Skip to content

[Bugfix] Lower gpt-oss max cudagraph size to 992 to be compatible with FA3 #25508

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
merged 1 commit into from
Sep 23, 2025
+5 −5
Merged

[Bugfix] Lower gpt-oss max cudagraph size to 992 to be compatible with FA3 #25508

Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
10 changes: 5 additions & 5 deletions vllm/model_executor/models/config.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -266,24 +266,24 @@ def verify_and_update_config(vllm_config: "VllmConfig") -> None:
if structured_outputs_config.reasoning_parser == "":
structured_outputs_config.reasoning_parser = "openai_gptoss"

# Increase the max capture size from 512 to 1024 for performance.
# Increase the max capture size from 512 to 992 for performance.
# NOTE(woosuk): This will increase the number of CUDA graphs
# from 67 to 83.
# from 67 to 81.
scheduler_config = vllm_config.scheduler_config
if len(scheduler_config.cuda_graph_sizes) == 1:
max_capture_size = scheduler_config.cuda_graph_sizes[0]
# FIXME(woosuk): When using full cuda graph with FA3, the max
# supported size is 992.
if max_capture_size < 1024:
if max_capture_size < 992:
cuda_graph_sizes = [1, 2, 4]
# Step size 8 for small batch sizes
cuda_graph_sizes += [i for i in range(8, 256, 8)]
# Step size 16 for larger batch sizes
cuda_graph_sizes += [i for i in range(256, 1025, 16)]
cuda_graph_sizes += [i for i in range(256, 993, 16)]
scheduler_config.cuda_graph_sizes = cuda_graph_sizes
logger.info(
"Overriding max cuda graph capture size to "
"%d for performance.", 1024)
"%d for performance.", 992)


class MambaModelConfig(VerifyAndUpdateConfig):
Expand Down
Loading