Skip to content

[Perf] Increase default max splits for FA3 full cudagraphs #25495

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
vllm-bot merged 2 commits into from
Sep 23, 2025
Merged

[Perf] Increase default max splits for FA3 full cudagraphs #25495

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
cc20802
increase default splits
LucasWilkinson Sep 23, 2025
1ae089a
change to 32
LucasWilkinson Sep 23, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 2 additions & 2 deletions vllm/envs.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -119,7 +119,7 @@
VLLM_SERVER_DEV_MODE: bool = False
VLLM_V1_OUTPUT_PROC_CHUNK_SIZE: int = 128
VLLM_MLA_DISABLE: bool = False
VLLM_FLASH_ATTN_MAX_NUM_SPLITS_FOR_CUDA_GRAPH: int = 16
VLLM_FLASH_ATTN_MAX_NUM_SPLITS_FOR_CUDA_GRAPH: int = 32
VLLM_RAY_PER_WORKER_GPUS: float = 1.0
VLLM_RAY_BUNDLE_INDICES: str = ""
VLLM_CUDART_SO_PATH: Optional[str] = None
Expand Down Expand Up @@ -1012,7 +1012,7 @@ def get_vllm_port() -> Optional[int]:
# max number splits for cuda graph decode
"VLLM_FLASH_ATTN_MAX_NUM_SPLITS_FOR_CUDA_GRAPH":
lambda: int(os.getenv("VLLM_FLASH_ATTN_MAX_NUM_SPLITS_FOR_CUDA_GRAPH",
"16")),
"32")),

# Number of GPUs per worker in Ray, if it is set to be a fraction,
# it allows ray to schedule multiple actors on a single GPU,
Expand Down