xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2025-12-30 11:05:16 +08:00
Code Issues Packages Projects Releases Wiki Activity

[Bug] Fix DeepGEMM Env Control (#23591)

Browse Source 
Signed-off-by: yewentao256 <zhyanwentao@126.com>
This commit is contained in:
Wentao Ye 2025-08-25 21:41:21 -04:00 committed by GitHub
parent ae067888d6
commit 56dcf4e7e9
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 4 additions and 7 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

11
vllm/utils/deep_gemm.py
View File

@ -27,7 +27,7 @@ def is_deep_gemm_supported() -> bool:
is_supported_arch = current_platform.is_cuda() and (
current_platform.is_device_capability(90)
or current_platform.is_device_capability(100))
return has_deep_gemm() and is_supported_arch
return envs.VLLM_USE_DEEP_GEMM and has_deep_gemm() and is_supported_arch
@functools.cache
@ -35,12 +35,9 @@ def is_blackwell_deep_gemm_e8m0_used() -> bool:
"""Return ``True`` if vLLM is configured to use DeepGEMM "
"E8M0 scale on a Blackwell-class GPU.
"""
if not (envs.VLLM_USE_DEEP_GEMM):
logger.debug_once("DeepGEMM E8M0 disabled: VLLM_USE_DEEP_GEMM=0.")
return False
if not has_deep_gemm():
logger.debug_once("DeepGEMM E8M0 disabled: DeepGEMM backend missing.")
if not is_deep_gemm_supported():
logger.debug_once(
"DeepGEMM E8M0 disabled: DeepGEMM not supported on this system.")
return False
if not envs.VLLM_USE_DEEP_GEMM_E8M0: