Skip to content

Conversation

lilinsiman
Copy link
Contributor

@lilinsiman lilinsiman commented Sep 25, 2025

What this PR does / why we need it?

Solve the problem that ACLgraph piecewise cannot be automatically enabled

Does this PR introduce any user-facing change?

no

How was this patch tested?

ut

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 introduces a fix to ensure ACLgraph piecewise is enabled correctly by downgrading CUDAGraphMode.FULL_AND_PIECEWISE to PIECEWISE for vLLM versions other than v0.10.2. While this addresses the immediate issue, the version check is brittle and may cause issues with future vLLM updates. I've added a comment with a suggestion to make the version check more robust for long-term maintainability.

@wangxiyuan wangxiyuan added the ready read for review label Sep 25, 2025
@wangxiyuan
Copy link
Collaborator

fixed here #3182

@wangxiyuan wangxiyuan closed this Sep 26, 2025
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
module:core ready read for review
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants