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 151.26 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 135.48 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqQ32Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 137.02 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 121.95 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 113.85 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 102.07 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 132.39 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 118.52 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqQ8Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 119.26 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.02 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 104.05 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 91.11 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqSkipsSoftmaxQ16Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 97.20 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.31 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 112.50 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.68 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqSkipsSoftmaxQ32Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 104.73 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqSkipsSoftmaxQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.17 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 113.31 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 103.25 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 99.91 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.27 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqSkipsSoftmaxQ8Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 93.23 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP16VarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 80.02 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 128.47 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvCgaVarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 121.84 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 140.01 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvCgaVarSeqQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 132.53 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 121.10 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvCgaVarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.74 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.05 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.12 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.58 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.59 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.55 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.92 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 97.00 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 99.30 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 127.12 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvVarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 120.37 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 137.02 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvVarSeqQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 129.67 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 120.38 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvVarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 113.89 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.45 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.53 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.69 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.82 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.70 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.06 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 126.30 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 113.62 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqQ16Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 119.66 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.98 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 139.48 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 124.73 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqQ32Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 132.15 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 117.24 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 102.48 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 91.29 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 120.75 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.91 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqQ8Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 114.38 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.30 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 98.41 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.05 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqSkipsSoftmaxQ16Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 94.77 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 80.88 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 106.74 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.61 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqSkipsSoftmaxQ32Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 102.29 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqSkipsSoftmaxQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.86 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 104.20 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 93.46 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 94.27 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.23 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqSkipsSoftmaxQ8Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 90.79 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP32VarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 77.58 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 123.76 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvCgaVarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 119.17 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 135.30 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvCgaVarSeqQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 129.87 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 116.39 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvCgaVarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.20 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.60 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.78 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.27 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.23 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.10 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.45 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 92.29 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 94.84 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 122.41 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvVarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 117.84 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 132.43 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvVarSeqQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 127.13 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 115.66 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvVarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 111.35 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.12 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.19 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.25 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.35 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 91.25 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.59 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 121.42 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.91 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqQ16Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 117.36 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.45 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 134.60 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 119.98 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqQ32Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 129.86 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.84 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 97.49 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 86.58 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 115.88 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.20 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqQ8Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 112.09 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.90 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 95.97 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.62 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqSkipsSoftmaxQ16Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 93.55 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.53 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 104.30 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.30 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqSkipsSoftmaxQ32Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 101.08 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqSkipsSoftmaxQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.39 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 99.84 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 89.01 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 91.83 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 78.92 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqSkipsSoftmaxQ8Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 89.58 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseP64VarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 76.11 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseStaticTokenSparseP1MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 113.57 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseStaticTokenSparseP1MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 120.23 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseStaticTokenSparseP1MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 109.12 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseStaticTokenSparseP1MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.46 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseStaticTokenSparseP1MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.77 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseStaticTokenSparseP1MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 117.66 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseStaticTokenSparseP1MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.83 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseStaticTokenSparseP1VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 112.16 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseStaticTokenSparseP1VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.58 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseStaticTokenSparseP1VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 120.02 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseStaticTokenSparseP1VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.27 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseStaticTokenSparseP1VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 115.76 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseStaticTokenSparseP1VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 102.98 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseStaticTokenSparseP1VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 107.81 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta128PagedKvDenseStaticTokenSparseP1VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.30 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 151.79 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvCgaVarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 135.30 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 171.20 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvCgaVarSeqQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 153.87 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 145.34 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvCgaVarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 129.12 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.95 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.77 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 122.53 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.38 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.38 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.36 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvGmemSepVarSeqQ64Kv128Static2CtaKeepsAbForGen.cubin 05-Jun-2026 22:36 108.52 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.01 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128Static2CtaKeepsAbForGen.cubin 05-Jun-2026 22:36 111.11 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.46 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 149.80 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvVarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 133.19 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 164.65 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvVarSeqQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 147.45 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 143.93 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvVarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 127.73 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.98 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.58 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 118.98 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.87 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.93 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.06 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 150.23 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 136.37 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqQ16Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 133.30 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 119.80 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 167.77 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 155.71 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqQ32Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 149.66 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 135.20 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqQ64Kv128Persistent2CtaKeepsAbForGen.cubin 05-Jun-2026 22:36 116.60 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 118.61 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqQ64Kv128Static2CtaKeepsAbForGen.cubin 05-Jun-2026 22:36 103.24 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.27 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 143.66 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 129.91 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqQ8Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 126.84 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 113.73 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 110.62 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.41 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ16Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 102.08 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.12 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 123.76 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.97 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ32Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 114.41 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.87 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ64Kv128Persistent2CtaKeepsAbForGen.cubin 05-Jun-2026 22:36 119.43 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 119.43 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ64Kv128Static2CtaKeepsAbForGen.cubin 05-Jun-2026 22:36 105.71 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.63 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 105.94 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.23 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ8Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 97.70 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.35 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 138.48 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvCgaVarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 129.82 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 157.77 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvCgaVarSeqQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 148.39 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 132.03 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvCgaVarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 123.65 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.86 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.99 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 115.45 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 109.64 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.29 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.58 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvGmemSepVarSeqQ64Kv128Static2CtaKeepsAbForGen.cubin 05-Jun-2026 22:36 101.42 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 97.88 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128Static2CtaKeepsAbForGen.cubin 05-Jun-2026 22:36 104.02 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 99.78 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 136.36 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvVarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 127.70 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 151.22 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvVarSeqQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 141.98 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 130.62 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvVarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 122.12 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.80 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.93 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 111.93 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.12 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.88 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.29 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 136.20 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 122.98 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqQ16Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 127.52 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.32 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 153.49 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 139.08 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqQ32Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 144.01 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 129.73 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqQ64Kv128Persistent2CtaKeepsAbForGen.cubin 05-Jun-2026 22:36 108.20 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 104.72 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqQ64Kv128Static2CtaKeepsAbForGen.cubin 05-Jun-2026 22:36 95.76 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 92.43 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 129.59 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 116.52 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqQ8Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 121.20 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.13 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 103.91 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.41 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ16Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 99.32 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.37 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 117.07 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.98 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ32Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 111.66 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.11 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ64Kv128Persistent2CtaKeepsAbForGen.cubin 05-Jun-2026 22:36 111.02 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 106.02 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ64Kv128Static2CtaKeepsAbForGen.cubin 05-Jun-2026 22:36 98.23 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 94.02 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 99.23 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.11 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ8Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 94.93 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.59 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 132.88 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvCgaVarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 126.84 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 152.41 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvCgaVarSeqQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 145.53 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 126.55 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvCgaVarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 120.66 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.09 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.50 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.70 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.99 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.51 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.97 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvGmemSepVarSeqQ64Kv128Static2CtaKeepsAbForGen.cubin 05-Jun-2026 22:36 96.56 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 90.60 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128Static2CtaKeepsAbForGen.cubin 05-Jun-2026 22:36 99.28 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 93.15 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 130.88 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvVarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 124.72 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 145.73 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvVarSeqQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 138.99 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 125.02 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvVarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 119.26 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.15 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.31 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 109.19 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.48 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.10 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.80 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 130.43 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 117.62 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqQ16Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 124.90 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 111.59 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 147.72 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 133.60 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqQ32Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 141.38 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 126.99 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqQ64Kv128Persistent2CtaKeepsAbForGen.cubin 05-Jun-2026 22:36 103.20 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 96.59 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqQ64Kv128Static2CtaKeepsAbForGen.cubin 05-Jun-2026 22:36 90.90 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 84.84 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 123.82 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 111.03 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqQ8Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 118.46 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.41 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 101.16 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.65 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ16Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 97.95 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.86 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 114.18 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.22 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ32Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 110.28 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 91.48 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ64Kv128Persistent2CtaKeepsAbForGen.cubin 05-Jun-2026 22:36 105.86 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 99.05 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ64Kv128Static2CtaKeepsAbForGen.cubin 05-Jun-2026 22:36 93.37 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 87.26 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 96.34 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.35 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ8Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 93.44 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.96 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseStaticTokenSparseP1MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 122.04 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseStaticTokenSparseP1MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 130.44 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseStaticTokenSparseP1MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 117.98 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseStaticTokenSparseP1MultiCtasKvGmemSepVarSeqQ64Kv128Static2CtaKeepsAbForGen.cubin 05-Jun-2026 22:36 106.85 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseStaticTokenSparseP1MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 110.73 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseStaticTokenSparseP1MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 119.15 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseStaticTokenSparseP1MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 127.55 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseStaticTokenSparseP1MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 115.83 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseStaticTokenSparseP1VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 119.80 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseStaticTokenSparseP1VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.27 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseStaticTokenSparseP1VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 129.86 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseStaticTokenSparseP1VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.65 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseStaticTokenSparseP1VarSeqQ64Kv128Persistent2CtaKeepsAbForGen.cubin 05-Jun-2026 22:36 117.75 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseStaticTokenSparseP1VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 120.55 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseStaticTokenSparseP1VarSeqQ64Kv128Static2CtaKeepsAbForGen.cubin 05-Jun-2026 22:36 100.74 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseStaticTokenSparseP1VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 105.06 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseStaticTokenSparseP1VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 115.15 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512HVPerCta256PagedKvDenseStaticTokenSparseP1VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.11 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 175.70 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16MultiCtasKvCgaVarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 152.78 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 167.22 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16MultiCtasKvCgaVarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 144.45 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 125.55 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 113.95 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 119.35 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.05 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 127.91 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 126.55 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 172.23 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16MultiCtasKvVarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 149.30 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 194.88 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16MultiCtasKvVarSeqQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 170.89 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 164.92 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16MultiCtasKvVarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 142.40 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 122.21 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.66 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 131.98 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 119.47 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 117.40 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.12 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 174.84 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 159.52 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16VarSeqQ16Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 150.94 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16VarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 136.25 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 198.37 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 182.54 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16VarSeqQ32Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 173.48 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16VarSeqQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 158.34 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 134.39 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 122.13 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 165.43 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 151.82 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16VarSeqQ8Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 141.91 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16VarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 128.58 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 123.64 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.95 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16VarSeqSkipsSoftmaxQ16Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 112.59 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16VarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.44 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 137.48 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 119.31 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16VarSeqSkipsSoftmaxQ32Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 125.90 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16VarSeqSkipsSoftmaxQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.85 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 130.33 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 120.95 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 116.63 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.91 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16VarSeqSkipsSoftmaxQ8Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 106.00 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP16VarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.55 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 158.10 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32MultiCtasKvCgaVarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 145.63 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 149.62 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32MultiCtasKvCgaVarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 137.30 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 117.18 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.55 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.99 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.65 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.59 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 109.69 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 154.74 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32MultiCtasKvVarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 142.16 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 177.04 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32MultiCtasKvVarSeqQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 163.74 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 147.44 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32MultiCtasKvVarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 135.12 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 113.88 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.25 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 123.66 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 115.95 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 109.20 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.72 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 156.09 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 141.44 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32VarSeqQ16Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 143.62 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32VarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 129.10 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 179.33 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 164.23 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32VarSeqQ32Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 166.17 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32VarSeqQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 151.19 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 115.45 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 102.80 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 146.80 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 133.74 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32VarSeqQ8Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 134.59 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32VarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 121.43 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 115.68 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.59 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32VarSeqSkipsSoftmaxQ16Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 109.05 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32VarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.91 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 129.55 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.95 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32VarSeqSkipsSoftmaxQ32Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 122.37 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32VarSeqSkipsSoftmaxQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.45 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 116.69 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 103.91 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 108.67 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.55 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32VarSeqSkipsSoftmaxQ8Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 102.47 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP32VarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.14 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 150.83 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64MultiCtasKvCgaVarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 141.88 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 142.47 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64MultiCtasKvCgaVarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 133.55 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 113.78 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.70 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.46 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.68 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 99.34 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 102.12 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 147.47 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64MultiCtasKvVarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 138.40 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 169.77 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64MultiCtasKvVarSeqQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 159.87 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 140.16 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64MultiCtasKvVarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 131.38 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.35 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.28 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 120.27 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.12 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.66 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.75 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 148.77 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 134.41 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64VarSeqQ16Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 140.23 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64VarSeqQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 125.48 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 172.02 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 157.08 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64VarSeqQ32Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 162.66 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64VarSeqQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 147.44 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 106.05 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 93.44 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 139.48 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 126.59 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64VarSeqQ8Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 131.09 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64VarSeqQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 117.80 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 112.15 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.06 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64VarSeqSkipsSoftmaxQ16Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 107.36 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64VarSeqSkipsSoftmaxQ16Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 91.95 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 126.02 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.42 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64VarSeqSkipsSoftmaxQ32Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 120.67 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64VarSeqSkipsSoftmaxQ32Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.49 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 108.91 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 96.34 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 105.14 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.14 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64VarSeqSkipsSoftmaxQ8Kv64PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 100.77 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseP64VarSeqSkipsSoftmaxQ8Kv64StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.19 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseStaticTokenSparseP1MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 138.39 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseStaticTokenSparseP1MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 131.79 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseStaticTokenSparseP1MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 124.58 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseStaticTokenSparseP1MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 133.62 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseStaticTokenSparseP1MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 144.62 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseStaticTokenSparseP1MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 128.41 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseStaticTokenSparseP1VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 140.29 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseStaticTokenSparseP1VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 121.80 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseStaticTokenSparseP1VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 150.59 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseStaticTokenSparseP1VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 132.76 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseStaticTokenSparseP1VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 133.16 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseStaticTokenSparseP1VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 118.60 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseStaticTokenSparseP1VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 127.95 KB
fmhaSm100fKernel_QkvE4m3OBfloat16HQk576HV512PagedKvDenseStaticTokenSparseP1VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 118.16 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 101.66 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 89.44 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 105.65 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 93.43 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 101.28 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PackedQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:36 109.27 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PackedQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:36 96.09 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PackedQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:36 115.33 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PackedQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:36 102.33 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PackedQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:36 111.22 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PackedQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:36 98.16 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 89.45 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 105.20 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:36 113.03 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:36 99.92 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:36 119.16 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:36 106.10 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:36 115.11 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:36 101.98 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 93.31 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 112.09 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 100.29 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 116.09 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 104.27 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 132.66 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 109.68 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 119.13 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP16MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 116.50 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.84 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 135.95 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.13 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.28 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 118.96 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.31 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 124.15 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.16 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 116.68 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP16MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.44 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.98 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 123.23 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.32 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.89 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 109.02 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 80.54 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 118.24 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 124.80 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 105.88 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.84 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 111.98 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.32 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 119.48 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.45 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 107.01 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 94.72 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 101.31 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.88 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 122.22 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 126.98 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 109.72 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.61 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 91.70 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 75.23 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 98.93 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.01 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 110.12 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 97.55 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 81.62 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 68.87 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 129.84 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.34 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.92 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP32MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 113.56 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.50 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 133.25 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.73 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.87 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 116.15 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 80.02 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 121.34 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.82 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.47 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP32MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 103.49 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.63 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 120.42 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.91 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 91.48 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.34 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.12 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 115.26 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 121.94 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 103.09 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 110.12 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 107.45 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.02 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 114.95 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.35 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 104.05 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 91.77 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 96.78 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.70 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 119.11 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 124.25 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 107.06 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 111.90 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 89.88 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 73.95 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 97.36 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.59 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 107.42 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 94.74 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 80.20 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 67.46 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 128.30 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.17 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.75 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP64MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.02 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.33 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 131.72 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.97 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.09 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.73 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.14 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 119.80 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.65 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.30 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP64MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 101.96 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.47 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 118.88 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.16 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.71 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 104.80 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 78.37 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 113.72 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 120.40 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 101.70 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.59 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 105.14 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.84 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 112.77 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.19 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 102.51 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 90.37 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 94.47 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.41 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 117.57 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 122.71 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 105.54 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 110.49 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 88.78 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 73.18 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 96.27 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 80.83 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 105.88 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 93.21 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 79.44 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 66.69 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalPagedMultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 131.70 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.14 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalPagedMultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 115.84 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalPagedMultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 115.81 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalPagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.30 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 134.55 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.61 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.77 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 118.01 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.78 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalPagedMultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 123.20 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalPagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.62 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalPagedMultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 113.27 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalPagedMultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 105.75 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalPagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.44 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 121.73 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.80 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.38 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 107.95 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 80.88 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 115.02 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalPagedVarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 122.52 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 102.57 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalPagedVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 110.24 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalPagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 107.37 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.25 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalPagedVarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 115.38 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalPagedVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.70 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalPagedVarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 104.75 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalPagedVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 92.76 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalPagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 96.65 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalPagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.69 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 118.88 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 124.91 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 106.41 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.62 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 91.73 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 75.38 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 99.22 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.03 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 107.99 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 95.73 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 84.27 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 68.89 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 137.02 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.91 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 118.05 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.93 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 139.72 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 128.64 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.55 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 115.66 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.06 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 126.88 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 123.55 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 111.85 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 109.43 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.12 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 115.30 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.51 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 99.89 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.85 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 125.58 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 113.54 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 134.20 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.56 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 113.72 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.71 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 136.91 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 125.83 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.34 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 111.45 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.72 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 124.20 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 120.56 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 109.04 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 104.89 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.87 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 110.77 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.13 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 95.36 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.60 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 122.72 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 110.81 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 132.67 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.40 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 111.55 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.41 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 135.50 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 124.29 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.17 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 109.15 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.55 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 122.66 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 119.02 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 107.63 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 102.71 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.57 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 108.46 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.96 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 93.05 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.31 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 121.18 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 109.31 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomPagedMultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 136.19 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.62 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomPagedMultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.35 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomPagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.52 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 138.37 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomPagedMultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 127.56 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomPagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.16 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomPagedMultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.26 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomPagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.78 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomPagedMultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 125.53 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomPagedVarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 121.12 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomPagedVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 109.24 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomPagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 104.81 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.91 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomPagedVarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 111.07 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomPagedVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.46 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomPagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 95.23 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomPagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.53 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomPagedVarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 123.28 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvCustomPagedVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 111.55 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 117.07 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 105.91 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 120.98 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 109.90 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 113.99 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 103.25 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 118.02 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 107.11 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 112.41 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 101.73 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 116.45 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 105.71 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvDensePagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 113.86 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvDensePagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 102.64 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 117.88 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 106.62 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 140.05 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.38 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 124.02 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 122.95 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.17 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 143.43 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 91.10 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.98 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 125.23 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.13 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 131.55 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.85 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 121.45 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.98 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.30 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 130.59 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.41 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.60 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 115.28 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.11 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 132.18 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 132.26 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 119.62 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 119.83 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 115.78 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.09 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 124.16 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.48 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 112.34 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 100.61 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 104.59 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.38 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 133.78 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 134.30 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 121.19 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 122.11 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 95.94 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.56 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 104.21 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.09 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 115.75 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 103.20 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 88.26 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 72.71 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 137.23 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.16 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 119.81 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 120.12 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.96 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 140.49 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.69 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.57 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 122.42 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.72 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 128.60 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.64 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 117.12 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 110.04 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.97 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 127.78 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.00 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.31 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.47 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.70 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 129.20 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 129.39 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 116.84 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 117.12 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 111.37 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.84 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 119.62 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.23 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 109.38 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 97.67 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 100.09 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.13 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 130.67 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 131.60 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 118.41 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 119.39 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 94.37 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 78.15 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 102.64 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.68 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 112.91 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 100.25 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 86.69 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 71.30 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 135.70 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.86 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 117.52 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 118.59 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.66 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 139.09 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.94 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.80 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 120.88 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.97 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 127.20 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.47 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 115.07 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.63 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.80 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 126.25 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.25 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.41 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 111.05 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.95 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 127.66 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 127.85 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 115.45 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 115.59 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 109.06 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.67 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 117.31 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.94 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 107.84 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 96.27 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 97.87 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.96 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 129.13 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 130.06 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 117.00 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 117.86 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 93.15 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 77.38 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 101.55 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.91 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 111.38 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 98.72 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 85.55 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 70.52 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 139.68 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.04 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 121.69 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 122.52 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.72 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 142.38 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.22 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.11 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 124.41 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.13 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 131.05 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.40 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 118.99 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.55 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.98 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 129.55 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.54 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.72 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.32 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.23 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 129.70 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 130.73 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 117.45 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 118.41 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 112.66 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.25 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 121.30 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 109.66 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 111.04 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 98.84 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 101.44 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.54 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 131.30 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 132.76 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 119.01 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 120.68 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 96.62 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 80.10 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 104.90 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.63 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 113.67 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 101.55 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 88.95 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 73.25 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128SeparateQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 101.41 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:36 109.40 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:36 96.22 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:36 115.45 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:36 102.33 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:36 111.34 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:36 98.16 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128SeparateQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 89.57 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 105.20 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:36 113.16 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:36 100.05 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:36 119.29 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:36 106.10 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:36 115.23 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:36 101.98 KB
