Updates for 3.2 release (#1065)

This commit is contained in:
ANIKET SHIVAM
2023-08-25 23:05:46 -04:00
committed by GitHub
parent 27de343535
commit a88c41cf8d
20 changed files with 904 additions and 257 deletions
+2
View File
@@ -4218,6 +4218,8 @@ def GenerateSM90_TensorOp_16b_WGMMA_gemm(manifest, cuda_version):
layout[2][1] = 8
CreateGemmUniversal3xOperator(manifest, layouts, tile_descriptions, data_type_mixed, schedules)
CreateGemmUniversal3xOperator(manifest, layouts, tile_descriptions, data_type_mixed, stream_k_schedules, tile_schedulers=[TileSchedulerType.StreamK])
# persistent kernels with TMA epilogues
if CudaToolkitVersionSatisfies(cuda_version, 12, 1):
CreateGemmUniversal3xOperator(manifest, layouts, tile_descriptions, data_type_mixed,