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

Remove VLLM_ASCEND_ENABLE_DENSE_OPTIMIZE (#5272)

Browse Source 
`VLLM_ASCEND_ENABLE_DENSE_OPTIMIZE` is only used together with
`VLLM_ASCEND_ENABLE_PREFETCH_MLP` which is useless totally. This PR
remove it.
- vLLM version: release/v0.13.0
- vLLM main:
ad32e3e19c

Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>
This commit is contained in:
wangxiyuan
2025-12-25 11:09:56 +08:00
committed by GitHub
parent 13cd6362c6
commit 2ae0bad96d
8 changed files with 7 additions and 21 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
tests/e2e/nightly/features/test_qwen3_32b_int8_a3_feature_stack3.py
View File

@@ -72,7 +72,6 @@ async def test_models(model: str, tp_size: int) -> None:
"OMP_PROC_BIND": "false",
"VLLM_ASCEND_ENABLE_TOPK_OPTIMIZE": "1",
"VLLM_ASCEND_ENABLE_FLASHCOMM": "1",
"VLLM_ASCEND_ENABLE_DENSE_OPTIMIZE": "1",
"VLLM_ASCEND_ENABLE_PREFETCH_MLP": "1"
}
server_args = [

1
tests/e2e/nightly/models/test_qwen3_32b_int8.py
View File

@@ -81,7 +81,6 @@ async def test_models(model: str, mode: str, tp_size: int) -> None:
port = get_open_port()
env_dict = {
"TASK_QUEUE_ENABLE": "1",
"VLLM_ASCEND_ENABLE_DENSE_OPTIMIZE": "1",
"HCCL_OP_EXPANSION_MODE": "AIV",
"VLLM_ASCEND_ENABLE_FLASHCOMM": "1",
"VLLM_ASCEND_ENABLE_PREFETCH_MLP": "1"