fmhaSm100fKernel_QkvE4m3OE2m1H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 93.44 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 97.30 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 85.48 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 100.05 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 88.20 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 98.23 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 86.06 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 101.23 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 88.90 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 109.01 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 96.28 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 111.70 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 99.13 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 129.05 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 126.21 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 137.30 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP16MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 118.98 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 119.66 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 131.51 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.81 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 109.09 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 121.03 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.64 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 116.76 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 123.48 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 132.45 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP16MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.28 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 117.91 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 119.34 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.54 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.30 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 110.12 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.14 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 119.49 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 115.32 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 107.14 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 101.19 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 129.70 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 115.62 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 139.41 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 126.80 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 108.88 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 89.68 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 115.79 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.28 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 122.20 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 117.98 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 109.73 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 104.57 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 105.44 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.41 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 114.84 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.33 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 112.34 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 92.59 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 93.71 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 80.69 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 122.03 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 118.84 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 130.91 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP32MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 111.96 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.28 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 125.75 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.73 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.01 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 115.25 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.44 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 109.75 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 116.55 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 126.07 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP32MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 101.40 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.85 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 113.59 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.34 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.22 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 104.34 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.94 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 112.48 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 107.62 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 100.98 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 95.03 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 122.42 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.97 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 132.13 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 119.17 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 101.95 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 83.65 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 108.93 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.70 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 115.19 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 111.15 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 103.57 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 98.81 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 101.69 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.31 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 110.97 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.25 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 104.94 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 86.85 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 90.50 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 77.59 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 119.09 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 115.65 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 127.85 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP64MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 109.02 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 109.22 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 122.81 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.21 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.47 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.30 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.04 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.80 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 113.35 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 123.01 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP64MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 98.33 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.66 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 110.65 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 91.94 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.68 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 101.40 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.54 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 109.40 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 104.41 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 97.91 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 92.09 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 119.34 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.91 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 128.92 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 115.98 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 98.84 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 80.71 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 105.70 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.76 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 112.11 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 108.20 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 100.51 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 95.88 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 100.18 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.89 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 109.30 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.71 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 101.96 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 83.91 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 88.95 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 76.05 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalPagedMultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 121.11 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 117.79 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalPagedMultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 129.95 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalPagedMultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 111.03 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalPagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 111.36 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 124.95 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.89 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.15 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.35 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 91.84 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalPagedMultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.95 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalPagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 115.62 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalPagedMultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 125.11 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalPagedMultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 100.46 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalPagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 109.80 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.79 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.62 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.49 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 103.56 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.34 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 112.42 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalPagedVarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 105.68 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 99.66 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalPagedVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 93.68 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalPagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 120.95 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.70 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalPagedVarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 130.53 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalPagedVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 117.88 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalPagedVarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 100.70 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalPagedVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 82.39 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalPagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 107.12 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalPagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.13 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 115.24 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 109.57 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 102.38 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 97.67 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 102.06 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.67 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 111.34 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.60 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 103.94 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 85.73 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 93.59 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 77.96 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 122.33 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 107.95 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 125.20 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 110.91 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 114.00 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 101.66 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 116.87 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 104.62 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 110.91 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 98.71 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 113.78 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 101.68 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvDensePagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 112.43 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvDensePagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 100.23 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 115.42 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 103.19 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 136.51 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 130.73 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 142.08 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 124.22 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 123.92 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 139.09 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.78 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 113.80 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 126.39 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.34 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 124.34 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 128.02 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 137.08 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 113.69 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 122.09 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 126.80 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.51 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.01 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 115.64 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.84 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 135.62 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 122.80 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 119.16 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 109.15 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 134.10 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 119.88 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 144.22 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 131.68 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 113.88 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 95.49 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 123.85 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.84 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 136.58 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 124.69 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 121.72 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.33 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 110.23 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.73 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 120.03 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.43 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 117.37 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 98.20 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 100.96 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.52 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 129.49 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 123.53 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 135.56 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 117.33 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 116.73 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 133.34 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.59 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.70 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 120.60 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.15 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 117.33 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 121.23 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 130.56 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.67 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 115.30 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 121.05 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.45 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.91 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 109.85 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.65 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 127.50 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 114.79 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 112.39 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 102.51 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 126.62 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.24 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 136.85 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 124.05 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 106.94 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 88.84 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 116.38 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.26 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 128.45 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 118.45 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 115.08 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.27 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 105.97 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.64 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 116.04 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.35 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 110.17 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 92.17 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 97.00 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.42 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 126.55 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 120.47 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 132.50 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.27 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 113.66 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 130.40 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.05 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 109.17 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 117.66 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.73 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.39 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 118.17 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 127.62 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 103.73 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.23 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 118.11 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.91 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.38 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.91 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 91.23 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 124.42 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 111.59 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 109.45 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 99.56 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 123.41 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 109.17 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 133.64 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 120.86 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 103.71 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 85.91 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 113.29 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.20 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 125.24 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 115.37 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 112.13 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 103.30 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 104.42 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.09 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 114.49 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.81 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 107.09 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 89.23 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 95.48 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.88 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 129.29 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 122.73 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 135.06 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 116.86 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 116.05 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 133.16 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.64 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.66 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 120.16 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.07 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 117.12 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 120.43 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 130.19 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.33 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.37 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 121.00 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.37 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.74 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 109.65 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.70 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 126.64 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 113.76 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 111.62 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 101.73 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 125.51 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 111.12 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 135.73 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 122.94 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 106.47 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 88.19 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 111.59 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.19 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 127.59 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 117.52 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 114.44 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 105.55 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 106.99 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.40 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 116.94 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.22 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 109.70 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 91.38 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 95.68 KB
fmhaSm100fKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.30 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 106.21 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 93.53 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 108.97 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 96.07 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 107.27 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 94.13 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 110.27 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 96.98 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP16MultiCtasKvGmemSepVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 123.91 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP16MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 115.30 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP16MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 125.25 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP16MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 116.02 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 135.88 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 131.13 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 122.37 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 117.34 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 125.69 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 105.92 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 138.71 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 131.87 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 125.04 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 118.30 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 128.52 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.30 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP32MultiCtasKvGmemSepVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.14 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP32MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 103.54 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP32MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 115.48 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP32MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.34 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 123.81 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 119.05 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 111.73 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 105.83 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 114.02 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 94.41 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 126.64 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 122.33 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 114.27 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 109.34 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 116.88 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 97.35 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP64MultiCtasKvGmemSepVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 107.42 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP64MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 98.95 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP64MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 111.27 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP64MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 102.00 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 118.94 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 114.17 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 107.01 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 101.11 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 109.13 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 89.82 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 121.77 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 117.95 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 109.55 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 105.00 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 112.38 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 93.04 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 137.70 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 123.42 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 140.56 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 126.27 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 125.48 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 112.45 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 128.47 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 115.28 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 120.73 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 107.73 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 123.59 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 110.56 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvSlidingOrChunkedCausalP16MultiCtasKvGmemSepVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 131.49 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvSlidingOrChunkedCausalP16MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 120.60 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvSlidingOrChunkedCausalP16MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 132.84 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvSlidingOrChunkedCausalP16MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 121.53 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 138.55 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 125.54 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvSlidingOrChunkedCausalP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 130.71 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvSlidingOrChunkedCausalP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 111.98 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 139.73 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 125.75 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 133.52 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 111.53 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvSlidingOrChunkedCausalP32MultiCtasKvGmemSepVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 119.60 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvSlidingOrChunkedCausalP32MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.84 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvSlidingOrChunkedCausalP32MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 123.07 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvSlidingOrChunkedCausalP32MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 111.73 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 126.55 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 113.17 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvSlidingOrChunkedCausalP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 119.13 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvSlidingOrChunkedCausalP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 99.61 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 129.70 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 116.79 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 122.00 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 102.67 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvSlidingOrChunkedCausalP64MultiCtasKvGmemSepVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 115.01 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvSlidingOrChunkedCausalP64MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 104.24 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvSlidingOrChunkedCausalP64MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 118.86 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvSlidingOrChunkedCausalP64MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 107.52 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 121.68 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.70 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 114.26 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 95.02 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 125.34 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.45 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 117.50 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 98.36 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 171.84 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 163.19 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 121.65 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.86 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 168.38 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 182.88 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 161.23 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 118.05 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 128.95 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.73 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 179.49 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 165.49 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 194.91 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 180.45 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 160.70 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 149.66 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 129.33 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 113.70 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 143.81 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 123.43 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 115.58 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.45 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 156.30 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 147.57 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 115.12 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.32 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 152.43 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 167.92 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 145.29 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 111.64 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 122.54 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.20 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 164.56 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 149.62 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 179.34 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 163.55 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 145.22 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 132.62 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 122.55 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.58 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 136.76 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 117.19 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 108.38 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.20 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 149.80 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 141.05 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.44 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.65 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 146.05 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 161.41 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 138.77 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.84 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 119.62 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.52 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 157.43 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 142.62 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 172.93 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 157.16 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 138.78 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 126.36 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 119.62 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.78 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 133.96 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.52 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 105.69 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvCausalP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.53 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 176.34 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 167.34 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 125.75 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 118.69 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 172.66 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 187.53 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 165.24 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 122.27 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 133.78 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 116.55 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 183.74 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 169.48 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 199.52 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 185.11 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 169.82 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 153.11 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 133.98 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 117.88 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 149.00 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 128.52 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 122.03 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.16 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 160.85 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 151.88 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 119.34 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.15 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 157.14 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 172.81 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 149.73 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 115.86 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 127.38 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.02 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 168.52 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 153.75 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 183.86 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 168.20 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 154.59 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 136.06 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 126.95 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 111.64 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 141.83 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 122.27 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 115.01 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.05 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 154.51 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 145.50 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 116.54 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 109.35 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 150.76 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 166.06 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 143.34 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.94 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 124.59 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.22 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 161.36 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 146.73 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 177.44 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 161.94 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 147.64 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 129.80 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 124.02 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.97 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 139.15 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 119.60 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 112.20 KB
fmhaSm100fKernel_QkvE4m3OE2m1H512PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.25 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 102.62 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 82.56 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 104.21 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 86.41 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 103.24 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PackedQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:36 100.49 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PackedQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:36 89.15 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PackedQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:36 106.51 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PackedQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:36 95.10 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PackedQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:36 102.46 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PackedQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:36 90.96 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 82.71 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 104.89 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:36 104.25 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:36 92.73 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:36 110.46 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:36 98.88 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:36 106.34 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:36 94.79 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 86.57 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 113.16 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 93.25 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 114.79 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 97.23 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 118.70 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.95 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 115.16 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP16MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 109.38 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.34 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 124.87 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.75 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.37 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.12 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 80.75 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 109.52 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.96 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 113.39 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP16MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 101.55 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.35 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 115.18 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.63 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.53 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 104.30 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.63 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 119.17 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 110.74 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 98.44 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 98.65 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 104.60 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.27 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 119.08 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.91 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 100.80 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 85.82 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 99.58 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.48 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 120.73 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 115.84 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 102.29 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 103.41 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 84.89 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 72.24 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 98.77 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.65 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 104.38 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 89.14 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 80.78 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 68.22 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 115.88 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.61 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.82 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP32MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.56 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.00 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 122.05 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.34 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.96 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 109.30 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.34 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.70 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.74 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 109.17 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP32MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 98.74 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.01 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.37 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.35 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.12 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 101.48 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 78.34 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 116.19 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 107.90 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 95.66 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 95.81 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 100.07 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.97 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 114.55 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.81 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 97.97 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 82.98 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 95.03 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.18 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 117.76 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 113.12 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 99.50 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 100.70 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 83.62 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 70.84 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 97.20 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.24 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 101.55 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 86.30 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 79.36 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 66.93 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.35 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.44 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.66 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP64MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 105.02 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.84 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 120.52 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.59 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.19 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 107.77 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 78.58 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 105.16 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.45 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.00 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP64MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 97.20 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.84 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 110.96 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 80.47 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.34 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 99.95 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 77.46 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 114.65 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 106.34 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 94.26 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 94.28 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 97.76 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.80 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 112.23 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.64 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 96.43 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 81.45 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 92.84 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.01 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 116.21 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 111.57 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 98.10 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 99.17 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 82.84 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 70.06 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 96.11 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 80.47 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 100.01 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 84.89 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 78.59 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 66.16 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalPagedMultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 117.87 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.41 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalPagedMultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 111.62 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalPagedMultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.81 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalPagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.80 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 123.36 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.10 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.86 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 111.04 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.22 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalPagedMultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.69 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalPagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.41 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalPagedMultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 109.84 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalPagedMultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 100.99 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalPagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.81 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 113.67 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.11 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.89 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 103.22 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 80.10 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 116.07 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalPagedVarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 108.45 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 95.66 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalPagedVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 96.41 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalPagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 99.81 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.20 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalPagedVarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 114.59 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalPagedVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.16 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalPagedVarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 98.67 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalPagedVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 83.59 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalPagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 95.02 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalPagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.29 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 117.52 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 113.53 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 99.50 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 101.43 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 87.82 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 72.27 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 98.81 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.70 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 102.25 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 87.15 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 83.28 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 68.36 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 123.29 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.91 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 113.98 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.43 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 128.52 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 113.86 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.79 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.42 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.44 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 118.96 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 109.16 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 97.63 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 101.88 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.94 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 114.56 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.91 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 98.12 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.37 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 114.45 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 102.34 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 120.35 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.57 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 109.64 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.09 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 125.84 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 111.05 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.58 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.09 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.22 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 116.15 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 106.31 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 94.81 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 97.35 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.70 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 110.16 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.54 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 93.59 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.12 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 111.58 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 99.49 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 118.95 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.40 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.47 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.92 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 124.30 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 109.52 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.41 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.91 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.05 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.61 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 104.77 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 93.28 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 95.16 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.52 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 107.84 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.38 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 91.41 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.95 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 110.04 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 97.99 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomPagedMultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 122.22 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.63 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomPagedMultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.57 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomPagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.02 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 127.16 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomPagedMultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 113.03 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomPagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.52 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomPagedMultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 111.15 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomPagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.03 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomPagedMultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 117.48 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomPagedVarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 107.09 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomPagedVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 95.39 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomPagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 97.34 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.74 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomPagedVarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 110.07 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomPagedVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.87 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomPagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 93.59 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomPagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.16 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomPagedVarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 112.25 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvCustomPagedVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 100.23 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 118.44 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 98.51 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 120.16 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 102.50 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 115.47 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 95.85 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 117.06 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 99.72 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 113.90 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 94.33 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 115.49 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 98.31 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvDensePagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 115.21 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvDensePagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 95.74 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 116.80 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 99.73 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 126.05 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 109.65 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 120.73 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 115.94 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.55 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 132.23 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.84 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.08 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 118.38 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.48 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 116.88 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.66 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 118.95 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.12 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.68 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 122.55 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.73 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.23 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 110.56 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.33 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 130.59 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 117.88 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 112.40 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.04 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 108.13 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.91 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 123.62 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 111.36 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 106.29 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 91.00 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 102.98 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.88 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 132.17 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 122.94 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 113.96 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 110.93 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 92.31 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 76.55 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 103.93 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.77 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 109.84 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 94.35 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 87.29 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 72.05 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 123.24 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.43 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 116.39 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 113.12 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.34 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 129.42 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.43 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.66 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 115.56 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.06 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.06 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.31 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.73 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 105.18 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.34 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 119.73 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.31 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.82 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 107.75 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.04 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 127.60 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 115.02 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 109.61 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 103.20 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 103.62 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 91.67 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 119.09 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.23 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 103.34 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 88.16 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 98.49 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.63 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 129.19 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 120.23 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 111.18 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.22 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 90.73 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 75.14 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 102.36 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.35 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 107.02 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 91.64 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 85.72 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 70.77 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 121.71 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.13 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.22 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 111.59 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.04 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 127.88 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.68 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.90 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.03 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.31 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.52 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.14 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.57 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 103.77 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.17 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 118.32 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.56 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.05 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.21 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.16 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 126.06 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 113.47 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 108.21 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 101.79 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 101.29 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.50 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 116.91 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.95 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 101.80 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 86.76 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 96.15 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.46 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 127.65 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 118.70 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 109.77 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.69 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 89.39 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 74.38 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 101.27 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.59 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 105.47 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 90.11 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 84.59 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 70.00 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 125.56 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.31 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 118.39 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 115.52 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.22 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 131.05 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.71 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.07 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 117.42 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.47 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 116.50 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.32 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 116.49 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 107.82 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.23 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 121.49 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.84 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.36 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 109.73 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.45 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 128.08 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 116.55 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 110.22 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 104.36 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 104.86 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.20 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 120.51 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.54 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 105.00 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 92.59 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 99.59 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.03 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 129.54 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 121.64 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 111.91 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 109.51 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 93.00 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 76.97 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 104.74 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.30 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 107.99 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 95.33 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 87.98 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 72.72 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:36 100.49 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:36 89.15 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:36 106.63 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:36 95.23 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:36 102.59 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:36 91.09 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:36 104.38 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:36 92.85 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:36 110.59 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:36 99.00 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:36 106.47 KB
