Skip to content

[ROCm] Allow full CUDA graphs with DCP for the a2a backend #53587

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

Closed
seungrokj wants to merge 1 commit into from
Closed

[ROCm] Allow full CUDA graphs with DCP for the a2a backend #53587

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: 8 additions & 2 deletions vllm/platforms/rocm.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -901,8 +901,14 @@ def check_and_update_config(cls, vllm_config: "VllmConfig") -> None:
parallel_config = vllm_config.parallel_config

if compilation_config.cudagraph_mode.has_full_cudagraphs():
# decode context parallel does not support full cudagraphs
if parallel_config.decode_context_parallel_size > 1:
# Decode context parallel does not support full cudagraphs, except
# for the all-to-all ("a2a") backend: its Triton pack/unpack kernels
# and private-pool send/recv buffers are capture-safe, matching CUDA
# (which does not downgrade this path).
if (
parallel_config.decode_context_parallel_size > 1
and parallel_config.dcp_comm_backend != "a2a"
):
logger.warning_once(
"Decode context parallel (DCP) is enabled, which is "
"incompatible with full CUDA graphs. "
Expand Down
Loading