EngineX/xc-llm-ascend
3
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Revert "GMM custom operator optimization in small batch scenarios (vllm-project#7100)" (#7557)

Browse Source 
### What this PR does / why we need it?
This reverts commit 42bcad7e9b. The commit
cause accuracy decrease of qwen3Next, 150 items of gsm8k, 98 -> 91.

- vLLM version: v0.18.0
- vLLM main:
6a9cceb219

Signed-off-by: Your Name <you@example.com>
Co-authored-by: Your Name <you@example.com>
This commit is contained in:
LeeWenquan
2026-03-24 14:24:44 +08:00
committed by GitHub
parent 83bd77c983
commit 475b4b0cea
3 changed files with 30 additions and 71 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
csrc/torch_binding.cpp
View File

@@ -691,7 +691,7 @@ std::vector<at::Tensor> moe_grouped_matmul(
y.emplace_back(y_0);
at::TensorList result = at::TensorList(y);
EXEC_NPU_CMD(aclnnMoeGroupedMatmul,
EXEC_NPU_CMD(aclnnMoeGroupedMatmulWeightNz,
x_list, weight_list, group_list, transpose_weight, result);
return y;