fmhaSm100fKernel_QkvE4m3OE2m1H64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:36 94.91 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 105.71 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 94.12 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 109.69 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 98.11 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 105.45 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PackedQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:36 120.28 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PackedQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:36 108.23 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PackedQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:36 126.34 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PackedQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:36 114.34 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PackedQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:36 122.23 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PackedQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:36 110.16 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 94.26 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 109.36 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:36 124.16 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:36 111.93 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:36 130.05 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:36 118.11 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:36 126.12 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:36 114.12 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 98.12 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 116.25 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 104.97 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 120.26 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 108.96 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 131.02 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.02 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 118.47 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP16MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.84 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.30 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 134.45 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.63 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.66 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 117.30 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.81 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 122.54 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.47 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 115.24 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP16MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 104.76 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.31 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 121.55 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.63 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 91.33 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 107.47 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.04 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 122.41 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 116.98 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 110.64 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 105.09 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 107.02 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.00 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 116.97 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.98 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 102.16 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 90.89 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 99.31 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.10 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 126.26 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 118.06 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 114.36 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 105.83 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 84.27 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 71.28 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 96.88 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 80.55 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 105.27 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 93.61 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 79.70 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 67.05 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 128.20 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.68 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.25 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP32MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.02 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.97 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 131.63 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.23 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.37 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.48 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 78.40 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 119.59 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.25 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.90 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP32MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 101.94 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.98 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 118.73 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.35 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.92 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 104.66 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 77.62 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 119.42 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 114.12 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 107.86 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 102.25 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 102.49 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.82 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 112.44 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.88 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 99.20 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 87.95 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 94.77 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.80 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 123.15 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 115.34 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 111.70 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 102.98 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 82.89 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 69.87 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 95.18 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.27 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 102.44 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 90.79 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 78.38 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 65.64 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 126.66 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.51 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 111.96 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP64MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 110.37 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.80 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 130.09 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.47 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.59 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 113.08 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 77.64 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 118.19 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.09 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.86 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP64MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 100.41 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.80 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 117.32 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.47 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.15 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 103.12 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 76.88 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 117.88 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 112.57 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 106.46 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 100.71 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 100.18 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.65 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 110.12 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.71 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 97.66 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 85.66 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 92.45 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 80.62 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 121.61 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 113.92 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 110.17 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 101.58 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 82.10 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 69.10 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 94.09 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 78.50 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 100.90 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 89.38 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 77.62 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 64.87 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalPagedMultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 130.19 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.60 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalPagedMultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 115.05 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalPagedMultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.15 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalPagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.64 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 132.93 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.11 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.14 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 116.35 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 80.28 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalPagedMultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 121.58 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalPagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.93 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalPagedMultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 111.70 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalPagedMultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 104.23 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalPagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.95 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 120.04 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.11 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 91.82 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.42 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.38 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 119.18 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalPagedVarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 114.58 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 107.38 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalPagedVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 102.51 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalPagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 102.41 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.06 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalPagedVarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 112.34 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalPagedVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.35 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalPagedVarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 99.90 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalPagedVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 87.63 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalPagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 94.63 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalPagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.91 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 123.04 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 116.01 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 111.10 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 103.84 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 87.39 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 71.42 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 97.04 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 80.70 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 103.02 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 90.73 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 82.18 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 67.20 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 135.50 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.40 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 116.55 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.41 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 138.23 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 126.90 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.02 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.38 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.55 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 125.29 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 115.59 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 104.10 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 104.40 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.70 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 112.19 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.78 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 97.55 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.75 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 118.27 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.66 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 132.69 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.18 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.33 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.08 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 135.41 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 124.09 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.80 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.16 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.34 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 122.48 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 112.61 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 101.29 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 99.87 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.33 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 107.66 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.41 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 93.13 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.50 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 115.40 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 103.94 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 131.16 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.88 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.16 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.91 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 133.88 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 122.55 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.51 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.86 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.04 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 121.07 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 111.07 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 99.76 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 97.56 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.16 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 105.34 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.11 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 90.82 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.33 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 113.86 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 102.44 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomPagedMultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 134.55 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.12 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomPagedMultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 113.39 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomPagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.14 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 136.87 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomPagedMultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 126.07 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomPagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.61 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomPagedMultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.97 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomPagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.15 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomPagedMultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 123.94 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomPagedVarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 113.17 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomPagedVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 101.51 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomPagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 99.78 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.48 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomPagedVarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 107.41 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomPagedVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.61 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomPagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 93.00 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomPagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.55 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomPagedVarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 115.96 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvCustomPagedVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 103.96 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 121.23 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 110.55 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 125.12 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 114.53 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 118.14 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 107.89 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 122.17 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 111.88 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 116.57 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 106.37 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 120.60 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 110.35 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvDensePagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 118.13 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvDensePagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 107.33 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 122.04 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 111.31 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 138.53 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.71 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 123.24 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 121.28 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.64 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 141.81 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.60 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.48 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 123.58 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.51 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 129.93 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 111.29 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 120.02 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 111.28 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.65 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 128.91 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.73 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.04 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 113.58 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.61 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 136.34 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 124.30 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 124.31 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.23 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 110.82 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.88 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 121.43 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.01 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 107.35 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 95.61 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 102.73 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.63 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 137.95 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 125.39 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 125.87 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 113.20 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 91.88 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 75.52 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 102.03 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.67 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 110.89 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 98.07 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 85.94 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 70.91 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 135.59 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.50 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 118.91 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 118.47 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.30 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 138.99 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.19 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.07 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 120.77 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.09 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 127.12 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.95 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 115.68 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.46 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.44 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 126.09 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.32 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.62 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 110.77 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.20 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 133.36 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 121.45 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 121.52 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 109.51 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 106.28 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 91.63 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 116.89 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.76 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 104.52 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 92.67 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 98.23 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.39 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 134.84 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 122.69 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 123.09 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 110.48 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 90.30 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 74.10 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 100.34 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.26 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 108.06 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 95.26 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 84.48 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 69.49 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 134.18 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.20 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 116.73 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 116.94 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.13 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 137.46 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.44 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.30 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 119.35 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.34 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 125.58 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.77 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 113.51 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.93 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.14 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 124.69 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.56 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.86 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 109.35 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 80.45 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 131.82 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 119.90 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 120.12 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 107.98 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 103.98 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.34 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 114.59 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.47 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 102.98 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 91.14 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 95.89 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.09 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 133.30 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 121.15 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 121.69 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.95 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 89.09 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 73.45 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 99.24 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.62 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 106.52 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 93.72 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 83.38 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 68.72 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 138.04 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.38 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 121.16 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 120.85 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.05 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 140.75 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.60 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.48 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 122.74 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.63 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 129.55 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.83 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 117.81 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 110.86 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.31 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 127.86 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.85 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.16 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.62 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.73 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 133.74 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 122.77 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 122.26 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 110.67 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 107.70 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.04 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 118.12 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.19 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 106.19 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 93.84 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 99.46 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.67 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 135.47 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 123.84 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 123.82 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 111.77 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 92.56 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 76.18 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 102.72 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.34 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 108.82 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 96.55 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 86.62 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 71.45 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128SeparateQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 105.57 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:36 120.41 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:36 108.35 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:36 126.47 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:36 114.46 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:36 122.36 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:36 110.29 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128SeparateQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 94.26 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 109.36 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:36 124.16 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:36 112.05 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:36 130.17 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:36 118.23 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:36 126.24 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:36 114.24 KB
fmhaSm100fKernel_QkvE4m3OE4m3H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 98.12 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 101.70 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 89.72 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 104.46 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 92.44 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 102.64 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 90.42 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 105.63 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 93.14 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 113.41 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 100.52 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 116.10 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 103.38 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 140.62 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 124.66 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 135.71 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP16MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 118.65 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 117.98 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 143.27 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.06 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.59 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 120.84 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.14 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 122.52 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 121.78 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 130.77 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP16MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.70 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 116.22 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 125.20 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.85 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.61 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.55 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.42 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 123.90 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 119.51 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 111.38 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 105.24 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 122.39 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.84 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 133.48 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 121.38 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 103.45 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 90.41 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 112.59 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.20 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 126.61 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 122.17 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 113.97 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.62 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 95.30 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.24 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 111.29 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.83 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 105.24 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 93.32 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 90.53 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 77.62 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 133.61 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 117.30 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 129.33 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP32MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 111.76 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.75 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 137.37 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.98 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.52 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.92 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.94 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 115.51 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.84 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 124.26 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP32MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 99.70 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 109.16 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 119.42 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 91.77 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.53 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 102.64 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.34 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 116.89 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 111.88 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 105.22 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 99.21 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 115.36 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.26 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 126.97 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 113.78 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 96.59 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 84.27 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 105.73 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.50 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 119.59 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 115.54 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 107.81 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 102.98 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 92.23 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.15 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 107.05 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.62 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 99.62 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 87.59 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 87.45 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 74.52 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 130.67 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.23 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 126.26 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP64MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.70 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.55 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 134.42 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.46 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.97 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.11 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.54 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.57 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 111.66 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 121.20 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP64MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 96.75 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.09 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 116.48 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.25 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.99 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 99.70 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.82 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 113.80 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 108.80 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 102.15 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 96.27 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 112.15 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.20 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 123.77 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.59 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 93.36 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 81.32 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 102.50 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.55 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 116.52 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 112.59 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 104.87 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 100.05 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 90.66 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 77.61 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 105.50 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.21 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 96.64 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 84.65 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 85.88 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 72.98 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalPagedMultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 132.68 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 116.26 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalPagedMultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 128.23 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalPagedMultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 110.83 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalPagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 109.82 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 136.56 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.27 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.78 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.02 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.22 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalPagedMultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.70 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalPagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 113.92 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalPagedMultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 123.30 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalPagedMultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 98.77 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalPagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.11 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 118.62 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.94 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.80 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 101.86 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.62 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 116.83 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalPagedVarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 110.07 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 103.89 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalPagedVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 97.85 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalPagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 113.63 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.57 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalPagedVarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 125.24 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalPagedVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.23 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalPagedVarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 95.26 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalPagedVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 83.00 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalPagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 103.91 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalPagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.06 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 119.65 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 113.96 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 106.61 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 101.84 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 95.93 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.39 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 107.54 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.97 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 98.48 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 86.34 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 90.14 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 74.87 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 126.53 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 112.30 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 129.52 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 115.14 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 118.41 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 105.89 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 121.39 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 108.85 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 115.32 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 102.95 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 118.19 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 105.91 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvDensePagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 116.84 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvDensePagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 104.46 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 119.83 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 107.42 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 148.05 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 129.19 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 140.36 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 124.02 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 122.25 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 150.67 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.16 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.30 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 125.94 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.84 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 130.08 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 126.20 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 135.27 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 111.98 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 120.39 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 132.76 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.95 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.32 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 113.94 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.12 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 139.81 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 127.00 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 123.39 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 113.32 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 126.79 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 109.66 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 138.55 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 126.13 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 108.34 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 96.10 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 120.04 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.64 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 140.78 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 129.09 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 125.95 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 116.38 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 103.77 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.45 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 116.60 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.82 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 110.38 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 98.81 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 97.59 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.46 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 141.04 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 122.00 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 133.98 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 117.01 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 115.19 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 144.88 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.09 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 109.08 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 120.16 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.66 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 123.06 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 119.54 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 128.88 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 104.97 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 113.60 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 126.98 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.76 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.23 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.15 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 91.05 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 131.91 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 119.20 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 116.62 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.68 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 119.30 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.96 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 131.69 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 118.54 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 101.46 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 89.46 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 112.61 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.05 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 132.85 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 122.73 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 119.31 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 110.44 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 99.84 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.35 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 112.36 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.74 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 104.60 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 92.79 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 93.67 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 78.37 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 138.10 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 118.93 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 130.91 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.06 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.00 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 141.95 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.55 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.55 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 117.21 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 91.23 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 120.12 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 116.34 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 125.81 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 102.03 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.54 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 124.03 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.34 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.70 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 105.21 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.52 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 128.82 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 115.99 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 113.69 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 103.73 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 116.23 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.02 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 128.61 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 115.47 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 98.23 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 86.52 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 109.52 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.99 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 129.65 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 119.77 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 116.37 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 107.47 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 98.30 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.94 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 110.82 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.20 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 101.77 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 89.97 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 92.13 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 76.82 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 140.70 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 121.20 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 133.35 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 116.66 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.38 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 144.54 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.02 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 111.16 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 119.97 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.57 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 122.72 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 118.73 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 129.18 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 104.62 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.80 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 126.76 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.68 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.05 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 107.96 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.10 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 131.05 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 118.04 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 115.87 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 105.90 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 118.20 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.34 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 130.44 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 117.30 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 100.99 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 88.80 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 108.39 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.97 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 131.99 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 121.80 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 118.67 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 109.60 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 97.89 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.23 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 113.26 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.61 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 104.33 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 92.11 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 92.60 KB
fmhaSm100fKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.24 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 110.55 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 97.67 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 113.31 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 100.24 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 111.49 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 98.27 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 114.61 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 101.10 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP16MultiCtasKvGmemSepVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 131.16 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP16MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 115.30 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP16MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 132.58 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP16MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 116.02 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 140.88 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 135.40 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 126.51 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 121.39 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 120.29 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.66 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 143.59 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 136.05 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 129.09 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 122.34 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 120.48 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.91 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP32MultiCtasKvGmemSepVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 119.28 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP32MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 103.54 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP32MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 122.80 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP32MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.34 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 128.91 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 123.31 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 115.87 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 109.88 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 108.67 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 95.15 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 131.62 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 126.59 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 118.32 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 113.52 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 111.44 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 98.09 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP64MultiCtasKvGmemSepVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.69 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP64MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 98.95 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP64MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 118.59 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP64MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 102.00 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 124.04 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 118.44 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 111.16 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 105.16 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 103.78 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 90.43 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 126.75 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 122.22 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 113.60 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 109.05 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 107.05 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 93.65 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 141.83 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 127.55 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 144.70 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 130.39 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 129.82 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 116.57 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 132.69 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 119.41 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 124.95 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 111.85 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 127.94 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 114.69 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvSlidingOrChunkedCausalP16MultiCtasKvGmemSepVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 138.62 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvSlidingOrChunkedCausalP16MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 120.60 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvSlidingOrChunkedCausalP16MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 140.20 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvSlidingOrChunkedCausalP16MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 121.53 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 142.74 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 129.59 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvSlidingOrChunkedCausalP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 125.42 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvSlidingOrChunkedCausalP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.59 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 143.92 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 129.80 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 125.61 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.27 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvSlidingOrChunkedCausalP32MultiCtasKvGmemSepVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 126.86 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvSlidingOrChunkedCausalP32MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.84 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvSlidingOrChunkedCausalP32MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 130.55 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvSlidingOrChunkedCausalP32MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 111.73 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 130.95 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 117.34 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvSlidingOrChunkedCausalP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 113.66 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvSlidingOrChunkedCausalP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 100.34 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 134.09 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 120.84 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 116.55 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 103.41 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvSlidingOrChunkedCausalP64MultiCtasKvGmemSepVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 122.27 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvSlidingOrChunkedCausalP64MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 104.24 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvSlidingOrChunkedCausalP64MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 126.21 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvSlidingOrChunkedCausalP64MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 107.52 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 126.07 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.76 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 108.79 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 95.75 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 129.60 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 116.50 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 112.05 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 98.98 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 170.67 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 161.80 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 120.16 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 113.23 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 166.82 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 179.76 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 159.55 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 116.49 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 127.26 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 111.07 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 167.19 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 150.18 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 184.04 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 169.59 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 155.13 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 143.22 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 117.62 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.78 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 132.83 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.53 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 110.02 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.37 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 154.77 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 145.91 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 113.62 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.82 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 150.87 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 164.80 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 143.60 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.08 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 120.85 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.66 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 152.13 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 133.35 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 168.90 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 153.64 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 139.78 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 126.41 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 110.64 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.66 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 125.91 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.30 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 103.16 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.25 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 148.26 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 139.40 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.94 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.02 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 144.35 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 158.42 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 137.21 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.28 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 118.06 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.86 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 145.71 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 126.97 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 162.73 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 146.82 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 133.22 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 120.02 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 107.98 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.87 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 122.98 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.49 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 100.37 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvCausalP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.45 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 175.00 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 165.74 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 124.25 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 117.06 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 171.09 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 184.41 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 163.69 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 120.71 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 132.09 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.90 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 171.55 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 153.98 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 191.38 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 174.22 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 163.66 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 146.66 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 123.16 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.08 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 138.15 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 119.54 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 116.46 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.20 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 159.48 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 150.35 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 117.84 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.65 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 155.57 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 169.45 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 148.04 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.17 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 125.69 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.49 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 156.33 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 137.16 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 176.42 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 158.48 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 148.44 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 129.86 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 116.12 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.96 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 131.09 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 113.30 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 109.55 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.09 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 152.97 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 143.84 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 115.04 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.85 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 149.06 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 163.07 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 141.66 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 111.38 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 122.90 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.69 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 149.79 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 130.77 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 169.30 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 151.53 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 142.15 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 123.59 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 113.32 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.16 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 128.30 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.62 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 106.76 KB
fmhaSm100fKernel_QkvE4m3OE4m3H512PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 91.30 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 94.76 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 82.56 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 98.73 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 86.55 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 94.37 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PackedQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:36 104.05 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PackedQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:36 92.33 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PackedQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:36 110.09 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PackedQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:36 98.28 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PackedQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:36 106.02 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PackedQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:36 94.14 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 82.72 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 98.41 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:36 107.93 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:36 96.04 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:36 114.04 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:36 102.18 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:36 110.02 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:36 98.10 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 86.70 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 105.30 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 93.38 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 109.31 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 97.37 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 117.20 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.41 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 113.87 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP16MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 107.71 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.69 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 123.24 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.12 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.88 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 110.46 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.16 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.35 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.30 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.05 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP16MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 99.98 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.82 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 113.49 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.13 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.84 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 102.73 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 78.13 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 111.43 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 106.02 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 98.45 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 93.95 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 102.81 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.36 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 115.09 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.88 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 97.91 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 82.94 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 97.86 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.62 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 115.27 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 110.98 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 102.30 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 98.56 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 83.34 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 70.53 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 95.10 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 78.88 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 101.49 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 86.25 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 79.09 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 66.52 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.38 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.08 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 109.52 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP32MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 104.90 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.34 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 120.43 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 80.72 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.46 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 107.65 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 77.74 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 105.53 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.09 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.84 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP32MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 97.05 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.48 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 110.80 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.73 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.43 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 99.92 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 76.72 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 108.32 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 103.16 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 95.66 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 91.12 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 98.28 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.18 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 110.69 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.77 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 95.09 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 80.10 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 93.31 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.45 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 112.28 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 108.26 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 99.63 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 95.85 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 81.95 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 69.12 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 93.41 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 77.59 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 98.66 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 83.54 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 77.66 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 65.10 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.85 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.91 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.35 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP64MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 103.37 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.17 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 119.02 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.96 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.69 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.11 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 76.98 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 104.00 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.79 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.54 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP64MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 95.52 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.30 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 109.27 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 78.97 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.66 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 98.52 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 75.96 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 106.77 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 101.61 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 94.27 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 89.59 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 95.97 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.88 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 108.38 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.57 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 93.54 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 78.56 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 91.00 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.15 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 110.73 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 106.72 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 98.11 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 94.31 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 81.16 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 68.35 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 92.31 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 76.83 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 97.12 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 82.01 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 76.89 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 64.33 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalPagedMultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 116.38 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.88 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalPagedMultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.33 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalPagedMultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 107.16 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalPagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.27 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 121.86 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.48 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 91.23 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 109.38 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.62 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalPagedMultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 107.52 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalPagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.88 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalPagedMultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.51 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalPagedMultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 99.43 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalPagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.45 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 111.98 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.61 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.33 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 101.70 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 78.60 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 108.20 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalPagedVarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 103.72 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 95.79 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalPagedVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 91.71 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalPagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 98.15 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.41 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalPagedVarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 110.60 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalPagedVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.25 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalPagedVarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 95.78 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalPagedVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 80.70 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalPagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 93.18 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalPagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.55 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 112.04 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 108.80 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 99.63 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 96.57 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 86.16 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 70.43 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 95.27 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.02 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 99.36 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 84.27 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 81.34 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 66.55 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 121.55 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.52 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.47 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.92 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 127.02 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.38 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.41 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.88 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.93 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 117.27 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 104.43 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 92.93 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 100.00 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.96 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 110.16 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.67 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 96.19 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.27 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 110.36 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 97.80 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 118.73 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.19 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.12 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.70 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 124.21 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 109.56 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.07 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.67 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.71 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.46 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 101.46 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 90.12 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 95.47 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.72 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 105.62 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.17 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 91.66 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.90 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 107.49 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 95.08 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 117.20 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.02 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.95 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.41 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 122.68 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.03 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.90 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.38 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 91.54 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 113.05 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 99.91 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 88.59 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 93.16 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.42 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 103.31 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 91.13 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 89.47 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 77.73 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 105.95 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 93.58 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomPagedMultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 120.72 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.00 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomPagedMultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 109.06 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomPagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.64 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 125.67 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomPagedMultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 111.55 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomPagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.01 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomPagedMultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 109.61 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomPagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.52 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomPagedMultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 115.80 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomPagedVarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 102.24 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomPagedVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 90.70 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomPagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 95.34 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.77 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomPagedVarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 105.66 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomPagedVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.50 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomPagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 91.52 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomPagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 80.07 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomPagedVarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 108.04 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvCustomPagedVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 95.73 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 109.14 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 98.64 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 113.13 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 102.63 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 106.18 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 95.86 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 110.17 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 99.85 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 104.61 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 94.46 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 108.60 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 98.32 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvDensePagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 106.05 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvDensePagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 95.75 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 110.04 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 99.74 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 124.55 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.11 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 119.07 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.28 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.02 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 130.61 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.09 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.58 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 116.84 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.85 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 115.87 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.12 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 117.25 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.55 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.15 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 120.73 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.23 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.55 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.99 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.83 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 125.24 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 113.15 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 112.41 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 101.34 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 106.47 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.12 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 119.52 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.43 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 103.41 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 88.12 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 101.13 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.02 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 126.82 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 118.09 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 114.09 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.08 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 90.65 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 74.77 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 100.59 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.06 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 106.95 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 91.46 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 85.23 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 70.24 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 121.62 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.77 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.73 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 111.47 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.68 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 127.80 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.68 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.16 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.03 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.44 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 113.05 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.78 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 113.04 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 103.62 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.81 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 117.92 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.81 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.13 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.18 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 80.41 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 122.13 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 110.30 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 109.74 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 98.50 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 101.96 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.88 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 114.98 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.19 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 100.45 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 85.41 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 96.62 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.77 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 123.84 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 115.51 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 111.31 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 103.37 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 89.07 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 73.35 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 99.02 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.65 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 104.12 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 88.75 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 83.64 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 68.95 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 120.21 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.60 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.55 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 109.94 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.51 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 126.39 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.93 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.40 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.50 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 80.69 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 111.52 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.61 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.87 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 102.08 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.64 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 116.51 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.06 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 91.37 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 104.65 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.66 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 120.59 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 108.74 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 108.22 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 96.97 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 99.62 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.71 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 112.67 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.89 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 98.91 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 83.87 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 94.30 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.60 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 122.30 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 113.97 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 109.78 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 101.84 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 87.85 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 72.59 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 97.93 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.88 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 102.59 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 87.22 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 82.55 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 68.19 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 124.06 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.66 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 116.73 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 113.98 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.69 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 129.55 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.09 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.58 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 115.88 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.98 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 115.37 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.79 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.80 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.12 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.70 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 119.80 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.23 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.67 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.16 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.95 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 122.61 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 111.82 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 110.35 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 99.66 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 103.20 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 91.41 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 116.52 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.61 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 102.11 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 89.70 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 97.73 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.30 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 124.31 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 116.79 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 111.91 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 104.66 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 91.19 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 75.30 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 101.41 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.60 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 105.11 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 92.45 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 85.91 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 70.91 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:36 104.17 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:36 92.33 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:36 110.21 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:36 98.41 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:36 106.14 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:36 94.27 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:36 108.05 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:36 96.16 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:36 114.16 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:36 102.30 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:36 110.15 KB
fmhaSm100fKernel_QkvE4m3OE4m3H64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:36 98.23 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 100.47 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 88.69 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 104.45 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 92.67 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 100.08 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PackedQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:36 107.20 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PackedQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:36 94.48 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PackedQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:36 113.27 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PackedQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:36 100.59 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PackedQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:36 109.16 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PackedQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:36 96.41 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 88.70 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 103.99 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:36 110.96 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:36 98.30 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:36 117.09 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:36 104.36 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:36 113.04 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:36 100.37 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 92.55 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 110.88 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 99.53 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 114.89 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 103.52 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 131.12 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.00 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 118.45 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP16MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.83 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.29 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 134.43 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.62 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.77 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 117.41 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.80 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 122.52 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.58 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 115.35 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP16MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 104.87 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.43 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 121.66 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.74 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 91.32 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 107.58 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.02 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 117.04 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 117.94 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 105.12 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 105.97 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 108.88 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.38 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 120.66 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.00 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 103.52 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 91.15 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 99.73 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.41 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 121.02 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 120.11 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 108.97 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.98 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 88.59 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 72.29 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 99.29 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.31 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 106.77 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 93.87 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 80.05 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 67.28 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 128.19 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.79 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.24 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP32MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.01 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.08 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 131.74 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.21 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.35 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.59 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 78.51 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 119.70 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.23 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 111.01 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP32MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 101.92 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.09 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 118.84 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.34 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.91 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 104.77 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 77.73 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 114.05 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 115.08 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 102.34 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 103.26 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 104.35 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.20 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 116.12 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.91 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 100.57 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 88.21 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 95.20 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.12 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 117.91 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 117.39 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 106.31 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.14 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 86.89 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 70.88 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 97.59 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 80.90 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 103.94 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 91.05 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 78.64 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 65.88 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 126.77 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.62 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.07 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP64MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 110.48 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.78 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 130.20 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.45 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.59 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 113.06 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 77.62 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 118.17 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.07 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.84 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP64MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 100.52 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.91 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 117.30 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.58 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.13 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 103.23 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 76.86 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 112.52 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 113.53 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 100.95 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 101.73 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 102.16 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.02 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 113.81 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.73 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 99.03 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 86.67 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 92.88 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 80.95 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 116.37 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 115.84 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 104.79 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 104.73 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 85.80 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 70.11 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 96.50 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 80.13 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 102.40 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 89.64 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 77.87 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 65.10 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalPagedMultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 130.17 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.59 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalPagedMultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 115.16 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalPagedMultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.14 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalPagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.75 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 133.04 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.09 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.25 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 116.46 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 80.27 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalPagedMultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 121.70 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalPagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.04 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalPagedMultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 111.81 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalPagedMultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 104.34 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalPagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.05 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 120.15 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.22 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 91.80 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.41 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.37 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 113.81 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalPagedVarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 115.66 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 101.82 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalPagedVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 103.38 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalPagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 104.27 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.43 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalPagedVarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 116.02 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalPagedVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.25 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalPagedVarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 101.27 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalPagedVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 89.06 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalPagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 95.06 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalPagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.23 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 117.67 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 117.93 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 105.66 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.18 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 88.51 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 72.43 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 99.46 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.24 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 104.52 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 92.04 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 82.41 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 67.45 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 135.48 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.51 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 116.53 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.40 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 138.21 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 127.01 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.12 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.48 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.66 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 125.41 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 116.68 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 104.98 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 105.96 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.98 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 114.87 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.57 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 97.95 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.05 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 118.72 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.06 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 132.67 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.16 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.31 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.19 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 135.39 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 124.20 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.79 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.14 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.32 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 122.59 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 113.82 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 102.17 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 101.43 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.61 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 110.34 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.07 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 93.41 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.80 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 115.85 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 105.34 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 131.14 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.99 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.15 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.89 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 133.98 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 122.66 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.62 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.97 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.15 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 121.05 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 112.28 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 100.77 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 99.12 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.44 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 108.02 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.90 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 91.23 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.63 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 114.31 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 103.84 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomPagedMultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 134.66 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.10 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomPagedMultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 113.38 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomPagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.12 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 136.85 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomPagedMultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 126.05 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomPagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.59 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomPagedMultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 111.08 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomPagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.13 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomPagedMultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 123.92 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomPagedVarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 114.26 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomPagedVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 102.38 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomPagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 101.34 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.78 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomPagedVarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 110.09 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomPagedVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.40 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomPagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 93.41 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomPagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.85 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomPagedVarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 116.41 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvCustomPagedVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 105.62 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 115.85 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 105.16 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 119.76 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 109.14 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 112.90 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 102.50 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 116.80 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 106.36 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 111.33 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 100.98 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 115.23 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 104.96 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvDensePagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 112.77 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvDensePagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 101.89 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 116.66 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 105.88 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 138.52 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.82 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 123.23 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 121.39 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.62 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 141.92 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.71 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.47 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 123.69 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.62 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 130.04 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 111.27 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 120.00 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 111.39 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.76 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 128.90 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.84 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.02 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 113.69 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.72 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 130.98 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 125.39 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 118.88 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.97 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 112.55 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.27 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 125.11 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 111.91 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 108.85 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 96.91 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 103.14 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.91 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 132.58 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 127.33 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 120.43 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 115.74 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 93.12 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 76.52 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 104.33 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.29 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 112.27 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 99.38 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 86.31 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 71.14 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 135.70 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.48 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 119.02 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 118.45 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.41 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 138.98 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.30 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.06 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 120.88 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.21 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 127.10 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.05 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 115.79 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.45 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.42 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 126.20 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.43 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.73 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 110.88 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.31 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 127.99 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 122.53 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 116.09 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 110.26 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 108.02 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.02 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 120.57 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.66 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 105.90 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 93.98 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 98.52 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.67 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 129.47 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 124.73 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 117.77 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.98 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 91.42 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 75.23 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 102.75 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.88 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 109.44 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 96.56 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 84.74 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 69.73 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 134.16 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.31 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 116.84 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 116.92 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.12 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 137.57 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.55 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.29 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 119.34 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.45 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 125.69 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.76 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 113.62 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.91 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.25 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 124.67 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.67 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.84 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 109.34 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 80.55 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 126.45 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 120.98 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 114.69 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.72 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 105.83 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.85 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 118.27 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.48 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 104.35 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 92.57 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 96.30 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.50 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 127.93 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 123.20 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 116.25 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 111.45 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 90.20 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 74.46 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 101.66 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.10 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 107.89 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 95.02 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 83.63 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 68.96 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 138.15 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.49 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 121.15 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 120.96 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.16 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 140.86 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.71 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.59 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 122.85 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.62 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 129.54 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.81 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 117.80 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 110.97 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.43 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 127.84 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.84 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.15 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.73 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.72 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 128.50 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 123.86 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 116.70 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 111.54 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 109.55 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.43 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 121.80 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 109.09 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 107.56 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 95.15 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 99.87 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.95 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 130.10 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 125.90 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 118.26 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.32 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 93.68 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 77.19 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 105.27 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.95 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 110.20 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 97.85 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 86.88 KB
fmhaSm100fKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 71.68 KB
fmhaSm100fKernel_QkvE4m3OFp16H128SeparateQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 100.20 KB
fmhaSm100fKernel_QkvE4m3OFp16H128SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:36 107.33 KB
fmhaSm100fKernel_QkvE4m3OFp16H128SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:36 94.60 KB
fmhaSm100fKernel_QkvE4m3OFp16H128SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:36 113.39 KB
fmhaSm100fKernel_QkvE4m3OFp16H128SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:36 100.59 KB
fmhaSm100fKernel_QkvE4m3OFp16H128SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:36 109.28 KB
fmhaSm100fKernel_QkvE4m3OFp16H128SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:36 96.41 KB
fmhaSm100fKernel_QkvE4m3OFp16H128SeparateQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 88.82 KB
fmhaSm100fKernel_QkvE4m3OFp16H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 103.99 KB
fmhaSm100fKernel_QkvE4m3OFp16H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:36 111.09 KB
fmhaSm100fKernel_QkvE4m3OFp16H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:36 98.43 KB
fmhaSm100fKernel_QkvE4m3OFp16H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:36 117.22 KB
fmhaSm100fKernel_QkvE4m3OFp16H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:36 104.36 KB
fmhaSm100fKernel_QkvE4m3OFp16H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:36 113.16 KB
fmhaSm100fKernel_QkvE4m3OFp16H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:36 100.37 KB
fmhaSm100fKernel_QkvE4m3OFp16H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 92.68 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 96.61 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 85.16 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 99.37 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 87.76 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 97.55 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 85.75 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 100.54 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 88.59 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 108.32 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 95.84 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 110.88 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 98.70 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 127.53 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 124.65 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 135.82 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP16MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 117.46 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 118.09 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 130.12 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.17 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.70 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 119.65 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.12 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 115.18 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 121.89 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 130.88 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP16MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.70 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 116.33 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 117.64 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.97 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.72 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.53 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.53 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 118.81 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 114.62 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 106.70 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 100.75 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 124.98 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.73 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 138.85 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 124.70 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 105.10 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 92.07 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 113.05 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.55 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 121.52 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 117.16 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 109.41 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 104.13 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 100.84 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.80 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 115.82 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.43 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 106.90 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 94.97 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 90.86 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 77.95 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 120.65 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 117.41 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 129.31 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP32MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 110.57 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.73 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 124.24 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.97 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.62 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 113.73 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.05 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.17 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.83 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 124.37 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP32MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 99.68 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 109.27 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 111.89 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 91.76 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.52 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 102.75 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.33 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 111.80 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 106.98 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 100.54 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 94.72 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 117.82 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.29 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 131.77 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 118.43 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 98.25 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 86.04 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 106.08 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.97 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 114.50 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 110.51 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 103.26 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 98.37 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 97.09 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 80.71 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 111.39 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.35 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 101.27 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 89.24 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 87.66 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 74.85 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 117.58 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.23 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 126.37 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP64MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 107.51 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.66 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 121.42 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.57 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.08 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 110.92 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.52 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 105.23 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 111.77 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 121.30 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP64MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 96.73 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.08 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.95 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.36 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.98 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 99.80 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.93 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 108.71 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 103.77 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 97.60 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 91.65 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 114.73 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.10 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 128.57 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 115.36 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 95.01 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 82.97 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 102.98 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.91 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 111.42 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 107.55 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 100.19 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 95.55 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 95.42 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.17 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 109.76 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.80 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 98.30 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 86.30 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 86.09 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 73.31 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalPagedMultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 119.59 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 116.37 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalPagedMultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 128.22 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalPagedMultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 109.64 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalPagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 109.81 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 123.44 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.38 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.77 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.84 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.20 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalPagedMultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 107.37 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalPagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 113.91 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalPagedMultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 123.41 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalPagedMultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 98.75 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalPagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.22 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 111.21 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.05 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.91 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 101.98 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.73 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 111.73 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalPagedVarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 105.04 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 99.21 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalPagedVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 93.36 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalPagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 116.22 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.10 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalPagedVarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 129.92 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalPagedVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 115.17 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalPagedVarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 96.91 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalPagedVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 84.77 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalPagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 104.38 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalPagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.41 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 114.55 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 108.93 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 102.05 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 97.35 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 97.71 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 80.95 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 111.88 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.55 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 100.12 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 87.98 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 90.36 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 75.22 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 121.58 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 107.62 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 124.45 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 110.59 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 113.31 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 101.21 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 116.30 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 104.17 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 110.23 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 98.27 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 113.09 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 101.23 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvDensePagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 111.75 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvDensePagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 99.91 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 114.73 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 102.74 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 134.99 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 129.30 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 140.47 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 122.83 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 122.36 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 137.71 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.27 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.28 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 124.75 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.83 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 122.77 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 126.31 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 135.38 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.09 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 120.50 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 125.23 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.05 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.43 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.05 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.23 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 134.74 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 122.09 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 118.71 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.83 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 129.38 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.96 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 143.80 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 128.30 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 109.98 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 97.88 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 120.39 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.98 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 135.70 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 124.05 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 121.40 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 111.88 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 105.73 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.04 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 121.14 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.51 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 112.02 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 100.59 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 97.95 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.79 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 128.11 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 122.11 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 133.96 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 115.82 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 115.17 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 131.83 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.08 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 109.19 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 118.97 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.77 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 115.75 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 119.52 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 128.86 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 105.08 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 113.71 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 119.48 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.87 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.21 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.13 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 91.04 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 126.81 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 114.15 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 112.07 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 102.19 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 121.89 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.65 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 136.21 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 121.91 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 103.11 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 91.23 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 113.08 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.41 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 127.64 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 117.80 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 114.63 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 105.82 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 101.62 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.95 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 116.70 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.43 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 106.26 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 94.56 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 93.91 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 78.70 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 125.16 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 118.91 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 131.02 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.88 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.11 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 129.02 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.54 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.66 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 116.02 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 91.22 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.81 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 116.46 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 125.92 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 102.14 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.65 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 116.54 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.33 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.81 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 105.32 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.62 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 123.61 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 110.95 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 109.13 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 99.25 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 118.80 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.59 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 132.87 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 118.97 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 99.89 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 88.29 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 109.88 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.46 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 124.55 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 114.73 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 111.69 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 102.98 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 99.95 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.41 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 115.16 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.89 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 103.42 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 91.62 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 92.37 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 77.16 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 127.90 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 121.18 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 133.46 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 115.47 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.37 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 131.66 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.12 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 111.14 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 118.78 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.68 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 115.55 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 118.84 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 129.16 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 104.73 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.78 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 119.42 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.79 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.16 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 107.95 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.09 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 125.95 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 113.12 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 111.31 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 101.41 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 120.78 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.28 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 135.47 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 120.79 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 102.64 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 90.45 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 108.87 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.31 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 126.90 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 116.88 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 113.99 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 105.12 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 102.52 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.83 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 117.50 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.17 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 105.98 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 93.76 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 92.94 KB
fmhaSm100fKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.58 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 105.52 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 93.18 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 108.41 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 95.75 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 106.59 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 93.79 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 109.58 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 96.62 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP16MultiCtasKvGmemSepVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 123.91 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP16MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 115.30 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP16MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 125.25 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP16MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 116.02 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 135.19 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 130.37 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 122.02 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 116.90 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 122.07 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.31 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 138.01 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 131.04 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 124.59 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 117.85 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 122.13 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.56 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP32MultiCtasKvGmemSepVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.14 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP32MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 103.54 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP32MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 115.48 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP32MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.34 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 123.11 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 118.41 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 111.26 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 105.38 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 110.33 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 96.80 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 125.95 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 121.56 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 113.83 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.90 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 113.09 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 99.73 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP64MultiCtasKvGmemSepVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 107.42 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP64MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 98.95 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP64MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 111.27 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP64MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 102.00 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 118.23 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 113.53 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 106.54 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 100.67 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 105.44 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 92.09 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 121.06 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 117.19 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 109.11 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 104.55 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 108.70 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 95.43 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 136.95 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 123.08 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 139.81 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 125.91 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 124.91 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 112.09 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 127.78 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 114.93 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 120.04 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 107.38 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 122.91 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 110.22 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP16MultiCtasKvGmemSepVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 131.49 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP16MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 120.60 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP16MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 132.84 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP16MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 121.53 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 137.73 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 125.09 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 127.07 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.23 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 138.91 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 125.30 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 127.26 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 113.92 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP32MultiCtasKvGmemSepVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 119.60 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP32MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.84 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP32MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 123.07 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP32MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 111.73 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 125.91 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.85 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 115.31 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 101.99 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 129.06 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 116.35 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 118.20 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 105.05 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP64MultiCtasKvGmemSepVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 115.01 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP64MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 104.24 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP64MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 118.86 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP64MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 107.52 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 121.04 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.27 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 110.44 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 97.41 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 124.70 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.01 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 113.71 KB
