Skip to content

GitLab

Unverified Commit 917fdae5 authored by Boyuan Feng's avatar Boyuan Feng Committed by GitHub
Browse files

[Log] Skip piecewise cudagraph warn when using full cudagraph (#30657) 


Signed-off-by: default avatarBoyuan Feng <boyuan@meta.com>
parent e2ed2388
Show whitespace changes
Inline Side-by-side
Showing with 7 additions and 3 deletions
vllm/config/compilation.py
View file @ 917fdae5
...@@ -932,6 +932,10 @@ class CompilationConfig: ...@@ -932,6 +932,10 @@ class CompilationConfig:
self.splitting_ops = list(self._attention_ops) self.splitting_ops = list(self._attention_ops)
added_default_splitting_ops = True added_default_splitting_ops = True
elif len(self.splitting_ops) == 0: elif len(self.splitting_ops) == 0:
if (
self.cudagraph_mode == CUDAGraphMode.PIECEWISE
or self.cudagraph_mode == CUDAGraphMode.FULL_AND_PIECEWISE
):
logger.warning_once( logger.warning_once(
"Using piecewise compilation with empty splitting_ops" "Using piecewise compilation with empty splitting_ops"
) )
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment