[v0.23.0][BugFix] Use grouped block sizes in KV transfer threads

[Pz1116]

What this PR does / why we need it?

Passes grouped_block_size to AscendStore's non-layerwise sending and asynchronous receiving threads instead of passing the first KV cache group's scalar block size.

Hybrid models such as DeepSeek V4 use different block sizes across cache groups. Reusing group 0's block size makes store/load mask indexing drop valid chunks from groups with smaller block sizes.

Does this PR introduce any user-facing change?

Yes. AscendStore correctly stores and asynchronously loads hybrid KV cache groups whose block sizes differ from group 0.

How was this patch tested?

• Added a unit test covering grouped block-size propagation to both transfer threads.

• Ran tests/ut/distributed/ascend_store/test_pool_worker.py: 66 passed.

• Ran all pre-commit hooks on the changed files.

• Validated the fix with the DeepSeek V4 W8A8 KV Pool workload that reproduced the issue.

• vLLM version: v0.23.0

• vLLM main: vllm-project/vllm@ee0da84

[gemini-code-assist]

Summary of Changes

Hello, I'm Gemini Code Assist¹! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!

This pull request addresses a bug in AscendStore where KV cache transfer threads were incorrectly using a single block size derived from the first group. By passing the complete grouped_block_size configuration, the system now correctly handles hybrid models that employ different block sizes across cache groups, preventing data loss during store and load operations.

Highlights

• Corrected Block Size Propagation: Updated AscendStore's non-layerwise sending and asynchronous receiving threads to utilize the full grouped_block_size list instead of defaulting to the first group's block size.
• Improved Hybrid Model Support: Ensured that hybrid models with varying block sizes across KV cache groups, such as DeepSeek V4, maintain correct store/load mask indexing.
• Enhanced Test Coverage: Added a new unit test to verify that the correct grouped block sizes are passed to the transfer threads.

New Features

🧠 You can now enable Memory (public preview) to help Gemini Code Assist learn from your team's feedback. This makes future code reviews more consistent and personalized to your project's style. Click here to enable Memory in your admin console.

Using Gemini Code Assist

The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.

Invoking Gemini

You can request assistance from Gemini at any point by creating a comment using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands on the current page.

Feature	Command	Description
Code Review	/gemini review	Performs a code review for the current pull request in its current state.
Pull Request Summary	/gemini summary	Provides a summary of the current pull request in its current state.
Comment	@gemini-code-assist	Responds in comments when explicitly tagged, both in pull request comments and review comments.
Help	/gemini help	Displays a list of available commands.

Customization

To customize the Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counterproductive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for GitHub and other Google products, sign up here.

Footnotes

1. Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩

[github-actions]

👋 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.

---

Tip

💡 Consider Linking a Related Issue or RFC

Your PR title contains the [BugFix] tag, indicating a bug fix or new feature.

Linking a related issue or RFC in the PR description is strongly encouraged — it gives reviewers helpful context and speeds up the review. You can use any of these keywords:

• Fixes #<issue_number>
• Closes #<issue_number>
• Resolves #<issue_number>
• Refs #<rfc_or_issue_number> (for RFCs)

🙏 Thanks for helping us keep the project well-organized!

[gemini-code-assist]

Code Review

Suggested PR Title:

[Ops][BugFix] Use grouped block sizes in KV transfer threads

Suggested PR Summary:

### What this PR does / why we need it?
This PR updates `_start_kv_transfer_threads` in `pool_worker.py` to pass `self.grouped_block_size` instead of `self.block_size` when initializing `KVCacheStoreSendingThread` and `KVCacheStoreRecvingThread`. This ensures that the transfer threads use the correct grouped block sizes.

### Does this PR introduce _any_ user-facing change?
No.

### How was this patch tested?
Added a unit test `test_transfer_threads_use_grouped_block_sizes` in `test_pool_worker.py` to verify that the transfer threads are initialized with the correct grouped block sizes.

I have no feedback to provide as there are no review comments.