xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-01-29 08:57:13 +08:00
Code Issues Packages Projects Releases Wiki Activity

[BugFix] Fix DeepGEMM over-allocating workspace (#28254)

Browse Source 
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
This commit is contained in:
Lucas Wilkinson 2025-11-10 17:01:17 -05:00 committed by GitHub
parent bf6a3d0ff5
commit 6dec9f6109
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 1 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
vllm/model_executor/layers/fused_moe/deep_gemm_moe.py
View File

@ -215,7 +215,7 @@ class DeepGemmExperts(mk.FusedMoEPermuteExpertsUnpermute):
)
assert M_sum % block_m == 0
workspace1 = (M_sum, max(N, K))
workspace1 = (M_sum, N)
workspace2 = (M_sum, max(N // 2, K))
output = (M, K)
return (workspace1, workspace2, output)