fmhaSm100fKernel_QkvE4m3OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 100.62 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 170.66 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 161.79 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 120.14 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 113.34 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 166.80 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 179.88 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 159.66 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 116.60 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 127.38 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 111.19 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 171.10 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 155.67 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 191.05 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 176.98 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 155.81 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 143.79 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 121.31 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.23 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 142.03 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 118.92 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 110.45 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.79 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 154.76 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 146.02 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 113.73 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.80 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 150.85 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 164.91 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 143.71 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.06 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 120.84 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.65 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 156.17 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 139.81 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 175.65 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 160.98 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 140.33 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 126.98 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 114.22 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.12 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 135.25 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.80 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 103.46 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.68 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 148.37 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 139.38 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.93 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.13 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 144.34 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 158.41 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 137.20 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.39 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 118.05 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.97 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 149.63 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 133.55 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 169.23 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 153.82 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 133.90 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 120.72 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 111.44 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.32 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 132.33 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.01 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 100.79 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvCausalP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.88 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 174.98 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 165.85 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 124.36 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 117.17 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 171.08 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 184.52 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 163.67 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 120.70 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 132.21 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 115.02 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 175.47 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 159.81 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 195.45 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 179.79 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 164.36 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 147.22 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 125.92 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.43 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 147.48 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 124.09 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 116.89 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.63 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 159.47 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 150.34 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 117.82 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.76 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 155.55 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 169.57 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 148.15 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.28 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 125.80 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.48 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 160.24 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 143.95 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 180.43 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 162.45 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 149.13 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 130.53 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 118.89 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.32 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 140.32 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 117.84 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 109.86 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.39 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 153.08 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 143.95 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 115.02 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.84 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 149.17 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 163.05 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 141.64 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 111.48 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 123.01 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.67 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 153.16 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 137.70 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 174.02 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 156.20 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 142.71 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 124.15 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 116.09 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.52 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 137.52 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 115.17 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 107.05 KB
fmhaSm100fKernel_QkvE4m3OFp16H512PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 91.72 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 95.64 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 83.45 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 99.62 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 87.43 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 95.25 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PackedQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:36 111.12 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PackedQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:36 87.59 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PackedQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:36 113.70 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PackedQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:36 93.66 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PackedQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:36 113.12 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PackedQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:36 89.53 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 83.60 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 99.30 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:36 111.66 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:36 91.30 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:36 117.52 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:36 97.45 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:36 113.89 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:36 93.36 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 87.59 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 106.18 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 94.27 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 110.20 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 98.25 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 117.18 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.40 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 113.85 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP16MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 107.82 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.67 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 123.35 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.11 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.86 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 110.57 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.14 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.46 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.41 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.16 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP16MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 99.97 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.80 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 113.60 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.12 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.83 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 102.72 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 78.12 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 112.31 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 107.70 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 99.33 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 94.93 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 102.98 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.56 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 116.23 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.98 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 98.84 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 83.66 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 98.01 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.90 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 116.15 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 112.11 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 103.18 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 99.70 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 83.32 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 70.62 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 95.62 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.62 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 102.29 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 86.98 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 79.21 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 66.65 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.37 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.19 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 109.51 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP32MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 104.88 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.45 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 120.54 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 80.83 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.45 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 107.76 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 77.85 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 105.64 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.07 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.82 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP32MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 97.16 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.59 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 110.79 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.84 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.41 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 100.03 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 76.83 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 109.20 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 104.84 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 96.55 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 92.09 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 98.45 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.26 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 111.82 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.88 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 95.88 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 80.83 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 93.46 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.59 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 113.16 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 109.39 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 100.39 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 96.98 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 81.94 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 69.22 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 94.05 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 78.21 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 99.46 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 84.27 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 77.77 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 65.23 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.84 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.89 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.34 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP64MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 103.35 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.28 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 119.01 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.95 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.68 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.23 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 76.97 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 104.11 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.90 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.65 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP64MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 95.62 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.29 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 109.26 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 78.95 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.64 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 98.50 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 75.95 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 107.66 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 103.30 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 95.15 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 90.56 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 96.14 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.09 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 109.51 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.71 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 94.34 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 79.30 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 91.27 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.43 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 111.62 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 107.85 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 98.99 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 95.45 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 81.15 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 68.45 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 92.96 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 77.45 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 97.91 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 82.74 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 77.01 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 64.46 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalPagedMultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 116.36 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.98 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalPagedMultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.44 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalPagedMultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 107.14 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalPagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.25 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 121.84 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.46 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 91.22 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 109.49 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.61 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalPagedMultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 107.62 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalPagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.87 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalPagedMultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.49 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalPagedMultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 99.41 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalPagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.44 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 111.97 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.59 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.31 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 101.69 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 78.59 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 109.09 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalPagedVarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 105.30 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 96.67 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalPagedVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 92.69 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalPagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 98.20 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.49 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalPagedVarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 111.73 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalPagedVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.23 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalPagedVarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 96.58 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalPagedVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 81.43 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalPagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 93.45 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalPagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.71 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 112.92 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 109.80 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 100.52 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 97.70 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 86.27 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 70.65 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 95.91 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.65 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 100.16 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 84.99 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 81.45 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 66.66 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 121.66 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.51 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.45 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.03 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 127.01 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.37 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.39 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.99 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.04 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 117.26 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 106.12 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 93.91 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 100.17 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.12 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 110.98 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.56 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 96.35 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.42 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 110.72 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 98.62 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 118.84 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.17 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.11 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.69 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 124.32 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 109.55 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.18 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.66 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.70 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.57 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 103.25 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 91.09 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 95.64 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.75 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 106.45 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.19 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 91.82 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 80.17 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 107.86 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 95.77 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 117.30 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.00 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.95 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.52 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 122.79 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.02 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.88 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.48 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 91.52 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 113.04 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 101.71 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 89.56 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 93.33 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.58 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 104.13 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.02 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 89.52 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 78.00 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 106.32 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 94.40 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomPagedMultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 120.70 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.11 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomPagedMultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 109.17 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomPagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.62 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 125.66 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomPagedMultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 111.54 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomPagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.12 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomPagedMultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 109.72 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomPagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.63 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomPagedMultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 115.78 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomPagedVarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 103.81 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomPagedVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 91.67 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomPagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 95.51 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.92 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomPagedVarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 106.48 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomPagedVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.39 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomPagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 91.69 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomPagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 80.22 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomPagedVarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 108.53 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvCustomPagedVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 96.51 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 110.02 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 99.52 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 114.02 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 103.39 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 107.06 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 96.74 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 111.05 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 100.73 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 105.49 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 95.22 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 109.48 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 99.20 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvDensePagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 106.93 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvDensePagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 96.63 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 110.92 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 100.62 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 124.54 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.22 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 119.05 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.39 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.00 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 130.72 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.08 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.56 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 116.83 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.96 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 115.85 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.10 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 117.36 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.66 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.13 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 120.84 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.21 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.53 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.98 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.94 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 126.12 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 114.84 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 113.29 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 102.32 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 106.52 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.30 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 120.77 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.30 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 104.20 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 88.84 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 101.39 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.29 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 127.70 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 119.22 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 114.85 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 107.21 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 90.63 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 74.88 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 101.24 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.68 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 107.76 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 92.20 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 85.33 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 70.47 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 121.73 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.88 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.84 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 111.45 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.79 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 127.91 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.66 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.16 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.14 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.55 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 113.03 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.89 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 113.15 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 103.73 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.92 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 118.03 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.80 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.25 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.16 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 80.53 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 123.02 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 111.98 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 110.50 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 99.48 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 102.02 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.05 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 116.24 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.18 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 101.24 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 86.01 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 96.88 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.05 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 124.72 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 116.52 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 112.20 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 104.38 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 89.06 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 73.47 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 99.67 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.27 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 104.92 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 89.48 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 83.87 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 69.06 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 120.20 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.71 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.66 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 110.05 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.62 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 126.38 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.91 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.38 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.61 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 80.80 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 111.50 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.59 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.98 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 102.20 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.75 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 116.49 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.05 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 91.35 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 104.76 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.77 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 121.47 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 110.45 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 109.10 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 98.07 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 99.80 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.88 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 113.93 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.88 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 99.83 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 84.60 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 94.56 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.88 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 123.18 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 114.98 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 110.66 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 102.97 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 87.84 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 72.82 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 98.57 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.50 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 103.38 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 87.95 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 82.65 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 68.29 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 124.17 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.64 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 116.72 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 113.97 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.67 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 129.54 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.20 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.56 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 115.87 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.09 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 115.48 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.77 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.91 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.23 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.68 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 119.79 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.34 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.66 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.14 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.94 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 123.49 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 113.39 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 111.23 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 100.64 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 103.37 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 91.59 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 117.78 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.48 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 102.91 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 90.44 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 97.99 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.45 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 125.07 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 117.80 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 112.80 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 105.79 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 91.32 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 75.41 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 102.05 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.09 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 105.91 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 93.17 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 86.01 KB
fmhaSm100fKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 71.01 KB
fmhaSm100fKernel_QkvE4m3OFp16H64SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:36 111.24 KB
fmhaSm100fKernel_QkvE4m3OFp16H64SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:36 87.59 KB
fmhaSm100fKernel_QkvE4m3OFp16H64SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:36 113.82 KB
fmhaSm100fKernel_QkvE4m3OFp16H64SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:36 93.79 KB
fmhaSm100fKernel_QkvE4m3OFp16H64SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:36 113.24 KB
fmhaSm100fKernel_QkvE4m3OFp16H64SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:36 89.66 KB
fmhaSm100fKernel_QkvE4m3OFp16H64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:36 111.78 KB
fmhaSm100fKernel_QkvE4m3OFp16H64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:36 91.42 KB
fmhaSm100fKernel_QkvE4m3OFp16H64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:36 117.65 KB
fmhaSm100fKernel_QkvE4m3OFp16H64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:36 97.57 KB
fmhaSm100fKernel_QkvE4m3OFp16H64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:36 114.02 KB
fmhaSm100fKernel_QkvE4m3OFp16H64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:36 93.48 KB
fmhaSm100fKernel_QkvFp16OFp16H128PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 106.30 KB
fmhaSm100fKernel_QkvFp16OFp16H128PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 94.67 KB
fmhaSm100fKernel_QkvFp16OFp16H128PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 111.46 KB
fmhaSm100fKernel_QkvFp16OFp16H128PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 100.09 KB
fmhaSm100fKernel_QkvFp16OFp16H128PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 105.82 KB
fmhaSm100fKernel_QkvFp16OFp16H128PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 94.62 KB
fmhaSm100fKernel_QkvFp16OFp16H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 110.88 KB
fmhaSm100fKernel_QkvFp16OFp16H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 100.10 KB
fmhaSm100fKernel_QkvFp16OFp16H128PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 116.84 KB
fmhaSm100fKernel_QkvFp16OFp16H128PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 105.65 KB
fmhaSm100fKernel_QkvFp16OFp16H128PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 122.03 KB
fmhaSm100fKernel_QkvFp16OFp16H128PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 111.15 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 139.20 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 122.39 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 133.04 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP16MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 125.80 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 115.28 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 149.18 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.27 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.32 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 128.93 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.18 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 131.08 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 121.20 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 129.63 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP16MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 115.68 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.66 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 136.52 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.36 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.94 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 118.93 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.51 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 128.28 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 126.93 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 116.04 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 113.49 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 123.87 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.70 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 136.41 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 120.86 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 115.18 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 101.16 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 114.93 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.31 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 133.44 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 135.73 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 121.39 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 122.95 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 94.80 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.29 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 105.69 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.19 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 118.16 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 104.27 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 85.61 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 73.98 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 134.06 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.83 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 125.22 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP32MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 120.67 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.72 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 144.55 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.70 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.76 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 123.89 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.62 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 125.81 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 113.45 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 121.57 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP32MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 110.55 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.91 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 131.51 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.83 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.41 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.02 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.98 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 122.33 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 121.03 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 110.77 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 107.73 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 116.07 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.25 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 127.34 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 113.03 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 109.89 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 95.91 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 105.74 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.02 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 127.51 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 129.88 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 116.12 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 117.31 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 91.84 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 75.41 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 102.23 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.27 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 113.51 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 99.40 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 82.30 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 70.12 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 130.54 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 109.66 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 120.17 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP64MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 117.15 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.55 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 141.02 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.05 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.23 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 120.26 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.09 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 122.72 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.11 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 116.95 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP64MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 107.20 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.55 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 128.28 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.18 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.88 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 110.42 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 80.32 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 119.09 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 117.55 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 107.13 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 104.79 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 110.76 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.66 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 121.77 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.59 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 105.98 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 92.34 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 100.40 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.43 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 124.25 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 126.61 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 112.65 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.24 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 89.29 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 73.87 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 99.99 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.73 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 109.32 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 95.56 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 80.75 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 68.59 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalPagedMultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 132.47 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 111.68 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalPagedMultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 122.06 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalPagedMultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 119.28 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalPagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.32 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 142.95 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.37 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.42 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 122.51 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.28 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalPagedMultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 124.52 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalPagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.00 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalPagedMultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 118.70 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalPagedMultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 109.48 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalPagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.75 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 130.09 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.49 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.07 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.59 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.51 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 119.76 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalPagedVarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 118.91 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 107.91 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalPagedVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.31 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalPagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 111.32 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.98 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalPagedVarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 122.88 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalPagedVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 109.02 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalPagedVarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 107.45 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalPagedVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 94.09 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalPagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 101.48 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalPagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.62 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 124.65 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 127.94 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 113.30 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 115.90 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 91.35 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 75.74 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 102.05 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.52 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 110.66 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 97.45 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 84.87 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 70.48 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 142.48 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 121.77 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 130.73 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.53 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 150.88 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 134.13 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 120.73 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 128.62 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.02 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 137.84 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 125.97 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 113.23 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 120.54 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.52 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 130.77 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.83 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 113.41 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.17 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 132.83 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 121.41 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 137.21 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.21 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 122.92 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.98 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 145.99 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 128.87 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.86 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 120.69 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.27 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 132.94 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 119.77 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 107.96 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 113.04 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.54 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 121.38 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.98 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 104.11 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.20 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 127.49 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 116.63 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 133.67 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 109.02 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 117.86 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.78 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 142.58 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 125.65 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.64 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 116.06 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.92 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 129.71 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 116.27 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 104.43 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 107.72 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.46 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 116.15 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.76 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 98.88 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.13 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 123.67 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.81 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomPagedMultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 135.60 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 111.17 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomPagedMultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 120.01 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomPagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.94 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 144.38 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomPagedMultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 127.57 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomPagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 109.67 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomPagedMultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 117.84 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomPagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.95 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomPagedMultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 131.64 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomPagedVarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 117.60 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomPagedVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 105.80 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomPagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 108.43 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.72 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomPagedVarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 117.00 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomPagedVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.02 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomPagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 99.84 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomPagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.27 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomPagedVarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 125.01 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvCustomPagedVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.34 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 127.88 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 116.16 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 132.92 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 121.55 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 121.93 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 110.73 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 126.49 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 115.73 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 118.34 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 107.27 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 123.39 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 112.81 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvDensePagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 118.98 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvDensePagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 107.94 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 124.05 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 113.48 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 146.59 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 127.09 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 137.82 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 131.88 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 119.62 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 154.94 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.34 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.02 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 135.18 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.02 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 138.59 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 125.90 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 134.41 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 121.96 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 119.11 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 142.16 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.44 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.64 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 124.73 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.09 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 139.82 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 134.19 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 127.55 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 120.73 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 127.70 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.47 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 140.51 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 125.42 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 120.59 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 107.05 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 118.01 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.69 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 144.75 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 141.34 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 132.62 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 128.84 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 99.08 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.49 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 111.09 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.16 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 124.00 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 109.86 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 92.20 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 77.68 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 141.45 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 119.53 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 130.00 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 127.24 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.05 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 150.21 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.78 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.47 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 130.24 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.34 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 133.34 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 118.15 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 126.48 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 116.95 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 111.23 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 137.17 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.78 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.11 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 120.09 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.56 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 133.88 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 128.30 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 122.28 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.97 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 119.77 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.20 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 131.59 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 117.82 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 115.33 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 101.80 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 109.16 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.58 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 138.92 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 135.63 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 127.23 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 123.08 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 96.24 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.61 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 107.52 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.25 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 119.11 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 105.12 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 89.12 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 73.95 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 137.94 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.36 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 124.95 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 123.59 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.88 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 146.56 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 91.13 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.94 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 126.48 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.81 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 130.11 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.80 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 121.73 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 113.59 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.89 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 133.95 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.26 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.59 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 116.48 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.91 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 130.63 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 124.82 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 118.66 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.02 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 114.33 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.61 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 126.14 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.51 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 111.41 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 98.23 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 103.92 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.98 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 135.54 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 132.37 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 123.75 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 120.13 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 93.68 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 78.06 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 105.02 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.71 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 114.95 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 101.17 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 86.72 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 72.41 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 140.57 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 117.59 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 128.31 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 126.09 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 109.86 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 149.17 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.84 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.77 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 128.86 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.65 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 132.62 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 115.78 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 124.95 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 115.97 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 109.12 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 136.55 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.09 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.42 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 118.73 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.75 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 131.88 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 126.92 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 120.23 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.27 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 116.62 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.23 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 128.46 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 115.02 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 113.85 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 100.17 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 106.28 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.48 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 136.80 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 134.33 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 125.34 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 122.25 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 96.38 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 80.45 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 107.84 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 91.10 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 116.73 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 103.36 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 89.30 KB
fmhaSm100fKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 74.67 KB
fmhaSm100fKernel_QkvFp16OFp16H128SeparateQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 105.95 KB
fmhaSm100fKernel_QkvFp16OFp16H128SeparateQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 94.74 KB
fmhaSm100fKernel_QkvFp16OFp16H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 111.01 KB
fmhaSm100fKernel_QkvFp16OFp16H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 100.23 KB
fmhaSm100fKernel_QkvFp16OFp16H256PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 102.35 KB
fmhaSm100fKernel_QkvFp16OFp16H256PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 87.88 KB
fmhaSm100fKernel_QkvFp16OFp16H256PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 105.07 KB
fmhaSm100fKernel_QkvFp16OFp16H256PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 91.05 KB
fmhaSm100fKernel_QkvFp16OFp16H256PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 102.02 KB
fmhaSm100fKernel_QkvFp16OFp16H256PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 90.91 KB
fmhaSm100fKernel_QkvFp16OFp16H256PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 105.12 KB
fmhaSm100fKernel_QkvFp16OFp16H256PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 93.87 KB
fmhaSm100fKernel_QkvFp16OFp16H256PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 112.66 KB
fmhaSm100fKernel_QkvFp16OFp16H256PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 98.48 KB
fmhaSm100fKernel_QkvFp16OFp16H256PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 115.70 KB
