frozenleaves/pytorch
1
0
Fork 0
mirror of https://github.com/pytorch/pytorch.git synced 2025-10-20 21:14:14 +08:00
Code Issues Packages Projects Releases Wiki Activity

[CUDA][float8][TF32] Disable tf32 for vs. emulated rowwise comparison (#162387)

Browse Source 
Pull Request resolved: https://github.com/pytorch/pytorch/pull/162387
Approved by: https://github.com/Skylion007
This commit is contained in:
Eddie Yan
2025-09-09 17:04:06 +00:00
committed by PyTorch MergeBot
parent f03d635dc6
commit 5eb35d2ab8
1 changed files with 2 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
test/test_matmul_cuda.py
View File

@ -32,6 +32,7 @@ from torch.testing._internal.common_cuda import (
PLATFORM_SUPPORTS_MX_GEMM,
PLATFORM_SUPPORTS_MXFP8_GROUPED_GEMM,
IS_SM90,
with_tf32_off,
)
from torch.testing._internal.common_device_type import (
dtypes,
@ -1510,6 +1511,7 @@ class TestFP8Matmul(TestCase):
@unittest.skipIf(not PLATFORM_SUPPORTS_FP8 or IS_WINDOWS, f8_msg)
@unittest.skipIf(not SM89OrLater, "rowwise implementation is currently sm89-sm100 specific")
@parametrize("base_dtype", [torch.bfloat16, torch.float32])
@with_tf32_off
def test_scaled_mm_vs_emulated_row_wise(self, base_dtype):
# Fp32 out_dtype is only supported by cuBLAS, which however only started
# shipping row-wise kernels in CUDA 12.9, and only for sm90+.