leavelet/cutlass
1
0
Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases Wiki Activity

v4.4 update. (#2979)

Browse Source
This commit is contained in:
Junkai-Wu
2026-01-25 00:46:17 +08:00
committed by GitHub
parent 2fafefb7b9
commit 9fba3195f9
293 changed files with 46344 additions and 2996 deletions
Download Patch File Download Diff File Expand all files Collapse all files

16
python/cutlass_library/library.py
View File

@@ -562,6 +562,12 @@ class KernelScheduleType(enum.Enum):
SparseNvf4TmaWarpSpecialized2SmSm100 = enum_auto()
SparseMxf8f6f4TmaWarpSpecialized1SmSm100 = enum_auto()
SparseMxf8f6f4TmaWarpSpecialized2SmSm100 = enum_auto()
InterleavedComplexTF32TmaWarpSpecialized1SmSm100 = enum_auto()
InterleavedComplexTF32TmaWarpSpecialized2SmSm100 = enum_auto()
TmaWarpSpecialized1SmFastFP32Sm100 = enum_auto()
TmaWarpSpecialized2SmFastFP32Sm100 = enum_auto()
# FP4 Ultra
MxNvf4UltraTmaWarpSpecialized1SmVs16Sm103 = enum_auto()
MxNvf4UltraTmaWarpSpecialized2SmVs16Sm103 = enum_auto()
@@ -679,7 +685,10 @@ KernelScheduleTag = {
KernelScheduleType.MxNvf4UltraTmaWarpSpecialized2SmVs16Sm103DisablePrefetch: 'cutlass::gemm::KernelTmaWarpSpecialized2SmBlockScaledMxNvf4UltraVs16Sm103DisablePrefetch',
KernelScheduleType.MxNvf4UltraTmaWarpSpecialized1SmVs32Sm103DisablePrefetch: 'cutlass::gemm::KernelTmaWarpSpecialized1SmBlockScaledMxNvf4UltraVs32Sm103DisablePrefetch',
KernelScheduleType.MxNvf4UltraTmaWarpSpecialized2SmVs32Sm103DisablePrefetch: 'cutlass::gemm::KernelTmaWarpSpecialized2SmBlockScaledMxNvf4UltraVs32Sm103DisablePrefetch',
KernelScheduleType.InterleavedComplexTF32TmaWarpSpecialized1SmSm100: 'cutlass::gemm::KernelTmaWarpSpecialized1SmInterleavedComplexTF32Sm100',
KernelScheduleType.InterleavedComplexTF32TmaWarpSpecialized2SmSm100: 'cutlass::gemm::KernelTmaWarpSpecialized2SmInterleavedComplexTF32Sm100',
KernelScheduleType.TmaWarpSpecialized1SmFastFP32Sm100: 'cutlass::gemm::KernelTmaWarpSpecialized1SmFastFP32Sm100',
KernelScheduleType.TmaWarpSpecialized2SmFastFP32Sm100: 'cutlass::gemm::KernelTmaWarpSpecialized2SmFastFP32Sm100',
KernelScheduleType.PtrArrayTmaWarpSpecializedCooperative: 'cutlass::gemm::KernelPtrArrayTmaWarpSpecializedCooperative',
KernelScheduleType.PtrArrayTmaWarpSpecializedCooperativeFP8FastAccum: 'cutlass::gemm::KernelPtrArrayTmaWarpSpecializedCooperativeFP8FastAccum',
KernelScheduleType.PtrArrayTmaWarpSpecializedPingpong: 'cutlass::gemm::KernelPtrArrayTmaWarpSpecializedPingpong',
@@ -799,7 +808,10 @@ KernelScheduleSuffixes = {
KernelScheduleType.MxNvf4UltraTmaWarpSpecialized2SmVs16Sm103TmaPrefetch: '_o_vs16_ultra_2sm_tmapf',
KernelScheduleType.MxNvf4UltraTmaWarpSpecialized1SmVs32Sm103TmaPrefetch: '_o_vs32_ultra_1sm_tmapf',
KernelScheduleType.MxNvf4UltraTmaWarpSpecialized2SmVs32Sm103TmaPrefetch: '_o_vs32_ultra_2sm_tmapf',
KernelScheduleType.InterleavedComplexTF32TmaWarpSpecialized1SmSm100: '_1sm',
KernelScheduleType.InterleavedComplexTF32TmaWarpSpecialized2SmSm100: '_2sm',
KernelScheduleType.TmaWarpSpecialized1SmFastFP32Sm100: '_FastF32_1sm',
KernelScheduleType.TmaWarpSpecialized2SmFastFP32Sm100: '_FastF32_2sm',
KernelScheduleType.PtrArrayTmaWarpSpecializedCooperative: '_warpspecialized_cooperative',
KernelScheduleType.PtrArrayTmaWarpSpecializedCooperativeFP8FastAccum: '_warpspecialized_cooperative_fp8_fastaccum',
KernelScheduleType.PtrArrayTmaWarpSpecializedPingpong: '_warpspecialized_pingpong',