fmhaSm100fKernel_QkvFp16OFp16H256PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 101.47 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 147.48 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 150.62 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 163.59 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP16MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 138.13 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 143.80 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 152.64 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.27 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 118.78 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 139.98 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.09 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 134.20 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 147.84 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 158.51 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP16MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 127.37 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 142.01 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 139.49 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.06 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.68 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 128.95 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.37 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 142.86 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 133.70 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 123.10 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 116.68 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 145.45 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 129.64 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 161.89 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 145.16 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 127.90 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 109.84 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 134.89 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 121.51 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 145.70 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 136.51 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 126.03 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 123.62 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 109.74 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.95 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 124.54 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.55 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 125.90 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 113.20 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 100.54 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.83 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 132.27 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 135.69 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 147.57 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP32MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 123.06 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 128.87 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 137.88 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.31 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.82 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 125.19 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.12 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 119.12 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 132.88 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 142.32 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP32MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.29 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 127.05 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 124.85 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.10 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.72 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.27 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.41 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 123.66 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 117.53 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 107.48 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 101.58 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 133.07 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 116.79 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 147.06 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 132.31 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 110.86 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 94.58 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 120.60 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.98 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 126.42 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 123.55 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 110.29 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.09 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 103.76 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.84 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 118.27 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.45 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 112.66 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 97.67 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 93.12 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 80.73 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 125.32 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 128.12 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 140.61 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP64MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 115.98 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 121.30 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 131.84 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.57 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.96 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 119.15 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.27 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.05 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 125.54 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 135.54 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP64MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 105.22 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 119.71 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 118.69 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.36 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.98 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.12 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.67 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 115.73 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 110.23 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 101.54 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 95.64 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 125.62 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 109.45 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 139.65 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 124.86 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 103.25 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 88.63 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 113.70 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.11 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 118.62 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 116.90 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 104.34 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 102.61 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 99.59 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.07 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 113.88 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.68 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 106.27 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 92.20 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 89.75 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 77.96 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalPagedMultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 125.67 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 128.61 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalPagedMultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 140.92 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalPagedMultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 116.58 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalPagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 121.78 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 132.43 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.89 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.16 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 119.74 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.46 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalPagedMultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.52 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalPagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 126.15 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalPagedMultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 135.98 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalPagedMultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 105.69 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalPagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 120.20 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 119.29 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.55 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.30 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.84 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.99 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 116.16 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalPagedVarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 109.85 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 102.16 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalPagedVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 96.21 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalPagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 125.55 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.09 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalPagedVarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 139.20 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalPagedVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 123.84 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalPagedVarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 103.34 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalPagedVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 89.33 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalPagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 113.30 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalPagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.50 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 119.11 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 116.73 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 105.21 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 103.17 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 101.27 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.37 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 115.42 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.96 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 106.56 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 92.62 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 93.76 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.38 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 143.16 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 128.01 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 146.16 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 131.09 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 124.84 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 112.19 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 127.82 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 115.15 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 116.58 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 105.03 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 119.57 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 108.12 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvDensePagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 116.29 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvDensePagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 105.01 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 119.41 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 108.09 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 155.00 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 154.41 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 168.21 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 143.51 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 147.20 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 157.25 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.34 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 123.37 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 145.09 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.80 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 141.76 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 151.87 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 163.10 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 132.77 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 145.78 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 144.01 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.14 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 119.39 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 134.37 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.09 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 154.24 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 141.55 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 134.53 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 124.04 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 149.48 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 134.01 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 166.62 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 149.86 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 132.95 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 116.61 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 140.22 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 124.94 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 156.97 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 140.91 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 137.34 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 128.10 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 114.48 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.18 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 130.32 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 111.64 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 130.91 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 118.90 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 106.62 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.65 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 139.80 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 140.47 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 152.20 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 128.30 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 133.26 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 142.48 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.41 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 115.44 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 130.29 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.75 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 126.55 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 137.36 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 146.91 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 117.57 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 131.28 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 129.23 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.20 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 111.46 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 119.57 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.16 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 135.60 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 125.36 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 118.77 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.93 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 137.09 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 121.28 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 151.55 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 136.48 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 115.76 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 99.90 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 127.89 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.41 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 138.45 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 127.82 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 121.58 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.59 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 108.33 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 91.20 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 123.75 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.53 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 117.75 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 102.87 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 100.29 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.55 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 132.85 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 132.83 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 145.23 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 121.36 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 125.62 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 136.45 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.65 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.54 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 124.25 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.98 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 119.61 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 130.37 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 140.13 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 110.62 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 124.16 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 123.08 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.44 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.69 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 113.53 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.38 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 127.55 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 117.41 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 112.83 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 102.98 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 129.56 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 113.94 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 143.89 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 128.91 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 108.13 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 93.83 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 120.30 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.66 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 130.38 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 121.05 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 115.76 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.99 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 104.12 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.30 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 119.32 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.77 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 111.27 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 97.39 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 96.13 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.78 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 133.91 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 133.42 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 146.01 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 122.41 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 126.34 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 137.62 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.62 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 115.64 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 125.44 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.95 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 120.66 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 131.09 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 141.72 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 111.68 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 124.76 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 124.38 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.29 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 111.54 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.84 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.36 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 128.09 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 117.91 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 114.15 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 104.29 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 129.74 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.40 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 144.60 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 129.20 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 109.20 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 95.12 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 118.01 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.12 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 130.93 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 121.66 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 117.09 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.12 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 106.07 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.10 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 121.02 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.44 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 112.30 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 98.42 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 96.11 KB
fmhaSm100fKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.70 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 110.20 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 98.59 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 112.92 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 101.74 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 109.89 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 101.75 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 112.88 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 104.59 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP16MultiCtasKvGmemSepVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 154.71 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP16MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 155.74 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP16MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 155.70 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP16MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 153.85 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 160.52 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 151.64 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 151.27 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 145.89 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 154.72 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 147.05 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 163.52 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 154.46 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 154.07 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 145.25 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 152.33 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 141.34 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP32MultiCtasKvGmemSepVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 129.91 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP32MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 126.84 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP32MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 134.80 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP32MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 129.20 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 137.91 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 132.36 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 124.80 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 119.19 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 131.43 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 118.42 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 140.79 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 136.35 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 127.61 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 127.67 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 132.71 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 119.70 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP64MultiCtasKvGmemSepVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 117.04 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP64MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.02 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP64MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 123.80 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP64MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 116.54 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 124.27 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 118.88 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 113.54 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 107.53 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 118.03 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 104.80 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 127.14 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 125.53 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 116.47 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.38 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 120.92 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 107.77 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 159.55 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 150.35 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 162.41 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 153.31 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 139.45 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 129.79 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 142.30 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 132.88 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 125.27 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 116.88 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 128.12 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 119.84 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP16MultiCtasKvGmemSepVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 162.14 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP16MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 161.03 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP16MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 160.34 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP16MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 159.24 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 160.43 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 153.26 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 159.84 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 151.94 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 157.27 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 149.73 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 157.48 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 145.61 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP32MultiCtasKvGmemSepVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 137.34 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP32MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 132.14 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP32MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 139.30 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP32MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 134.58 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 140.54 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 126.55 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 136.57 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 123.09 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 140.96 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 132.31 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 137.85 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 124.40 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP64MultiCtasKvGmemSepVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 124.34 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP64MultiCtasKvGmemSepVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 119.31 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP64MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 128.31 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP64MultiCtasKvGmemSepVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 121.92 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 126.38 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 115.00 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 123.03 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 110.00 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 130.15 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 119.01 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 126.05 KB
fmhaSm100fKernel_QkvFp16OFp16H512HVPerCta256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.97 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 231.46 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 222.33 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 141.95 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 134.99 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 226.73 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 241.30 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 219.57 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 139.53 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 150.30 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 133.95 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 202.04 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 192.23 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 224.45 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 216.98 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 190.55 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 184.47 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 140.94 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 125.62 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 159.98 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 140.65 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 129.61 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 117.73 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 192.86 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 183.61 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 128.88 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 121.68 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 188.41 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 204.53 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 180.99 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 125.13 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 135.91 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 119.45 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 176.65 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 167.05 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 201.89 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 188.76 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 163.98 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 155.32 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 127.41 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.62 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 148.87 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 127.56 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 116.17 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.72 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 177.77 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 168.52 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 122.11 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 115.06 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 173.19 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 188.89 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 165.91 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 118.57 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 129.35 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.90 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 162.83 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 150.33 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 185.70 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 171.71 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 149.19 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 138.55 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 120.80 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.98 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 141.49 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 120.93 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 109.54 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvCausalP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.08 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 235.86 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 226.48 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 146.16 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 138.84 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 231.25 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 245.80 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 223.71 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 143.74 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 155.14 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 137.67 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 206.40 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 196.48 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 229.96 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 220.68 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 195.05 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 188.02 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 143.43 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 129.95 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 165.38 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 145.75 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 133.88 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 121.66 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 197.26 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 187.76 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 133.06 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 125.51 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 192.80 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 208.79 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 185.15 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 129.34 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 140.74 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 123.30 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 181.71 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 171.32 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 206.15 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 192.43 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 170.34 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 158.75 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 132.35 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 116.82 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 154.30 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 132.60 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 122.80 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.54 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 182.16 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 172.67 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 126.32 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 118.91 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 177.71 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 193.52 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 170.06 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 122.66 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 134.19 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 116.74 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 166.79 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 154.47 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 190.08 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 175.54 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 155.42 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 141.99 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 125.23 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.19 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 146.94 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 126.09 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 115.55 KB
fmhaSm100fKernel_QkvFp16OFp16H512PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.91 KB
fmhaSm100fKernel_QkvFp16OFp16H64PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 102.44 KB
fmhaSm100fKernel_QkvFp16OFp16H64PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 90.35 KB
fmhaSm100fKernel_QkvFp16OFp16H64PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 109.85 KB
fmhaSm100fKernel_QkvFp16OFp16H64PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 97.94 KB
fmhaSm100fKernel_QkvFp16OFp16H64PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 101.59 KB
fmhaSm100fKernel_QkvFp16OFp16H64PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 90.09 KB
fmhaSm100fKernel_QkvFp16OFp16H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 109.32 KB
fmhaSm100fKernel_QkvFp16OFp16H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 97.63 KB
fmhaSm100fKernel_QkvFp16OFp16H64PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 112.98 KB
fmhaSm100fKernel_QkvFp16OFp16H64PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 101.16 KB
fmhaSm100fKernel_QkvFp16OFp16H64PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 120.30 KB
fmhaSm100fKernel_QkvFp16OFp16H64PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 108.93 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 122.77 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.37 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 117.38 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP16MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 111.33 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.36 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 133.02 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.95 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.70 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 114.20 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 80.84 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 113.77 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.25 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 115.44 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP16MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 103.48 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.37 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 123.28 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.05 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.78 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.22 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.81 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 118.21 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 112.99 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 106.17 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 100.88 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 106.70 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.41 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 119.21 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.02 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 101.83 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 87.03 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 101.57 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.34 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 126.25 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 121.77 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 114.27 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 109.99 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 84.84 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 72.58 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 97.27 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.48 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 105.66 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 90.99 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 80.47 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 68.23 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 119.96 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.02 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 113.03 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP32MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.52 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.02 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 130.33 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.53 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 91.28 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 111.76 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.42 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 110.96 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.03 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 111.22 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP32MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 100.66 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.03 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 120.47 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.63 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.38 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 103.91 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 78.53 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 115.86 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 110.12 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 103.89 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 98.03 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 102.16 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.10 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 115.48 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.91 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 99.50 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 84.32 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 97.04 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.04 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 123.14 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 119.15 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 111.49 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 107.16 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 83.54 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 71.17 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 95.58 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 80.19 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 103.21 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 88.15 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 79.17 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 66.95 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 118.43 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.85 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.87 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP64MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.98 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.85 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 128.79 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.78 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.53 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 110.23 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 78.67 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 109.43 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.73 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.93 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP64MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 99.13 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.86 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 119.06 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 80.88 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.62 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 102.25 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 77.65 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 114.31 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 108.59 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 102.37 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 96.50 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 99.85 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.93 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 113.17 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.62 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 97.96 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 82.79 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 94.73 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.87 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 121.60 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 117.62 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 109.97 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 105.62 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 82.79 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 70.40 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 94.48 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.41 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 101.66 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 86.74 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 78.40 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 66.17 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalPagedMultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 121.95 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.82 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalPagedMultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 113.84 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalPagedMultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 110.77 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalPagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.82 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 131.66 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.41 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.04 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 113.49 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.31 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalPagedMultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.70 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalPagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.83 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalPagedMultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.02 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalPagedMultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 102.92 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalPagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.88 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 121.92 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.52 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 91.26 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 105.81 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 80.29 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 115.74 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalPagedVarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 110.71 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 103.64 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalPagedVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 98.41 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalPagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 102.16 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.34 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalPagedVarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 115.52 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalPagedVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.26 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalPagedVarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 100.20 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalPagedVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 85.05 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalPagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 96.91 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalPagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.28 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 123.03 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 119.61 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 111.24 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 107.65 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 87.79 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 72.48 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 97.45 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.49 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 103.91 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 88.88 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 82.82 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 68.38 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 125.78 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.72 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 115.74 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.98 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 134.45 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 116.78 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.60 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.20 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.12 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 124.84 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 111.69 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 99.97 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 103.99 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.12 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 114.44 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.43 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 99.94 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.99 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 118.95 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 107.37 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 122.97 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.26 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.16 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.52 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 131.76 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 113.97 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.14 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 110.62 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.53 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 122.02 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 108.70 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 97.16 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 99.46 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.88 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 110.87 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.18 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 95.41 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.74 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 116.09 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 104.64 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 121.44 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.96 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 109.98 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.23 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 130.23 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.44 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.97 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.45 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.36 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 120.49 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 107.16 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 95.75 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 97.27 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.58 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 108.55 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.89 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 93.22 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.45 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 114.55 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 103.14 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomPagedMultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 124.95 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.07 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomPagedMultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 113.09 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomPagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.46 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 133.09 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomPagedMultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 115.96 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomPagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.08 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomPagedMultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 111.55 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomPagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.34 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomPagedMultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 123.36 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomPagedVarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 109.52 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomPagedVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 97.48 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomPagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 99.45 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.80 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomPagedVarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 110.91 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomPagedVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.23 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomPagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 95.27 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomPagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.79 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomPagedVarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 116.63 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvCustomPagedVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 105.00 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 116.65 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 106.05 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 125.54 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 114.15 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 114.32 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 103.77 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 122.55 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 111.37 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 112.75 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 102.24 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 121.02 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 109.97 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvDensePagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 114.06 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvDensePagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 103.30 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 122.45 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 111.03 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 130.16 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 112.06 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 122.73 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 117.90 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.70 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 138.77 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.02 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.27 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 120.30 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.55 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 121.17 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 111.07 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 120.64 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 110.17 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.70 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 129.16 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.12 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.49 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.58 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.53 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 130.01 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 120.27 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 117.77 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 107.79 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 110.37 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.27 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 123.98 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 111.95 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 107.31 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 92.34 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 104.96 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.86 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 137.09 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 127.29 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 124.74 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 115.42 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 92.14 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 76.83 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 102.56 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.38 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 111.03 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 96.17 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 86.70 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 72.15 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 127.35 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.73 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 118.40 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 115.09 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 103.35 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 135.99 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.61 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.87 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 117.87 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.14 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 118.35 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 106.73 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 116.42 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 107.23 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 102.48 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 126.38 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.71 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.08 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 110.14 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.24 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 127.52 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 117.40 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 115.48 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 104.95 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 105.87 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.02 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 120.24 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 107.70 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 104.86 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 89.50 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 100.44 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.62 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 134.10 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 124.66 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 122.09 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.59 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 90.59 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 75.42 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 100.87 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.10 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 108.58 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 93.45 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 85.12 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 70.74 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 125.82 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.55 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 116.23 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 113.55 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.18 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 134.58 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.85 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.11 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 116.46 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.38 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 116.82 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.56 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 114.13 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 105.83 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 100.31 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 124.85 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.95 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.32 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 108.61 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.36 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 125.98 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 115.86 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 114.09 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 103.54 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 103.65 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 91.85 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 117.93 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.53 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 103.45 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 88.09 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 98.13 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.45 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 132.56 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 123.14 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 120.56 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 111.18 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 89.36 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 74.65 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 99.77 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.33 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 107.03 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 91.92 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 84.02 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 69.97 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 129.67 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 109.35 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 120.38 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 117.48 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 105.36 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 137.73 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.14 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.40 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 119.84 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.55 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 120.67 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.48 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 118.69 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 109.62 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 104.24 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 128.12 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.12 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.48 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 112.12 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.65 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 128.16 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 118.81 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 116.22 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 106.23 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 107.22 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 95.41 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 121.78 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 108.98 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 106.53 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 94.07 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 101.56 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.15 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 134.73 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 125.99 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 122.70 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 113.88 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 92.83 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 77.24 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 103.38 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.92 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ64Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 109.66 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ64Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 97.73 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 87.38 KB
fmhaSm100fKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 72.69 KB
fmhaSm100fKernel_QkvFp16OFp16H80PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 106.00 KB
fmhaSm100fKernel_QkvFp16OFp16H80PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 89.27 KB
fmhaSm100fKernel_QkvFp16OFp16H80PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 112.42 KB
fmhaSm100fKernel_QkvFp16OFp16H80PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 97.13 KB
fmhaSm100fKernel_QkvFp16OFp16H80PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 106.28 KB
fmhaSm100fKernel_QkvFp16OFp16H80PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 89.02 KB
fmhaSm100fKernel_QkvFp16OFp16H80PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 113.62 KB
fmhaSm100fKernel_QkvFp16OFp16H80PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 96.83 KB
fmhaSm100fKernel_QkvFp16OFp16H80PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 116.66 KB
fmhaSm100fKernel_QkvFp16OFp16H80PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 100.20 KB
fmhaSm100fKernel_QkvFp16OFp16H80PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 123.12 KB
fmhaSm100fKernel_QkvFp16OFp16H80PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 108.03 KB
fmhaSm100fKernel_QkvFp16OFp16H80PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 127.68 KB
fmhaSm100fKernel_QkvFp16OFp16H80PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 110.94 KB
fmhaSm100fKernel_QkvFp16OFp16H80PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 134.23 KB
fmhaSm100fKernel_QkvFp16OFp16H80PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 118.86 KB
fmhaSm100fKernel_QkvFp16OFp16H80PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 121.89 KB
fmhaSm100fKernel_QkvFp16OFp16H80PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 105.55 KB
fmhaSm100fKernel_QkvFp16OFp16H80PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 128.43 KB
fmhaSm100fKernel_QkvFp16OFp16H80PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 113.60 KB
fmhaSm100fKernel_QkvFp16OFp16H80PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 118.65 KB
fmhaSm100fKernel_QkvFp16OFp16H80PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 102.05 KB
fmhaSm100fKernel_QkvFp16OFp16H80PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 125.06 KB
fmhaSm100fKernel_QkvFp16OFp16H80PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 109.98 KB
fmhaSm100fKernel_QkvFp16OFp16H80PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 128.20 KB
fmhaSm100fKernel_QkvFp16OFp16H80PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 110.66 KB
fmhaSm100fKernel_QkvFp16OFp16H80PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 135.63 KB
fmhaSm100fKernel_QkvFp16OFp16H80PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 118.59 KB
fmhaSm100fKernel_QkvFp16OFp16H80PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 121.69 KB
fmhaSm100fKernel_QkvFp16OFp16H80PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 104.88 KB
fmhaSm100fKernel_QkvFp16OFp16H80PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 129.00 KB
fmhaSm100fKernel_QkvFp16OFp16H80PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 112.82 KB
fmhaSm100fKernel_QkvFp16OFp16H80PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 118.58 KB
fmhaSm100fKernel_QkvFp16OFp16H80PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 101.92 KB
fmhaSm100fKernel_QkvFp16OFp16H80PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 125.88 KB
fmhaSm100fKernel_QkvFp16OFp16H80PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 109.86 KB
fmhaSm100fKernel_QkvFp16OFp16H80PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 139.23 KB
fmhaSm100fKernel_QkvFp16OFp16H80PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 122.67 KB
fmhaSm100fKernel_QkvFp16OFp16H80PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 145.67 KB
fmhaSm100fKernel_QkvFp16OFp16H80PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 129.28 KB
fmhaSm100fKernel_QkvFp16OFp16H80PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 133.43 KB
fmhaSm100fKernel_QkvFp16OFp16H80PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 117.29 KB
fmhaSm100fKernel_QkvFp16OFp16H80PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 139.87 KB
fmhaSm100fKernel_QkvFp16OFp16H80PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 123.91 KB
fmhaSm100fKernel_QkvFp16OFp16H80PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 130.20 KB
fmhaSm100fKernel_QkvFp16OFp16H80PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 113.66 KB
fmhaSm100fKernel_QkvFp16OFp16H80PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 136.62 KB
fmhaSm100fKernel_QkvFp16OFp16H80PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 120.41 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H128PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 93.45 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H128PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 86.01 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H128PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 91.54 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H128PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 83.76 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H128PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 89.29 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H128PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 81.64 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H128PagedKvCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 90.41 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H128PagedKvCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 82.46 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H128PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 93.79 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H128PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 87.00 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H128PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 91.79 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H128PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 84.35 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H128PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 89.29 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H128PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 82.22 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H128PagedKvDensePagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 90.41 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H128PagedKvDensePagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 82.93 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H128PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 100.41 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H128PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 92.93 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H128PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 98.54 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H128PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 90.56 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H128PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 96.16 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H128PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 88.68 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 98.16 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 90.43 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H256PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 105.33 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H256PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 92.06 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H256PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 97.84 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H256PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 84.93 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H256PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 93.84 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H256PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 80.68 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H256PagedKvCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 94.95 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H256PagedKvCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 81.80 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H256PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 105.59 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H256PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 99.82 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H256PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 98.09 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H256PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 92.70 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H256PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 94.22 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H256PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 88.20 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H256PagedKvDensePagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 95.32 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H256PagedKvDensePagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 89.32 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 111.97 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 98.57 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 104.72 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 91.70 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 100.60 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 87.32 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 102.95 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 89.07 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H64PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 88.10 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H64PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 82.79 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H64PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 84.72 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H64PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 76.55 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H64PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 82.35 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H64PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 74.43 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H64PagedKvCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 83.47 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H64PagedKvCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 75.43 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H64PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 87.85 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H64PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 83.24 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H64PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 84.72 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H64PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 77.31 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H64PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 82.22 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H64PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 75.06 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H64PagedKvDensePagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 83.60 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H64PagedKvDensePagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 76.30 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 95.09 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 87.95 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 91.72 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 83.69 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 89.22 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 81.69 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 91.34 KB
