xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-04-01 02:47:04 +08:00
Code Issues Packages Projects Releases Wiki Activity

[Misc] Avoid cuda graph log when sizes still match (#18202)

Browse Source 
Signed-off-by: NickLucche <nlucches@redhat.com>
This commit is contained in:
Nicolò Lucchesi 2025-05-15 18:59:38 +02:00 committed by GitHub
parent 92540529c0
commit e3f3aee6f4
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 6 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

11
vllm/config.py
View File

@ -3950,11 +3950,12 @@ class CompilationConfig:
self.cudagraph_capture_sizes = cudagraph_capture_sizes
else:
# de-duplicate the sizes provided by the config
self.cudagraph_capture_sizes = list(
set(self.cudagraph_capture_sizes))
logger.info(("cudagraph sizes specified by model runner"
" %s is overridden by config %s"),
cudagraph_capture_sizes, self.cudagraph_capture_sizes)
dedup_sizes = list(set(self.cudagraph_capture_sizes))
if len(dedup_sizes) < len(self.cudagraph_capture_sizes):
logger.info(("cudagraph sizes specified by model runner"
" %s is overridden by config %s"),
cudagraph_capture_sizes, dedup_sizes)
self.cudagraph_capture_sizes = dedup_sizes
computed_compile_sizes = []
if self.compile_sizes is not None: