diff --git a/tools/library/scripts/generator.py b/tools/library/scripts/generator.py index 2da327a5..161110d9 100644 --- a/tools/library/scripts/generator.py +++ b/tools/library/scripts/generator.py @@ -4614,14 +4614,14 @@ def GenerateSM90_TensorOp_fp8_WGMMA_gemm(manifest, cuda_version): # some schedules disabled to save on library size if CudaToolkitVersionSatisfies(cuda_version, 12, 1): schedules = [ - [KernelScheduleType.ScheduleAuto, EpilogueScheduleType.ScheduleAuto], - [KernelScheduleType.TmaWarpSpecializedCooperative, EpilogueScheduleType.NoSmemWarpSpecialized], - # [KernelScheduleType.TmaWarpSpecialized, EpilogueScheduleType.NoSmemWarpSpecialized], - [KernelScheduleType.TmaWarpSpecializedPingpongFP8FastAccum, EpilogueScheduleType.NoSmemWarpSpecialized], + #[KernelScheduleType.ScheduleAuto, EpilogueScheduleType.ScheduleAuto], + [KernelScheduleType.TmaWarpSpecializedCooperative, EpilogueScheduleType.NoSmemWarpSpecialized], + [KernelScheduleType.TmaWarpSpecialized, EpilogueScheduleType.NoSmemWarpSpecialized], + [KernelScheduleType.TmaWarpSpecializedPingpongFP8FastAccum, EpilogueScheduleType.NoSmemWarpSpecialized], [KernelScheduleType.TmaWarpSpecializedCooperativeFP8FastAccum, EpilogueScheduleType.NoSmemWarpSpecialized], - # [KernelScheduleType.TmaWarpSpecializedFP8FastAccum, EpilogueScheduleType.NoSmemWarpSpecialized] + [KernelScheduleType.TmaWarpSpecializedFP8FastAccum, EpilogueScheduleType.NoSmemWarpSpecialized] ] - stream_k_schedules = [[KernelScheduleType.TmaWarpSpecializedCooperative, EpilogueScheduleType.NoSmemWarpSpecialized], + stream_k_schedules = [[KernelScheduleType.TmaWarpSpecializedCooperative, EpilogueScheduleType.NoSmemWarpSpecialized], [KernelScheduleType.TmaWarpSpecializedCooperativeFP8FastAccum, EpilogueScheduleType.NoSmemWarpSpecialized]] else: schedules = [