fmhaSm103aKernel_QE4m3KvE2m1OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 83.57 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 96.41 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 88.71 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 94.41 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 86.46 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 92.16 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 84.34 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 93.29 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 85.16 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 95.23 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 74.02 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 80.91 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 69.11 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 97.27 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 87.80 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 72.45 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 78.21 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 68.54 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 89.84 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 77.40 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 69.43 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 66.99 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 58.95 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 74.54 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 65.84 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 64.71 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 56.91 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 79.69 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 71.72 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 93.10 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 71.89 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 78.91 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 66.99 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 95.27 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 85.68 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 70.32 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 76.09 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 66.41 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 87.72 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 74.60 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 67.30 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 64.74 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 57.07 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 72.51 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 63.71 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 62.46 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 55.04 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 76.64 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 69.59 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 91.48 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 70.27 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 77.29 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 65.36 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 93.52 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 84.05 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 68.70 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 74.46 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 64.79 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 86.09 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 71.98 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 65.68 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 62.49 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 55.32 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 69.63 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 62.09 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 60.09 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 53.29 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 73.89 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 67.72 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomPagedMultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 93.35 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 72.39 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomPagedMultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.41 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomPagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 67.61 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 95.52 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomPagedMultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 86.05 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomPagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 70.82 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomPagedMultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 76.59 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomPagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 67.04 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomPagedMultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 87.97 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomPagedVarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 73.60 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomPagedVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 67.18 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomPagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 63.99 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 56.95 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomPagedVarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 71.13 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomPagedVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 63.71 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomPagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 61.71 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomPagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 54.91 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomPagedVarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 75.77 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvCustomPagedVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 69.34 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvDenseP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 73.39 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvDenseP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 68.74 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvDenseP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 71.95 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvDenseP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 68.16 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 96.66 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 89.72 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvDenseP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 65.49 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvDenseP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 58.70 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvDenseP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 63.21 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvDenseP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 56.54 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvDenseP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 71.27 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvDenseP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 66.61 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvDenseP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 69.82 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvDenseP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 66.04 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 94.66 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 86.85 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvDenseP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 63.49 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvDenseP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 56.82 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvDenseP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 61.34 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvDenseP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 54.66 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvDenseP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 69.64 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvDenseP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 65.11 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvDenseP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 68.20 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvDenseP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 64.54 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 92.16 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 84.72 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvDenseP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 61.37 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvDenseP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 55.07 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvDenseP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 59.09 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvDenseP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 52.91 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvDensePagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 71.89 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvDensePagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 67.36 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvDensePagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 70.32 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvDensePagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 66.66 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvDensePagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 93.29 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvDensePagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 85.55 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvDensePagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 63.37 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvDensePagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 56.70 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvDensePagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 61.21 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvDensePagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 54.66 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 76.27 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 71.61 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 74.82 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 70.91 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 103.54 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 95.56 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 71.10 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 60.43 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 67.82 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 58.27 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 74.27 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 69.49 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 72.70 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 68.79 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 101.54 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 93.06 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 68.60 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 58.43 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 65.44 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 56.14 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 72.64 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 67.86 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 71.07 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 67.29 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 99.16 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 91.31 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 66.10 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 56.43 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 62.94 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 54.14 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 75.27 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 70.61 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 73.70 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 69.91 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 101.16 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 93.05 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 66.60 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 58.80 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 64.19 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 56.52 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 108.20 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 94.43 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 100.72 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 87.30 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 96.72 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 82.93 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 97.95 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 84.43 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvDenseP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 98.27 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvDenseP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.61 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvDenseP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.45 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvDenseP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 93.41 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 108.47 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 102.20 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvDenseP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 90.37 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvDenseP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.20 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvDenseP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 87.84 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvDenseP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 79.66 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvDenseP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 91.52 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvDenseP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.74 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvDenseP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.57 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvDenseP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.66 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 100.97 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 95.20 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvDenseP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 81.62 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvDenseP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 75.70 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvDenseP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 79.21 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvDenseP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 73.16 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvDenseP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.89 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvDenseP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.11 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvDenseP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.07 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvDenseP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 82.04 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 97.09 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 90.70 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvDenseP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 77.62 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvDenseP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 71.32 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvDenseP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 75.09 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvDenseP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 68.79 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvDensePagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.02 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvDensePagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 85.24 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvDensePagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.95 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvDensePagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.91 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvDensePagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 98.45 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvDensePagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 91.70 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvDensePagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 79.24 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvDensePagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 72.32 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvDensePagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 76.71 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvDensePagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 69.91 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 101.52 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 97.61 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 99.45 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 96.41 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 115.10 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 100.95 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 94.10 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 83.80 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 90.82 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 81.39 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 94.64 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.74 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.57 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 89.54 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 107.72 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 94.07 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 87.47 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 76.30 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 84.19 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 73.89 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.02 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 86.11 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.07 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 84.91 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 103.60 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 89.82 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 83.47 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 71.93 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 80.19 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 69.52 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 92.64 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 88.74 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 90.70 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 87.41 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 105.95 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 91.45 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 82.35 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 73.55 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 79.82 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 71.14 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 87.72 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 82.37 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 84.35 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 76.43 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 81.97 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 74.30 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 83.10 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 75.43 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP16MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 90.49 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 72.02 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP16MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 78.66 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 70.11 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP16MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 92.53 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP16MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 84.95 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 71.32 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP16MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 77.34 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 69.41 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP16MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 86.98 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP16VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 74.79 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP16VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 67.45 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 66.12 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 59.95 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP16VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 71.01 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP16VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 64.59 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 63.96 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 57.79 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP16VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 76.96 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP16VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 69.36 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP32MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 85.99 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 67.52 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP32MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 74.41 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 65.49 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP32MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 88.03 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP32MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 80.45 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 66.82 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP32MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 72.96 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 64.91 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP32MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 82.61 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP32VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 70.74 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP32VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 62.82 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 62.62 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 55.70 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP32VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 67.51 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP32VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 60.34 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 60.46 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 53.54 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP32VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 72.66 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP32VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 65.11 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP64MultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 84.37 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 65.89 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP64MultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 72.79 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 63.99 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP64MultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 86.28 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP64MultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 78.95 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 65.20 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP64MultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 71.34 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 63.29 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP64MultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 80.73 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP64VarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 68.12 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP64VarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 61.32 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 60.24 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 53.95 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP64VarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 65.13 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP64VarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 58.71 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 58.09 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 51.79 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP64VarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 70.03 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomP64VarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 63.23 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomPagedMultiCtasKvCgaVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 86.37 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 68.39 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomPagedMultiCtasKvCgaVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 75.04 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomPagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 66.36 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomPagedMultiCtasKvCgaVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 88.41 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomPagedMultiCtasKvVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 80.95 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomPagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 67.57 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomPagedMultiCtasKvVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 73.46 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomPagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 65.66 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomPagedMultiCtasKvVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 82.98 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomPagedVarSeqQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 69.74 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomPagedVarSeqQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 62.95 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomPagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 62.12 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 55.57 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomPagedVarSeqQ32Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 67.01 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomPagedVarSeqQ32Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 60.34 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomPagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 59.96 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomPagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 53.66 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomPagedVarSeqSkipsSoftmaxQ128Kv128PersistentKeepsAbForGen.cubin 05-Jun-2026 22:36 71.78 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvCustomPagedVarSeqSkipsSoftmaxQ128Kv128StaticKeepsAbForGen.cubin 05-Jun-2026 22:36 65.11 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvDenseP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 71.52 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvDenseP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 69.74 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvDenseP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 70.82 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvDenseP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 69.04 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 87.60 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 82.87 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvDenseP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 64.74 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvDenseP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 59.57 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvDenseP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 62.71 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvDenseP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 57.54 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvDenseP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 67.02 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvDenseP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 65.24 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvDenseP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 66.32 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvDenseP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 64.41 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 84.47 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 76.81 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvDenseP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 61.49 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvDenseP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 55.32 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvDenseP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 59.46 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvDenseP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 53.29 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvDenseP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 65.39 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvDenseP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 63.61 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvDenseP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 64.70 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvDenseP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 62.79 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 81.97 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 74.68 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvDenseP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 59.37 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvDenseP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 53.57 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvDenseP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 57.21 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvDenseP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 51.54 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvDensePagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 67.77 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvDensePagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 65.86 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvDensePagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 66.95 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvDensePagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 65.29 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvDensePagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 83.22 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvDensePagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 75.80 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvDensePagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 61.49 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvDensePagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 55.32 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvDensePagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 59.21 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvDensePagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 53.41 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 74.52 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 72.49 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 73.70 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvSlidingOrChunkedCausalP16MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 71.79 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 94.72 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 87.45 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 70.10 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvSlidingOrChunkedCausalP16VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 59.93 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 67.07 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvSlidingOrChunkedCausalP16VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 57.77 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 70.02 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 67.99 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 69.20 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvSlidingOrChunkedCausalP32MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 67.29 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 91.34 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 83.32 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 66.35 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvSlidingOrChunkedCausalP32VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 56.18 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 63.32 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvSlidingOrChunkedCausalP32VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 54.02 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 68.39 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 66.36 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 67.57 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvSlidingOrChunkedCausalP64MultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 65.66 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 88.84 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 81.32 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 63.85 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvSlidingOrChunkedCausalP64VarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 54.18 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 60.82 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvSlidingOrChunkedCausalP64VarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 52.02 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 71.27 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvCgaVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 69.36 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 70.45 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvSlidingOrChunkedCausalPagedMultiCtasKvVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 68.54 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 90.97 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 83.19 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ16Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 64.72 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ16Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 56.68 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ8Kv128PersistentSwapsAbForGen.cubin 05-Jun-2026 22:36 62.32 KB
fmhaSm103aKernel_QE4m3KvE2m1OE4m3H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ8Kv128StaticSwapsAbForGen.cubin 05-Jun-2026 22:36 54.52 KB
fmhaSm103aKernel_QkBfloat16VE4m3OBfloat16H128SeparateQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 105.48 KB
fmhaSm103aKernel_QkBfloat16VE4m3OBfloat16H128SeparateQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 93.98 KB
fmhaSm103aKernel_QkBfloat16VE4m3OBfloat16H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 109.47 KB
fmhaSm103aKernel_QkBfloat16VE4m3OBfloat16H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 97.95 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 105.27 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 94.23 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 108.98 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 97.96 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 105.96 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 95.25 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 109.85 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 99.05 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 115.97 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 104.88 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 119.55 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 108.73 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 127.40 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 116.85 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 131.08 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 120.52 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 122.24 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 110.93 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 125.93 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 114.60 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 118.19 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 106.92 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 121.88 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 110.75 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PagedKvCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 119.66 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PagedKvCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 108.97 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 123.22 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 112.67 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 127.90 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 117.49 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 131.66 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 121.23 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 122.99 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 111.71 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 126.88 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 115.57 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 118.46 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 107.98 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 122.35 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 111.73 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PagedKvDensePagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 120.05 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PagedKvDensePagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 109.49 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 123.95 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 113.49 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 138.84 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 128.39 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 142.38 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 131.81 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 133.70 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 122.46 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 137.37 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 125.98 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 129.64 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 118.59 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 133.19 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 122.13 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 131.81 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 120.93 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 135.36 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 124.65 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128SeparateQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 105.40 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128SeparateQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 94.40 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128SeparateQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 109.09 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128SeparateQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 98.12 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128SeparateQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 106.09 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128SeparateQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 95.25 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 109.98 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 99.05 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 101.21 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 86.74 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 103.91 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 89.90 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 100.87 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 89.76 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 103.97 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 92.70 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 111.52 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 97.34 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 114.54 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 100.61 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 141.59 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 122.23 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 144.66 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 125.02 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 122.52 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 106.34 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 125.38 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 109.12 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 114.59 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 100.39 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 117.59 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 103.18 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PagedKvCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 115.01 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PagedKvCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 101.14 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 117.95 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 104.05 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 142.15 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 126.99 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 144.99 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 129.94 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 123.69 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 111.04 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 126.66 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 113.98 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 115.44 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 104.01 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 118.41 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 106.96 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PagedKvDensePagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 115.27 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PagedKvDensePagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 103.86 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 118.24 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 106.94 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 152.97 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 133.53 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 156.06 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 136.45 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 134.33 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 117.62 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 137.41 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 120.42 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 126.27 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 111.68 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 129.35 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 114.60 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 126.82 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 113.12 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 129.77 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 115.92 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256SeparateQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 100.92 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256SeparateQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 86.90 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256SeparateQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 103.99 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256SeparateQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 89.81 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256SeparateQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 100.99 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256SeparateQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 89.88 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 104.09 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H256SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 92.83 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H512HVPerCta256PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 109.94 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H512HVPerCta256PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 98.20 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H512HVPerCta256PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 112.64 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H512HVPerCta256PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 101.45 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H512HVPerCta256PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 109.62 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H512HVPerCta256PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 101.36 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H512HVPerCta256PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 112.60 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H512HVPerCta256PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 104.30 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 160.38 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 150.98 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 163.48 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 153.89 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 137.66 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 124.66 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 140.88 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 127.58 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 123.76 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 113.27 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 126.98 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 116.06 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 159.41 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 153.42 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 162.38 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 156.37 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 139.30 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 128.62 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 142.27 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 131.57 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 124.99 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 116.74 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 127.84 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 119.69 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 100.44 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 88.52 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 105.99 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 94.08 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 100.84 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 89.48 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 106.59 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 95.00 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 111.10 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 99.74 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 116.57 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 105.07 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 116.34 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 104.47 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 122.40 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 110.41 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 113.98 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 102.20 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 119.29 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 107.63 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 112.44 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 100.79 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 117.74 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 106.23 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PagedKvCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 113.74 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PagedKvCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 101.82 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 119.17 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 107.38 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 116.45 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 105.32 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 122.81 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 111.39 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 114.50 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 103.04 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 119.83 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 108.73 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 112.96 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 101.52 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 118.28 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 107.21 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PagedKvDensePagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 114.51 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PagedKvDensePagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 102.58 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 119.73 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 108.27 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 127.88 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 115.57 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 133.95 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 121.73 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 125.40 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 113.30 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 130.96 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 118.95 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 123.86 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 111.89 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 129.42 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 117.55 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 126.65 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 114.02 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 132.00 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 119.68 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H80PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 103.47 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H80PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 88.45 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H80PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 109.08 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H80PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 94.03 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H80PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 103.88 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H80PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 89.30 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H80PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 109.64 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H80PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 94.82 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H80PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 114.14 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H80PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 99.59 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H80PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 119.65 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H80PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 105.05 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H80PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 125.15 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H80PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 110.43 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H80PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 130.90 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H80PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 115.75 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H80PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 119.48 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H80PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 105.05 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H80PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 125.09 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H80PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 110.50 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H80PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 116.24 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H80PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 101.43 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H80PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 121.73 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H80PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 106.88 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H80PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 125.46 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H80PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 110.88 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H80PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 131.32 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H80PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 116.58 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H80PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 118.83 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H80PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 105.23 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H80PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 124.81 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H80PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 110.94 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H80PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 115.84 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H80PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 102.28 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H80PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 121.70 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H80PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 107.86 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H80PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 136.70 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H80PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 121.24 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H80PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 142.20 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H80PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 127.02 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H80PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 130.91 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H80PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 115.86 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H80PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 136.40 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H80PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 121.65 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H80PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 127.66 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H80PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 112.37 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H80PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 133.16 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16H80PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 118.03 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16HQk128HV64SeparateQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 107.50 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16HQk128HV64SeparateQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 96.31 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16HQk128HV64SeparateQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 112.69 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16HQk128HV64SeparateQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 101.36 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16HQk128HV64SeparateQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 108.30 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16HQk128HV64SeparateQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 97.04 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16HQk128HV64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 113.65 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16HQk128HV64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 102.28 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16HQk192HV128SeparateQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 112.43 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16HQk192HV128SeparateQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 99.55 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16HQk192HV128SeparateQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 116.23 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16HQk192HV128SeparateQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 103.16 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16HQk192HV128SeparateQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 113.23 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16HQk192HV128SeparateQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 101.82 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16HQk192HV128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 117.10 KB
fmhaSm103aKernel_QkvBfloat16OBfloat16HQk192HV128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 105.64 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 101.89 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 90.56 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 105.59 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 94.16 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 102.56 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PackedQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:36 112.39 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PackedQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:36 101.09 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PackedQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:36 118.45 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PackedQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:36 107.33 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PackedQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:36 114.22 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PackedQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:36 103.03 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 91.33 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 106.32 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:36 115.91 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:36 104.73 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:36 122.12 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:36 111.24 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:36 117.99 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:36 106.80 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 95.38 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 112.46 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 101.26 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 116.16 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 105.24 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 118.27 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 106.60 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 121.95 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 110.46 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 115.31 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 103.79 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 118.99 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 107.65 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 113.77 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 102.26 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 117.45 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 106.11 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PagedKvCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 116.13 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PagedKvCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 104.95 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 119.81 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 108.80 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 118.65 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 107.81 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 122.53 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 111.85 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 115.82 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 104.99 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 119.70 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 108.91 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 114.27 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 103.46 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 118.16 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 107.51 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PagedKvDensePagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 116.64 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PagedKvDensePagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 105.99 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 120.52 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 109.91 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 130.22 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 119.03 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 133.38 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 121.98 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 127.27 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 116.09 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 130.42 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 119.28 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 125.73 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 114.68 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 128.88 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 117.75 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 128.80 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 117.38 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 131.95 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 120.58 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128SeparateQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 102.02 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128SeparateQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 90.56 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128SeparateQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 105.70 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128SeparateQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 94.29 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128SeparateQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 102.56 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:36 112.39 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:36 101.09 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:36 118.58 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:36 107.45 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:36 114.34 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:36 103.16 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128SeparateQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 91.45 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 106.45 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:36 116.04 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:36 104.86 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:36 122.25 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:36 111.37 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:36 118.12 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:36 106.92 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 95.51 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 95.80 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 84.91 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 98.77 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 87.05 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 97.29 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 85.36 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 99.87 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 88.02 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 107.94 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 95.57 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 110.61 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 98.29 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 118.55 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 106.61 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 120.92 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 108.75 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 111.53 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 100.15 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 113.91 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 102.30 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 108.45 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 97.21 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 110.83 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 99.23 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PagedKvCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 111.35 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PagedKvCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 98.95 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 114.27 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 101.09 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 121.04 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 107.37 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 123.87 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 110.02 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 112.89 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 100.95 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 115.73 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 103.73 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 109.81 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 98.01 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 112.52 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 100.80 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PagedKvDensePagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 111.33 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PagedKvDensePagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 99.64 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 114.16 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 102.18 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 132.59 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 118.41 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 135.25 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 120.91 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 124.54 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 111.80 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 127.16 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 114.31 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 121.45 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 108.87 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 123.96 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 111.38 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 123.67 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 111.05 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 126.30 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 113.68 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256SeparateQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 97.46 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256SeparateQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 84.91 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256SeparateQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 98.88 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256SeparateQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 87.17 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256SeparateQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 97.41 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256SeparateQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 85.48 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 99.99 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H256SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 88.15 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H512HVPerCta256PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 105.59 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H512HVPerCta256PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 93.80 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H512HVPerCta256PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 108.69 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H512HVPerCta256PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 95.80 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H512HVPerCta256PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 107.20 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H512HVPerCta256PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 94.28 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H512HVPerCta256PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 109.91 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H512HVPerCta256PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 97.00 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 135.80 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 123.16 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 138.18 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 125.18 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 123.73 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 111.87 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 126.12 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 113.88 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 118.85 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 107.15 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 121.23 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 109.17 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 137.27 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 123.56 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 140.11 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 126.29 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 125.25 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 112.59 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 128.08 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 115.31 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 120.38 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 107.87 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 123.20 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 110.59 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 95.72 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 83.52 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 99.72 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 87.51 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 96.24 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PackedQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:36 111.62 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PackedQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:36 88.94 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PackedQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:36 114.19 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PackedQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:36 94.89 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PackedQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:36 113.62 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PackedQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:36 90.75 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 84.33 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 100.40 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:36 112.16 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:36 92.52 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:36 118.02 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:36 98.66 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:36 114.39 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:36 94.58 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 88.31 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 106.13 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 94.40 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 110.29 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 98.38 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 112.42 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 99.58 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 116.40 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 103.55 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 109.34 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 96.77 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 113.31 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 100.75 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 107.80 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 95.38 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 111.77 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 99.34 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PagedKvCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 109.07 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PagedKvCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 97.19 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 113.17 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 101.04 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 112.66 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 100.57 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 116.80 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 104.55 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 109.57 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 97.89 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 113.72 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 101.76 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 108.16 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 96.24 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 112.18 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 100.23 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PagedKvDensePagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 109.68 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PagedKvDensePagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 97.80 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 113.70 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 101.78 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 124.33 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 112.20 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 127.70 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 115.53 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 121.24 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 109.41 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 124.73 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 112.86 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 119.70 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 108.00 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 123.20 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 111.33 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 121.84 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 110.05 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 125.32 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 113.50 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:36 111.74 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:36 88.94 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:36 114.31 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:36 95.02 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:36 113.74 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:36 90.88 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:36 112.28 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:36 92.64 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:36 118.14 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:36 98.79 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:36 114.52 KB
fmhaSm103aKernel_QkvE4m3OBfloat16H64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:36 94.70 KB
fmhaSm103aKernel_QkvE4m3OBfloat16HQk128HV64SeparateQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 99.27 KB
fmhaSm103aKernel_QkvE4m3OBfloat16HQk128HV64SeparateQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 87.30 KB
fmhaSm103aKernel_QkvE4m3OBfloat16HQk128HV64SeparateQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 103.22 KB
fmhaSm103aKernel_QkvE4m3OBfloat16HQk128HV64SeparateQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 91.15 KB
fmhaSm103aKernel_QkvE4m3OBfloat16HQk128HV64SeparateQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 99.80 KB
fmhaSm103aKernel_QkvE4m3OBfloat16HQk128HV64SeparateQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 87.98 KB
fmhaSm103aKernel_QkvE4m3OBfloat16HQk128HV64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 103.91 KB
fmhaSm103aKernel_QkvE4m3OBfloat16HQk128HV64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 91.96 KB
fmhaSm103aKernel_QkvE4m3OBfloat16HQk192HV128SeparateQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 107.60 KB
fmhaSm103aKernel_QkvE4m3OBfloat16HQk192HV128SeparateQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 94.32 KB
fmhaSm103aKernel_QkvE4m3OBfloat16HQk192HV128SeparateQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 111.15 KB
fmhaSm103aKernel_QkvE4m3OBfloat16HQk192HV128SeparateQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 98.04 KB
fmhaSm103aKernel_QkvE4m3OBfloat16HQk192HV128SeparateQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 108.16 KB
fmhaSm103aKernel_QkvE4m3OBfloat16HQk192HV128SeparateQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 96.56 KB
fmhaSm103aKernel_QkvE4m3OBfloat16HQk192HV128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 111.90 KB
fmhaSm103aKernel_QkvE4m3OBfloat16HQk192HV128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 100.61 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 103.19 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 91.77 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 106.89 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 95.50 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 103.86 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PackedQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:36 114.46 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PackedQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:36 102.72 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PackedQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:36 120.65 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PackedQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:36 109.08 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PackedQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:36 116.54 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PackedQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:36 104.78 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 92.66 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 107.75 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:36 118.11 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:36 106.48 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:36 124.45 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:36 112.98 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:36 120.19 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:36 108.54 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 96.71 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 113.88 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 102.38 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 117.46 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 106.24 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 119.84 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 107.85 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 123.52 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 111.59 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 116.89 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 104.91 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 120.57 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 108.77 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 115.35 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 103.51 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 119.03 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 107.23 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PagedKvCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 117.59 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PagedKvCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 106.20 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 121.15 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 110.05 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 120.23 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 108.94 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 123.98 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 112.98 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 117.40 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 106.12 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 121.16 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 110.16 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 115.86 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 104.59 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 119.62 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 108.63 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PagedKvDensePagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 118.10 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PagedKvDensePagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 106.99 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 121.98 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 110.91 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 131.80 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 120.21 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 134.84 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 123.41 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 128.84 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 117.40 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 132.01 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 120.59 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 127.30 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 115.98 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 130.47 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 119.18 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 130.38 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 118.81 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 133.41 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 122.01 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128SeparateQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 103.98 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:36 114.59 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:36 102.84 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:36 120.65 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:36 109.08 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:36 116.54 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:36 104.78 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128SeparateQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 92.79 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 107.88 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:36 118.23 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:36 106.60 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:36 124.45 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:36 112.98 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:36 120.19 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:36 108.54 KB
