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

[bugfix] Fix warning bug: model config is None. (#3238)

Browse Source 
Cleanup wrong warning log error: model config is None

- vLLM version: v0.10.2
- vLLM main:
https://github.com/vllm-project/vllm/commit/releases/v0.11.0

---------

Signed-off-by: weijinqian_v1 <weijinqian@huawei.com>
Co-authored-by: weijinqian_v1 <weijinqian@huawei.com>
This commit is contained in:
weijinqian0
2025-09-29 09:44:49 +08:00
committed by GitHub
parent 15b8aff582
commit 8870966031
2 changed files with 6 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
vllm_ascend/ascend_forward_context.py
View File

@@ -106,7 +106,7 @@ def set_ascend_forward_context(
# Currently, it is an empirical value. In normal scenarios, if the concurrency exceeds this threshold, # Currently, it is an empirical value. In normal scenarios, if the concurrency exceeds this threshold,
# the performance benefits can be maximized. Conversely, if the concurrency is below the threshold, # the performance benefits can be maximized. Conversely, if the concurrency is below the threshold,
# the performance may degrade due to the switching of communication methods. # the performance may degrade due to the switching of communication methods.
sp_enabled = enable_sp() and \ sp_enabled = enable_sp(vllm_config) and \
tp_world_size > 1 and \ tp_world_size > 1 and \
num_tokens is not None and num_tokens > 1000 num_tokens is not None and num_tokens > 1000

9
vllm_ascend/utils.py
View File

@@ -597,11 +597,12 @@ def dense_optim_enable() -> bool:
return envs_ascend.VLLM_ASCEND_ENABLE_DENSE_OPTIMIZE return envs_ascend.VLLM_ASCEND_ENABLE_DENSE_OPTIMIZE
def enable_sp() -> bool: def enable_sp(vllm_config=None) -> bool:
from vllm.config import get_cached_compilation_config if vllm_config is None:
from vllm.config import get_current_vllm_config
vllm_config = get_current_vllm_config()
return ( return (
get_cached_compilation_config().pass_config.enable_sequence_parallelism vllm_config.compilation_config.pass_config.enable_sequence_parallelism
or envs_ascend.VLLM_ASCEND_ENABLE_FLASHCOMM) or envs_ascend.VLLM_ASCEND_ENABLE_FLASHCOMM)