NVIDIA/cutlass
1
0
Fork 0
mirror of https://github.com/NVIDIA/cutlass.git synced 2026-04-19 22:38:56 +00:00
Code Issues Packages Projects Releases Wiki Activity

v3.9 update (#2203)

Browse Source 
* v3.9 update

* voidD

---------

Co-authored-by: yuzhai <yuzhai@nvidia.com>
This commit is contained in:
Yujia Zhai
2025-04-02 12:11:18 -07:00
committed by GitHub
parent 62750a2b75
commit 6f4921858b
129 changed files with 7719 additions and 2036 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
python/cutlass_library/emit_kernel_listing.py
View File

@@ -282,6 +282,8 @@ def _computeFlopsPerByte(operation, m, n, k, batch_count=1, beta=0.0):
def emit_gemm_kernel_testlist(manifest, curr_build_dir, arch, mode
):
profiler_reference_computing = "--verification-providers=device --providers=cutlass"
# beta values for L0 and L1
# TODO: randomize beta values for wider coverage
beta_values = [0.5]

3
python/cutlass_library/generator.py
View File

@@ -10025,7 +10025,8 @@ def GenerateSM120_TensorOp_fp4_UMMA_gemm_with_block_scaled(manifest, cuda_versio
tile_sizes_cooperative = [
[128, 128, 128],
[128, 128, 256]
[128, 128, 256],
[256, 128, 128]
]
tile_sizes_pingpong = [