Skip to content

Conversation

@MengqingCao
Copy link
Collaborator

@MengqingCao MengqingCao commented Oct 30, 2025

What this PR does / why we need it?

Fix 2 breaks of aclgraph with MTP:

  1. deepseekmtp in vllm 0.11.0 does not support aclgraph and lack the support_torch_compile decorator
  2. There is a d2h synchornization in the original forward of mtp predictor. The fix pr in vllm [MTP] Refactor mtp predictor to avoid d2h operation vllm#27643

As we'll fix it in vllm main, this fix pr is only needed in branch v0.11.0-dev

The profling shows that MTP replays in aclgraph now:
a7d7f04155df4ed454b7eb20a92b2e2a

How was this patch tested?

Signed-off-by: MengqingCao <cmq0113@163.com>
@github-actions
Copy link

👋 Hi! Thank you for contributing to the vLLM Ascend project. The following points will speed up your PR merge:‌‌

  • A PR should do only one thing, smaller PRs enable faster reviews.
  • Every PR should include unit tests and end-to-end tests ‌to ensure it works and is not broken by other future PRs.
  • Write the commit message by fulfilling the PR description to help reviewer and future developers understand.

If CI fails, you can run linting and testing checks locally according Contributing and Testing.

Copy link
Contributor

@gemini-code-assist gemini-code-assist bot left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Code Review

This pull request addresses two issues related to aclgraph support with MTP in vllm 0.11.0. It introduces a patch to avoid d2h synchronization in the MTP predictor's forward pass and adds the support_torch_compile decorator for deepseekmtp. The changes are targeted for the v0.11.0-dev branch.

Signed-off-by: MengqingCao <cmq0113@163.com>
Signed-off-by: MengqingCao <cmq0113@163.com>
Signed-off-by: MengqingCao <cmq0113@163.com>
@MengqingCao
Copy link
Collaborator Author

@wangxiyuan @zzzzwwjj This pr is ready for merge, plz take a look, thanks!

@wangxiyuan wangxiyuan merged commit 7cc6208 into vllm-project:v0.11.0-dev Nov 3, 2025
16 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants