[Performance][EPD] Reduce Python proxy serialization overhead - #56657
Merged
Merged
Conversation
Signed-off-by: Tianyu Guo <guoty@inferact.ai>
![claude[bot]](https://avatars.githubusercontent.com/in/1236702?s=60&v=4){kind=small}
Contributor
|
Documentation preview: https://vllm--56657.org.readthedocs.build/en/56657/ |
mergify
Bot
added
documentation
Collaborator
Author
|
/ci run |
|
✅ Triggered Buildkite CI #88591 for commit |
Collaborator
Author
|
/ci run |
|
✅ Triggered Buildkite CI #88622 for commit |
ywang96
approved these changes
Sep 13, 2026
Shreya-gaur
pushed a commit
to Shreya-gaur/vllm_private
that referenced
this pull request
Sep 14, 2026
…roject#56657) Signed-off-by: Tianyu Guo <guoty@inferact.ai>
jiaran-king
added a commit
to jiaran-king/vllm
that referenced
this pull request
Sep 15, 2026
Merge upstream main at 3192898. Include the EPD proxy serialization optimization (vllm-project#56657), JSON-array multimodal metadata (vllm-project#56090), and media processing options forwarding (vllm-project#56786). Git merged the overlapping Mooncake KV worker and tests without conflicts; no manual code adjustments were needed. Validation: 61 EC/cache/runner/proxy CPU tests and 171 Mooncake KV Store worker CPU tests passed. PR-file pre-commit, including mypy 3.10, passed. Co-authored-by: Codex <noreply@openai.com> Signed-off-by: Zhou ziheng <jiaranran2@gmail.com>
ItsRoy69
pushed a commit
to ItsRoy69/vllm
that referenced
this pull request
Sep 15, 2026
…roject#56657) Signed-off-by: Tianyu Guo <guoty@inferact.ai>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Purpose
Reduce serialization overhead in the Python EPD proxy:
msgspecdependency for JSON requests and E/P metadata.Checked #52409 and open proxy PRs; no duplicate found. Unlike #54140 (error propagation), this targets serialization overhead.
Validation
MUIRBench: Qwen3.5-35B-A3B, 4E + 4PD across two GB200 nodes, Mooncake RDMA, concurrency 96, V2 runner, CUDA graphs up to 16384 tokens (no eager). 200 warmup requests + 5 runs of the same first 1200 questions, caches retained; 6000/6000 formal requests succeeded.
Command used with the local frozen benchmark harness (five repetitions):
Historical reference only: nodes and main revisions differ, so the full gain cannot be attributed to this change. Prompt/completion token counts match for every paired request.
AI assistance was used for implementation and validation.