fmhaSm103aKernel_QkvE4m3OE2m1H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 96.71 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 96.86 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 85.48 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 99.71 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 87.74 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 98.23 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 86.18 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 100.93 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 88.72 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 109.00 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 96.27 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 111.67 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 99.11 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 119.48 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 107.43 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 121.98 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 109.32 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 112.48 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 100.98 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 114.98 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 102.87 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 109.39 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 97.91 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 111.89 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 99.92 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PagedKvCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 112.41 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PagedKvCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 99.77 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 115.22 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 101.79 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 122.29 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 108.06 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 124.88 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 110.59 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 113.96 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 101.77 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 116.66 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 104.30 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 110.75 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 98.83 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 113.46 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 101.37 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PagedKvDensePagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 112.39 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PagedKvDensePagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 100.22 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 115.10 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 102.88 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 133.72 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 119.10 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 136.38 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 121.73 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 125.60 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 112.38 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 128.23 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 115.02 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 122.39 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 109.44 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 125.15 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 112.07 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 124.74 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 111.74 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 127.37 KB
fmhaSm103aKernel_QkvE4m3OE2m1H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 114.25 KB
fmhaSm103aKernel_QkvE4m3OE2m1H512HVPerCta256PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 106.66 KB
fmhaSm103aKernel_QkvE4m3OE2m1H512HVPerCta256PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 94.40 KB
fmhaSm103aKernel_QkvE4m3OE2m1H512HVPerCta256PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 109.50 KB
fmhaSm103aKernel_QkvE4m3OE2m1H512HVPerCta256PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 96.49 KB
fmhaSm103aKernel_QkvE4m3OE2m1H512HVPerCta256PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 108.27 KB
fmhaSm103aKernel_QkvE4m3OE2m1H512HVPerCta256PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 95.01 KB
fmhaSm103aKernel_QkvE4m3OE2m1H512HVPerCta256PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 110.85 KB
fmhaSm103aKernel_QkvE4m3OE2m1H512HVPerCta256PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 97.57 KB
fmhaSm103aKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 136.75 KB
fmhaSm103aKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 123.89 KB
fmhaSm103aKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 139.25 KB
fmhaSm103aKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 125.88 KB
fmhaSm103aKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 124.68 KB
fmhaSm103aKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 112.60 KB
fmhaSm103aKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 127.18 KB
fmhaSm103aKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 114.59 KB
fmhaSm103aKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 119.80 KB
fmhaSm103aKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 107.88 KB
fmhaSm103aKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 122.30 KB
fmhaSm103aKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 109.87 KB
fmhaSm103aKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 138.53 KB
fmhaSm103aKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 124.30 KB
fmhaSm103aKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 141.24 KB
fmhaSm103aKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 126.86 KB
fmhaSm103aKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 126.31 KB
fmhaSm103aKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 113.31 KB
fmhaSm103aKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 129.02 KB
fmhaSm103aKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 115.88 KB
fmhaSm103aKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 121.44 KB
fmhaSm103aKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 108.59 KB
fmhaSm103aKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 124.15 KB
fmhaSm103aKernel_QkvE4m3OE2m1H512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 111.16 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 100.10 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 83.97 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 103.69 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 87.95 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 100.72 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PackedQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:36 101.75 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PackedQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:36 90.37 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PackedQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:36 107.89 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PackedQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:36 96.45 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PackedQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:36 103.84 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PackedQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:35 92.30 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 84.53 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 104.24 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:36 105.63 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:36 94.20 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:36 111.84 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:36 100.23 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:36 107.85 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:35 96.27 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 88.64 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 110.64 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 94.56 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 114.26 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 98.55 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 116.65 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 99.77 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 120.09 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 103.73 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 113.54 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 97.09 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 116.98 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 100.93 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 111.99 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 95.56 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 115.44 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 99.53 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PagedKvCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 113.42 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PagedKvCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 97.38 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 116.87 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 101.35 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 115.91 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 100.88 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 119.39 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 104.87 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 112.82 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 98.08 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 116.41 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 102.07 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 111.24 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 96.55 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 114.84 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 100.54 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PagedKvDensePagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 112.68 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PagedKvDensePagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 98.11 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 116.28 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 102.09 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 128.06 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 112.24 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 131.40 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 115.57 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 124.95 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 109.44 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 128.41 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 112.89 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 123.41 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 108.04 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 126.87 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 111.37 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 125.55 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 110.09 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 129.02 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 113.54 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:35 101.88 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:35 90.49 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:35 107.89 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:35 96.45 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:36 103.84 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:35 92.30 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:35 105.76 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:35 94.33 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:36 111.84 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:35 100.23 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:35 107.85 KB
fmhaSm103aKernel_QkvE4m3OE2m1H64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:35 96.27 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 107.76 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 96.15 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:36 111.45 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 100.00 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 108.43 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PackedQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:35 125.34 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PackedQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:35 114.17 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PackedQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:36 131.40 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PackedQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:35 120.53 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PackedQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:35 127.29 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PackedQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:35 116.11 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 97.16 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 112.31 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:36 128.98 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:35 117.81 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:36 135.20 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:35 124.32 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:35 130.94 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:35 119.88 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 101.09 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 118.45 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 106.77 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 122.02 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 110.75 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 124.41 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 112.23 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 128.09 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 116.09 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 121.45 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 109.42 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 125.13 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 113.28 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 119.91 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 107.89 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 123.59 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 111.74 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PagedKvCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 122.16 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PagedKvCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 110.58 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 125.71 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:36 114.44 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 124.80 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 113.45 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 128.55 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 117.36 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 121.96 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 110.62 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 125.73 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 114.55 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 120.42 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 109.09 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 124.18 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 113.02 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PagedKvDensePagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 122.66 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PagedKvDensePagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 111.38 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 126.55 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 115.26 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 136.37 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 124.59 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 139.41 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 127.91 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 133.41 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 121.90 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 136.45 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 124.97 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 131.88 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 120.24 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 134.91 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 123.44 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 134.95 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 123.07 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 137.98 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 126.39 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128SeparateQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 108.55 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:35 125.46 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:35 114.30 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:35 131.52 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:35 120.53 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:35 127.41 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:35 116.11 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128SeparateQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 97.29 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 112.44 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:35 128.98 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:35 117.94 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:35 135.32 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:35 124.32 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:35 131.06 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:35 119.88 KB
fmhaSm103aKernel_QkvE4m3OE4m3H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 101.09 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 101.27 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 89.84 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 104.12 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 91.98 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 102.63 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 90.41 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 105.34 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 92.95 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 113.41 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 100.51 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 116.08 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 103.35 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 123.89 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 111.66 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 126.39 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 113.68 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 116.88 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 105.21 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 119.38 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 107.23 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 113.80 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 102.14 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 116.30 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 104.16 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PagedKvCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 116.82 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PagedKvCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 104.01 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 119.62 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 106.03 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 126.49 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 112.30 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 129.20 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 114.96 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 118.36 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 106.01 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 121.07 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 108.55 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 115.28 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 103.07 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 117.87 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 105.60 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PagedKvDensePagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 116.80 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PagedKvDensePagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 104.45 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 119.51 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 107.23 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 137.91 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 123.34 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 140.70 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 125.97 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 130.01 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 116.62 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 132.64 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 119.38 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 126.80 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 113.67 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 129.55 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 116.30 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 129.15 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 115.98 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 131.77 KB
fmhaSm103aKernel_QkvE4m3OE4m3H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 118.49 KB
fmhaSm103aKernel_QkvE4m3OE4m3H512HVPerCta256PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 110.99 KB
fmhaSm103aKernel_QkvE4m3OE4m3H512HVPerCta256PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 98.54 KB
fmhaSm103aKernel_QkvE4m3OE4m3H512HVPerCta256PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 113.84 KB
fmhaSm103aKernel_QkvE4m3OE4m3H512HVPerCta256PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 100.66 KB
fmhaSm103aKernel_QkvE4m3OE4m3H512HVPerCta256PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 112.48 KB
fmhaSm103aKernel_QkvE4m3OE4m3H512HVPerCta256PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 99.13 KB
fmhaSm103aKernel_QkvE4m3OE4m3H512HVPerCta256PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 115.07 KB
fmhaSm103aKernel_QkvE4m3OE4m3H512HVPerCta256PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 101.73 KB
fmhaSm103aKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 141.87 KB
fmhaSm103aKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 128.03 KB
fmhaSm103aKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 144.24 KB
fmhaSm103aKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 130.05 KB
fmhaSm103aKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 129.79 KB
fmhaSm103aKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 116.74 KB
fmhaSm103aKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 132.29 KB
fmhaSm103aKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 118.76 KB
fmhaSm103aKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 124.91 KB
fmhaSm103aKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 112.02 KB
fmhaSm103aKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 127.41 KB
fmhaSm103aKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 114.04 KB
fmhaSm103aKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 142.67 KB
fmhaSm103aKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 128.42 KB
fmhaSm103aKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 145.38 KB
fmhaSm103aKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 131.02 KB
fmhaSm103aKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 130.66 KB
fmhaSm103aKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 117.44 KB
fmhaSm103aKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 133.37 KB
fmhaSm103aKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 120.04 KB
fmhaSm103aKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 125.78 KB
fmhaSm103aKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 112.72 KB
fmhaSm103aKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 128.49 KB
fmhaSm103aKernel_QkvE4m3OE4m3H512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 115.32 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 95.72 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 83.81 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 99.72 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 87.80 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 96.12 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PackedQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:35 105.42 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PackedQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:35 93.55 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PackedQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:35 111.47 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PackedQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:35 99.62 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PackedQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:35 107.39 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PackedQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:35 95.49 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 84.38 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 100.27 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:35 109.30 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:35 97.38 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:35 115.42 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:35 103.53 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:35 111.40 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:35 99.45 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 88.48 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 106.13 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 94.40 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 110.29 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 98.51 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 112.30 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 99.73 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 116.27 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 103.58 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 109.21 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 96.92 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 113.19 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 100.90 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 107.67 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 95.40 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 111.65 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 99.38 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PagedKvCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 109.20 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PagedKvCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 97.22 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 113.17 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 101.19 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 112.66 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 100.72 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 116.80 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 104.71 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 109.70 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 97.91 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 113.72 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 102.03 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 108.16 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 96.39 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 112.18 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 100.38 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PagedKvDensePagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 109.68 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PagedKvDensePagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 97.95 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 113.58 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 102.05 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 124.20 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 112.08 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 127.70 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 115.41 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 121.12 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 109.28 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 124.61 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 112.73 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 119.58 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 107.75 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 123.07 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 111.20 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 121.84 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 109.92 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 125.32 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 113.38 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:35 105.55 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:35 93.68 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:35 111.47 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:35 99.62 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:35 107.39 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:35 95.49 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:35 109.43 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:35 97.51 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:35 115.42 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:35 103.53 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:35 111.40 KB
fmhaSm103aKernel_QkvE4m3OE4m3H64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:35 99.45 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 102.14 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 90.81 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 105.84 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 94.55 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 102.81 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PackedQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:35 112.39 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PackedQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:35 101.10 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PackedQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:35 118.58 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PackedQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:35 107.46 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PackedQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:35 114.34 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PackedQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:35 103.16 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 91.71 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 106.70 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:35 116.04 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:35 104.86 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:35 122.25 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:35 111.37 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:35 118.12 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:35 106.92 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 95.76 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 112.84 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 101.51 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 116.41 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 105.49 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 118.64 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 106.98 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 122.32 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 110.84 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 115.69 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 104.17 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 119.37 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 108.02 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 114.15 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 102.63 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 117.83 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 106.49 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PagedKvCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 116.52 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PagedKvCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 105.33 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 120.07 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 109.18 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 119.02 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 108.19 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 122.78 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 112.23 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 116.20 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 105.38 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 119.95 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 109.29 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 114.66 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 103.84 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 118.41 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 107.88 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PagedKvDensePagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 116.90 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PagedKvDensePagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 106.24 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 120.78 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 110.16 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 130.60 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 119.28 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 133.76 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 122.35 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 127.65 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 116.34 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 130.80 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 119.66 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 126.10 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 114.94 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 129.27 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 118.13 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 129.18 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 117.77 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 132.34 KB
fmhaSm103aKernel_QkvE4m3OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 120.96 KB
fmhaSm103aKernel_QkvE4m3OFp16H128SeparateQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 102.94 KB
fmhaSm103aKernel_QkvE4m3OFp16H128SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:35 112.52 KB
fmhaSm103aKernel_QkvE4m3OFp16H128SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:35 101.23 KB
fmhaSm103aKernel_QkvE4m3OFp16H128SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:35 118.58 KB
fmhaSm103aKernel_QkvE4m3OFp16H128SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:35 107.46 KB
fmhaSm103aKernel_QkvE4m3OFp16H128SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:35 114.34 KB
fmhaSm103aKernel_QkvE4m3OFp16H128SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:35 103.16 KB
fmhaSm103aKernel_QkvE4m3OFp16H128SeparateQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 91.84 KB
fmhaSm103aKernel_QkvE4m3OFp16H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 106.83 KB
fmhaSm103aKernel_QkvE4m3OFp16H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:35 116.16 KB
fmhaSm103aKernel_QkvE4m3OFp16H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:35 104.98 KB
fmhaSm103aKernel_QkvE4m3OFp16H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:35 122.25 KB
fmhaSm103aKernel_QkvE4m3OFp16H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:35 111.37 KB
fmhaSm103aKernel_QkvE4m3OFp16H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:35 118.12 KB
fmhaSm103aKernel_QkvE4m3OFp16H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:35 106.92 KB
fmhaSm103aKernel_QkvE4m3OFp16H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 95.76 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 96.17 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 85.16 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 99.02 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 87.42 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 97.54 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 85.73 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 100.24 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 88.27 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 108.31 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 95.83 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 110.86 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 98.67 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 118.80 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 106.98 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 121.30 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 109.00 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 111.79 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 100.53 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 114.29 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 102.55 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 108.70 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 97.59 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 111.20 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 99.48 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PagedKvCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 111.73 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PagedKvCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 99.33 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 114.53 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 101.35 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 121.41 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 107.74 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 124.12 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 110.28 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 113.27 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 101.33 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 115.98 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 103.99 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 110.19 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 98.39 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 112.77 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 101.05 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PagedKvDensePagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 111.70 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PagedKvDensePagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 99.90 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 114.41 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 102.55 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 132.84 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 118.66 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 135.62 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 121.29 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 124.79 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 112.06 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 127.55 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 114.70 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 121.71 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 109.12 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 124.34 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 111.76 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 124.05 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 111.43 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 126.55 KB
fmhaSm103aKernel_QkvE4m3OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 113.94 KB
fmhaSm103aKernel_QkvE4m3OFp16H512HVPerCta256PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 105.97 KB
fmhaSm103aKernel_QkvE4m3OFp16H512HVPerCta256PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 94.05 KB
fmhaSm103aKernel_QkvE4m3OFp16H512HVPerCta256PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 108.95 KB
fmhaSm103aKernel_QkvE4m3OFp16H512HVPerCta256PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 96.17 KB
fmhaSm103aKernel_QkvE4m3OFp16H512HVPerCta256PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 107.59 KB
fmhaSm103aKernel_QkvE4m3OFp16H512HVPerCta256PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 94.66 KB
fmhaSm103aKernel_QkvE4m3OFp16H512HVPerCta256PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 110.16 KB
fmhaSm103aKernel_QkvE4m3OFp16H512HVPerCta256PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 97.26 KB
fmhaSm103aKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 136.06 KB
fmhaSm103aKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 123.54 KB
fmhaSm103aKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 138.56 KB
fmhaSm103aKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 125.55 KB
fmhaSm103aKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 123.98 KB
fmhaSm103aKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 112.25 KB
fmhaSm103aKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 126.49 KB
fmhaSm103aKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 114.27 KB
fmhaSm103aKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 119.10 KB
fmhaSm103aKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 107.53 KB
fmhaSm103aKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 121.62 KB
fmhaSm103aKernel_QkvE4m3OFp16H512HVPerCta256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 109.55 KB
fmhaSm103aKernel_QkvE4m3OFp16H512HVPerCta256PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 137.66 KB
fmhaSm103aKernel_QkvE4m3OFp16H512HVPerCta256PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 123.95 KB
fmhaSm103aKernel_QkvE4m3OFp16H512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 140.37 KB
fmhaSm103aKernel_QkvE4m3OFp16H512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 126.55 KB
fmhaSm103aKernel_QkvE4m3OFp16H512HVPerCta256PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 125.62 KB
fmhaSm103aKernel_QkvE4m3OFp16H512HVPerCta256PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 112.96 KB
fmhaSm103aKernel_QkvE4m3OFp16H512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 128.34 KB
fmhaSm103aKernel_QkvE4m3OFp16H512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 115.56 KB
fmhaSm103aKernel_QkvE4m3OFp16H512HVPerCta256PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 120.75 KB
fmhaSm103aKernel_QkvE4m3OFp16H512HVPerCta256PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 108.24 KB
fmhaSm103aKernel_QkvE4m3OFp16H512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 123.46 KB
fmhaSm103aKernel_QkvE4m3OFp16H512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 110.84 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 96.60 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 84.41 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 100.59 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 88.39 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 97.00 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PackedQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:35 112.50 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PackedQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:35 88.94 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PackedQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:35 115.07 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PackedQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:35 95.02 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PackedQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:35 114.38 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PackedQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:35 90.88 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 85.09 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 101.15 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:35 113.04 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:35 92.65 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:35 118.90 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:35 98.79 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:35 115.27 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:35 94.71 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 89.07 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 107.02 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 95.28 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 111.17 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 99.27 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 113.18 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 100.34 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 117.16 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 104.30 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 110.09 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 97.53 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 114.07 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 101.51 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 108.55 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 96.13 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 112.53 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 100.10 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PagedKvCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 109.95 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PagedKvCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 97.95 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 114.05 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 101.80 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 113.54 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 101.33 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 117.69 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 105.44 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 110.45 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 98.65 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 114.60 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 102.64 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 109.04 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 97.00 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 113.06 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 101.11 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PagedKvDensePagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 110.56 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PagedKvDensePagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 98.68 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 114.46 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 102.66 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 125.09 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 112.96 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 128.45 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 116.29 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 122.00 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 110.16 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 125.49 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 113.62 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 120.46 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 108.76 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 123.95 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 112.09 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 122.72 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 110.80 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 126.20 KB
fmhaSm103aKernel_QkvE4m3OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 114.26 KB
fmhaSm103aKernel_QkvE4m3OFp16H64SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:35 112.62 KB
fmhaSm103aKernel_QkvE4m3OFp16H64SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:35 89.06 KB
fmhaSm103aKernel_QkvE4m3OFp16H64SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:35 115.07 KB
fmhaSm103aKernel_QkvE4m3OFp16H64SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:35 95.02 KB
fmhaSm103aKernel_QkvE4m3OFp16H64SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:35 114.50 KB
fmhaSm103aKernel_QkvE4m3OFp16H64SeparateQkvDenseVarSeqQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:35 90.88 KB
fmhaSm103aKernel_QkvE4m3OFp16H64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1PersistentContext.cubin 05-Jun-2026 22:35 113.16 KB
fmhaSm103aKernel_QkvE4m3OFp16H64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK16SageV1StaticContext.cubin 05-Jun-2026 22:35 92.77 KB
fmhaSm103aKernel_QkvE4m3OFp16H64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1PersistentContext.cubin 05-Jun-2026 22:35 118.90 KB
fmhaSm103aKernel_QkvE4m3OFp16H64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK1SageV1StaticContext.cubin 05-Jun-2026 22:35 98.79 KB
fmhaSm103aKernel_QkvE4m3OFp16H64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1PersistentContext.cubin 05-Jun-2026 22:35 115.27 KB
fmhaSm103aKernel_QkvE4m3OFp16H64SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128SageQ1SageK4SageV1StaticContext.cubin 05-Jun-2026 22:35 94.71 KB
fmhaSm103aKernel_QkvFp16OFp16H128PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 106.41 KB
fmhaSm103aKernel_QkvFp16OFp16H128PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 95.36 KB
fmhaSm103aKernel_QkvFp16OFp16H128PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 110.11 KB
fmhaSm103aKernel_QkvFp16OFp16H128PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 99.09 KB
fmhaSm103aKernel_QkvFp16OFp16H128PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 107.09 KB
fmhaSm103aKernel_QkvFp16OFp16H128PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 96.38 KB
fmhaSm103aKernel_QkvFp16OFp16H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 110.99 KB
fmhaSm103aKernel_QkvFp16OFp16H128PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 100.18 KB
fmhaSm103aKernel_QkvFp16OFp16H128PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 117.10 KB
fmhaSm103aKernel_QkvFp16OFp16H128PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 106.01 KB
fmhaSm103aKernel_QkvFp16OFp16H128PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 120.69 KB
fmhaSm103aKernel_QkvFp16OFp16H128PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 109.88 KB
fmhaSm103aKernel_QkvFp16OFp16H128PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 128.53 KB
fmhaSm103aKernel_QkvFp16OFp16H128PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 117.98 KB
fmhaSm103aKernel_QkvFp16OFp16H128PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 132.22 KB
fmhaSm103aKernel_QkvFp16OFp16H128PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 121.66 KB
fmhaSm103aKernel_QkvFp16OFp16H128PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 123.38 KB
fmhaSm103aKernel_QkvFp16OFp16H128PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 112.06 KB
fmhaSm103aKernel_QkvFp16OFp16H128PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 127.06 KB
fmhaSm103aKernel_QkvFp16OFp16H128PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 115.73 KB
fmhaSm103aKernel_QkvFp16OFp16H128PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 119.32 KB
fmhaSm103aKernel_QkvFp16OFp16H128PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 108.18 KB
fmhaSm103aKernel_QkvFp16OFp16H128PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 123.01 KB
fmhaSm103aKernel_QkvFp16OFp16H128PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 111.88 KB
fmhaSm103aKernel_QkvFp16OFp16H128PagedKvCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 120.80 KB
fmhaSm103aKernel_QkvFp16OFp16H128PagedKvCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 110.11 KB
fmhaSm103aKernel_QkvFp16OFp16H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 124.35 KB
fmhaSm103aKernel_QkvFp16OFp16H128PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 113.80 KB
fmhaSm103aKernel_QkvFp16OFp16H128PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 129.16 KB
fmhaSm103aKernel_QkvFp16OFp16H128PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 118.63 KB
fmhaSm103aKernel_QkvFp16OFp16H128PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 132.80 KB
fmhaSm103aKernel_QkvFp16OFp16H128PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 122.49 KB
fmhaSm103aKernel_QkvFp16OFp16H128PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 124.25 KB
fmhaSm103aKernel_QkvFp16OFp16H128PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 112.84 KB
fmhaSm103aKernel_QkvFp16OFp16H128PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 128.02 KB
fmhaSm103aKernel_QkvFp16OFp16H128PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 116.83 KB
fmhaSm103aKernel_QkvFp16OFp16H128PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 119.72 KB
fmhaSm103aKernel_QkvFp16OFp16H128PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 109.12 KB
fmhaSm103aKernel_QkvFp16OFp16H128PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 123.48 KB
fmhaSm103aKernel_QkvFp16OFp16H128PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 112.99 KB
fmhaSm103aKernel_QkvFp16OFp16H128PagedKvDensePagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 121.20 KB
fmhaSm103aKernel_QkvFp16OFp16H128PagedKvDensePagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 110.63 KB
fmhaSm103aKernel_QkvFp16OFp16H128PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 125.08 KB
fmhaSm103aKernel_QkvFp16OFp16H128PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 114.48 KB
fmhaSm103aKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 139.97 KB
fmhaSm103aKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 129.53 KB
fmhaSm103aKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 143.52 KB
fmhaSm103aKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 132.95 KB
fmhaSm103aKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 134.84 KB
fmhaSm103aKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 123.59 KB
fmhaSm103aKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 138.50 KB
fmhaSm103aKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 127.12 KB
fmhaSm103aKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 130.77 KB
fmhaSm103aKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 119.73 KB
fmhaSm103aKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 134.32 KB
fmhaSm103aKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 123.27 KB
fmhaSm103aKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 132.95 KB
fmhaSm103aKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 122.07 KB
fmhaSm103aKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 136.62 KB
fmhaSm103aKernel_QkvFp16OFp16H128PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 125.78 KB
fmhaSm103aKernel_QkvFp16OFp16H128SeparateQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 107.22 KB
fmhaSm103aKernel_QkvFp16OFp16H128SeparateQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 96.51 KB
fmhaSm103aKernel_QkvFp16OFp16H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 111.12 KB
fmhaSm103aKernel_QkvFp16OFp16H128SeparateQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 100.18 KB
fmhaSm103aKernel_QkvFp16OFp16H256PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 102.34 KB
fmhaSm103aKernel_QkvFp16OFp16H256PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 87.88 KB
fmhaSm103aKernel_QkvFp16OFp16H256PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 105.05 KB
fmhaSm103aKernel_QkvFp16OFp16H256PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 91.03 KB
fmhaSm103aKernel_QkvFp16OFp16H256PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 102.01 KB
fmhaSm103aKernel_QkvFp16OFp16H256PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 90.90 KB
fmhaSm103aKernel_QkvFp16OFp16H256PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 105.10 KB
fmhaSm103aKernel_QkvFp16OFp16H256PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 93.84 KB
fmhaSm103aKernel_QkvFp16OFp16H256PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 112.66 KB
fmhaSm103aKernel_QkvFp16OFp16H256PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 98.48 KB
fmhaSm103aKernel_QkvFp16OFp16H256PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 115.67 KB
fmhaSm103aKernel_QkvFp16OFp16H256PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 101.57 KB
fmhaSm103aKernel_QkvFp16OFp16H256PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 142.73 KB
fmhaSm103aKernel_QkvFp16OFp16H256PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 123.37 KB
fmhaSm103aKernel_QkvFp16OFp16H256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 145.67 KB
fmhaSm103aKernel_QkvFp16OFp16H256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 126.16 KB
fmhaSm103aKernel_QkvFp16OFp16H256PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 123.65 KB
fmhaSm103aKernel_QkvFp16OFp16H256PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 107.48 KB
fmhaSm103aKernel_QkvFp16OFp16H256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 126.40 KB
fmhaSm103aKernel_QkvFp16OFp16H256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 110.27 KB
fmhaSm103aKernel_QkvFp16OFp16H256PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 115.73 KB
fmhaSm103aKernel_QkvFp16OFp16H256PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 101.53 KB
fmhaSm103aKernel_QkvFp16OFp16H256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 118.59 KB
fmhaSm103aKernel_QkvFp16OFp16H256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 104.32 KB
fmhaSm103aKernel_QkvFp16OFp16H256PagedKvCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 116.14 KB
fmhaSm103aKernel_QkvFp16OFp16H256PagedKvCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 102.27 KB
fmhaSm103aKernel_QkvFp16OFp16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 119.09 KB
fmhaSm103aKernel_QkvFp16OFp16H256PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 105.19 KB
fmhaSm103aKernel_QkvFp16OFp16H256PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 143.28 KB
fmhaSm103aKernel_QkvFp16OFp16H256PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 128.12 KB
fmhaSm103aKernel_QkvFp16OFp16H256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 146.13 KB
fmhaSm103aKernel_QkvFp16OFp16H256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 131.07 KB
fmhaSm103aKernel_QkvFp16OFp16H256PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 124.83 KB
fmhaSm103aKernel_QkvFp16OFp16H256PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 112.18 KB
fmhaSm103aKernel_QkvFp16OFp16H256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 127.80 KB
fmhaSm103aKernel_QkvFp16OFp16H256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 115.12 KB
fmhaSm103aKernel_QkvFp16OFp16H256PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 116.57 KB
fmhaSm103aKernel_QkvFp16OFp16H256PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 105.15 KB
fmhaSm103aKernel_QkvFp16OFp16H256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 119.55 KB
fmhaSm103aKernel_QkvFp16OFp16H256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 108.09 KB
fmhaSm103aKernel_QkvFp16OFp16H256PagedKvDensePagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 116.41 KB
fmhaSm103aKernel_QkvFp16OFp16H256PagedKvDensePagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 105.00 KB
fmhaSm103aKernel_QkvFp16OFp16H256PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 119.38 KB
fmhaSm103aKernel_QkvFp16OFp16H256PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 108.07 KB
fmhaSm103aKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 154.11 KB
fmhaSm103aKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 134.67 KB
fmhaSm103aKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 157.20 KB
fmhaSm103aKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 137.71 KB
fmhaSm103aKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 135.46 KB
fmhaSm103aKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 118.76 KB
fmhaSm103aKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 138.55 KB
fmhaSm103aKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 121.68 KB
fmhaSm103aKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 127.41 KB
fmhaSm103aKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 112.82 KB
fmhaSm103aKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 130.48 KB
fmhaSm103aKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 115.86 KB
fmhaSm103aKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 127.96 KB
fmhaSm103aKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 114.27 KB
fmhaSm103aKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 130.91 KB
fmhaSm103aKernel_QkvFp16OFp16H256PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 117.06 KB
fmhaSm103aKernel_QkvFp16OFp16H512HVPerCta256PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 111.07 KB
fmhaSm103aKernel_QkvFp16OFp16H512HVPerCta256PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 99.34 KB
fmhaSm103aKernel_QkvFp16OFp16H512HVPerCta256PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 113.77 KB
fmhaSm103aKernel_QkvFp16OFp16H512HVPerCta256PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 102.59 KB
fmhaSm103aKernel_QkvFp16OFp16H512HVPerCta256PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 110.77 KB
fmhaSm103aKernel_QkvFp16OFp16H512HVPerCta256PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 102.49 KB
fmhaSm103aKernel_QkvFp16OFp16H512HVPerCta256PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 113.73 KB
fmhaSm103aKernel_QkvFp16OFp16H512HVPerCta256PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 105.45 KB
fmhaSm103aKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 161.52 KB
fmhaSm103aKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 152.12 KB
fmhaSm103aKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 164.62 KB
fmhaSm103aKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 155.03 KB
fmhaSm103aKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 138.80 KB
fmhaSm103aKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 125.80 KB
fmhaSm103aKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 142.02 KB
fmhaSm103aKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 128.72 KB
fmhaSm103aKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 124.89 KB
fmhaSm103aKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 114.41 KB
fmhaSm103aKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 128.12 KB
fmhaSm103aKernel_QkvFp16OFp16H512HVPerCta256PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 117.20 KB
fmhaSm103aKernel_QkvFp16OFp16H512HVPerCta256PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 160.54 KB
fmhaSm103aKernel_QkvFp16OFp16H512HVPerCta256PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 154.55 KB
fmhaSm103aKernel_QkvFp16OFp16H512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 163.52 KB
fmhaSm103aKernel_QkvFp16OFp16H512HVPerCta256PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 157.50 KB
fmhaSm103aKernel_QkvFp16OFp16H512HVPerCta256PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 140.44 KB
fmhaSm103aKernel_QkvFp16OFp16H512HVPerCta256PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 129.76 KB
fmhaSm103aKernel_QkvFp16OFp16H512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 143.41 KB
fmhaSm103aKernel_QkvFp16OFp16H512HVPerCta256PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 132.70 KB
fmhaSm103aKernel_QkvFp16OFp16H512HVPerCta256PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 126.13 KB
fmhaSm103aKernel_QkvFp16OFp16H512HVPerCta256PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 117.88 KB
fmhaSm103aKernel_QkvFp16OFp16H512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 128.98 KB
fmhaSm103aKernel_QkvFp16OFp16H512HVPerCta256PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 120.82 KB
fmhaSm103aKernel_QkvFp16OFp16H64PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 102.08 KB
fmhaSm103aKernel_QkvFp16OFp16H64PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 90.28 KB
fmhaSm103aKernel_QkvFp16OFp16H64PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 109.20 KB
fmhaSm103aKernel_QkvFp16OFp16H64PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 97.28 KB
fmhaSm103aKernel_QkvFp16OFp16H64PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 102.48 KB
fmhaSm103aKernel_QkvFp16OFp16H64PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 91.12 KB
fmhaSm103aKernel_QkvFp16OFp16H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 109.55 KB
fmhaSm103aKernel_QkvFp16OFp16H64PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 98.20 KB
fmhaSm103aKernel_QkvFp16OFp16H64PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 112.74 KB
fmhaSm103aKernel_QkvFp16OFp16H64PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 101.38 KB
fmhaSm103aKernel_QkvFp16OFp16H64PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 119.65 KB
fmhaSm103aKernel_QkvFp16OFp16H64PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 108.27 KB
fmhaSm103aKernel_QkvFp16OFp16H64PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 117.98 KB
fmhaSm103aKernel_QkvFp16OFp16H64PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 106.11 KB
fmhaSm103aKernel_QkvFp16OFp16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 125.60 KB
fmhaSm103aKernel_QkvFp16OFp16H64PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 113.62 KB
fmhaSm103aKernel_QkvFp16OFp16H64PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 115.62 KB
fmhaSm103aKernel_QkvFp16OFp16H64PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 103.84 KB
fmhaSm103aKernel_QkvFp16OFp16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 122.49 KB
fmhaSm103aKernel_QkvFp16OFp16H64PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 110.84 KB
fmhaSm103aKernel_QkvFp16OFp16H64PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 114.08 KB
fmhaSm103aKernel_QkvFp16OFp16H64PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 102.43 KB
fmhaSm103aKernel_QkvFp16OFp16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 120.95 KB
fmhaSm103aKernel_QkvFp16OFp16H64PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 109.31 KB
fmhaSm103aKernel_QkvFp16OFp16H64PagedKvCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 115.51 KB
fmhaSm103aKernel_QkvFp16OFp16H64PagedKvCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 103.46 KB
fmhaSm103aKernel_QkvFp16OFp16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 122.38 KB
fmhaSm103aKernel_QkvFp16OFp16H64PagedKvCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 110.47 KB
fmhaSm103aKernel_QkvFp16OFp16H64PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 118.09 KB
fmhaSm103aKernel_QkvFp16OFp16H64PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 106.96 KB
fmhaSm103aKernel_QkvFp16OFp16H64PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 126.02 KB
fmhaSm103aKernel_QkvFp16OFp16H64PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 114.47 KB
fmhaSm103aKernel_QkvFp16OFp16H64PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 116.14 KB
fmhaSm103aKernel_QkvFp16OFp16H64PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 104.68 KB
fmhaSm103aKernel_QkvFp16OFp16H64PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 123.03 KB
fmhaSm103aKernel_QkvFp16OFp16H64PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 111.69 KB
fmhaSm103aKernel_QkvFp16OFp16H64PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 114.60 KB
fmhaSm103aKernel_QkvFp16OFp16H64PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 103.16 KB
fmhaSm103aKernel_QkvFp16OFp16H64PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 121.49 KB
fmhaSm103aKernel_QkvFp16OFp16H64PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 110.29 KB
fmhaSm103aKernel_QkvFp16OFp16H64PagedKvDensePagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 116.15 KB
fmhaSm103aKernel_QkvFp16OFp16H64PagedKvDensePagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 104.34 KB
fmhaSm103aKernel_QkvFp16OFp16H64PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 122.80 KB
fmhaSm103aKernel_QkvFp16OFp16H64PagedKvDensePagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 111.48 KB
fmhaSm103aKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 129.52 KB
fmhaSm103aKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 117.21 KB
fmhaSm103aKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 136.44 KB
fmhaSm103aKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 124.22 KB
fmhaSm103aKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 127.04 KB
fmhaSm103aKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 114.94 KB
fmhaSm103aKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 133.32 KB
fmhaSm103aKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 121.43 KB
fmhaSm103aKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 125.50 KB
fmhaSm103aKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 113.53 KB
fmhaSm103aKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 131.78 KB
fmhaSm103aKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 119.91 KB
fmhaSm103aKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 128.29 KB
fmhaSm103aKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 115.66 KB
fmhaSm103aKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 134.36 KB
fmhaSm103aKernel_QkvFp16OFp16H64PagedKvSlidingOrChunkedCausalPagedVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 122.16 KB
fmhaSm103aKernel_QkvFp16OFp16H80PackedQkvCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 105.37 KB
fmhaSm103aKernel_QkvFp16OFp16H80PackedQkvCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 89.47 KB
fmhaSm103aKernel_QkvFp16OFp16H80PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 111.89 KB
fmhaSm103aKernel_QkvFp16OFp16H80PackedQkvCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 96.48 KB
fmhaSm103aKernel_QkvFp16OFp16H80PackedQkvDenseVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 105.77 KB
fmhaSm103aKernel_QkvFp16OFp16H80PackedQkvDenseVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 90.18 KB
fmhaSm103aKernel_QkvFp16OFp16H80PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 112.97 KB
fmhaSm103aKernel_QkvFp16OFp16H80PackedQkvDenseVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 97.27 KB
fmhaSm103aKernel_QkvFp16OFp16H80PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 116.03 KB
fmhaSm103aKernel_QkvFp16OFp16H80PackedQkvSlidingOrChunkedCausalVarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 100.48 KB
fmhaSm103aKernel_QkvFp16OFp16H80PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 122.47 KB
fmhaSm103aKernel_QkvFp16OFp16H80PackedQkvSlidingOrChunkedCausalVarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 107.51 KB
fmhaSm103aKernel_QkvFp16OFp16H80PagedKvCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 127.05 KB
fmhaSm103aKernel_QkvFp16OFp16H80PagedKvCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 111.31 KB
fmhaSm103aKernel_QkvFp16OFp16H80PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 133.46 KB
fmhaSm103aKernel_QkvFp16OFp16H80PagedKvCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 118.20 KB
fmhaSm103aKernel_QkvFp16OFp16H80PagedKvCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 121.38 KB
fmhaSm103aKernel_QkvFp16OFp16H80PagedKvCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 105.93 KB
fmhaSm103aKernel_QkvFp16OFp16H80PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 127.66 KB
fmhaSm103aKernel_QkvFp16OFp16H80PagedKvCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 112.95 KB
fmhaSm103aKernel_QkvFp16OFp16H80PagedKvCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 118.14 KB
fmhaSm103aKernel_QkvFp16OFp16H80PagedKvCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 102.32 KB
fmhaSm103aKernel_QkvFp16OFp16H80PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 124.41 KB
fmhaSm103aKernel_QkvFp16OFp16H80PagedKvCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 109.34 KB
fmhaSm103aKernel_QkvFp16OFp16H80PagedKvDenseP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 127.44 KB
fmhaSm103aKernel_QkvFp16OFp16H80PagedKvDenseP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 111.77 KB
fmhaSm103aKernel_QkvFp16OFp16H80PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 134.98 KB
fmhaSm103aKernel_QkvFp16OFp16H80PagedKvDenseP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 118.91 KB
fmhaSm103aKernel_QkvFp16OFp16H80PagedKvDenseP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 120.93 KB
fmhaSm103aKernel_QkvFp16OFp16H80PagedKvDenseP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 106.12 KB
fmhaSm103aKernel_QkvFp16OFp16H80PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 128.47 KB
fmhaSm103aKernel_QkvFp16OFp16H80PagedKvDenseP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 113.27 KB
fmhaSm103aKernel_QkvFp16OFp16H80PagedKvDenseP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 117.81 KB
fmhaSm103aKernel_QkvFp16OFp16H80PagedKvDenseP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 103.16 KB
fmhaSm103aKernel_QkvFp16OFp16H80PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 125.36 KB
fmhaSm103aKernel_QkvFp16OFp16H80PagedKvDenseP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 110.18 KB
fmhaSm103aKernel_QkvFp16OFp16H80PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 138.59 KB
fmhaSm103aKernel_QkvFp16OFp16H80PagedKvSlidingOrChunkedCausalP16VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 122.12 KB
fmhaSm103aKernel_QkvFp16OFp16H80PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 144.90 KB
fmhaSm103aKernel_QkvFp16OFp16H80PagedKvSlidingOrChunkedCausalP16VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 128.63 KB
fmhaSm103aKernel_QkvFp16OFp16H80PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 132.80 KB
fmhaSm103aKernel_QkvFp16OFp16H80PagedKvSlidingOrChunkedCausalP32VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 116.74 KB
fmhaSm103aKernel_QkvFp16OFp16H80PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 139.21 KB
fmhaSm103aKernel_QkvFp16OFp16H80PagedKvSlidingOrChunkedCausalP32VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 123.25 KB
fmhaSm103aKernel_QkvFp16OFp16H80PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 129.55 KB
fmhaSm103aKernel_QkvFp16OFp16H80PagedKvSlidingOrChunkedCausalP64VarSeqQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 113.26 KB
fmhaSm103aKernel_QkvFp16OFp16H80PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128PersistentContext.cubin 05-Jun-2026 22:35 135.85 KB
fmhaSm103aKernel_QkvFp16OFp16H80PagedKvSlidingOrChunkedCausalP64VarSeqSkipsSoftmaxQ128Kv128StaticContext.cubin 05-Jun-2026 22:35 119.76 KB
Artifactory/7.117.25 Server