biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[Refactor] Use DeepGEMM Col Major TMA Aligned Tensor (#25517)

Browse Source 
Signed-off-by: yewentao256 <zhyanwentao@126.com>
This commit is contained in:
Wentao Ye
2025-09-24 18:52:36 -04:00
committed by GitHub
parent 6160ba4151
commit 1f29141258
6 changed files with 34 additions and 78 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
tests/kernels/quantization/test_block_fp8.py
View File

@@ -11,11 +11,12 @@ from tests.kernels.quant_utils import (native_per_token_group_quant_fp8,
native_w8a8_block_matmul)
from vllm.config import VllmConfig
from vllm.model_executor.layers.quantization.utils.fp8_utils import (
cutlass_scaled_mm, get_col_major_tma_aligned_tensor,
per_token_group_quant_fp8, w8a8_triton_block_scaled_mm)
cutlass_scaled_mm, per_token_group_quant_fp8, w8a8_triton_block_scaled_mm)
from vllm.platforms import current_platform
from vllm.utils import has_deep_gemm
from vllm.utils.deep_gemm import fp8_gemm_nt, per_block_cast_to_fp8
from vllm.utils.deep_gemm import (fp8_gemm_nt,
get_col_major_tma_aligned_tensor,
per_block_cast_to_fp8)
if current_platform.get_device_capability() < (9, 0):
pytest.skip("FP8 Triton requires CUDA 9.0 or higher",