[Improvement] KVCache: optimize coll communication

[DwyaneShi]

Pull Request Description

Use all reduce with tensor instead of all gather objects, in which, based on our profiling results, the frequent conversion of python objects to tensors would cause performance issue.

Related Issues

Resolves: #[Insert issue number(s)]

Important: Before submitting, please complete the description above and review the checklist below.

---

Contribution Guidelines (Expand for Details)

We appreciate your contribution to aibrix! To ensure a smooth review process and maintain high code quality, please adhere to the following guidelines:

Pull Request Title Format

Your PR title should start with one of these prefixes to indicate the nature of the change:

• [Bug]: Corrections to existing functionality
• [CI]: Changes to build process or CI pipeline
• [Docs]: Updates or additions to documentation
• [API]: Modifications to aibrix's API or interface
• [CLI]: Changes or additions to the Command Line Interface
• [Misc]: For changes not covered above (use sparingly)

Note: For changes spanning multiple categories, use multiple prefixes in order of importance.

Submission Checklist

• PR title includes appropriate prefix(es)
• Changes are clearly explained in the PR description
• New and existing tests pass successfully
• Code adheres to project style and best practices
• Documentation updated to reflect changes (if applicable)
• Thorough testing completed, no regressions introduced

By submitting this PR, you confirm that you've read these guidelines and your changes align with the project's contribution standards.

[gemini-code-assist]

Summary of Changes

Hello @DwyaneShi, 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 significantly optimizes the KVCache's collective communication by transitioning from dist.all_gather_object to dist.all_reduce with tensors. This change addresses performance bottlenecks identified in profiling, which were caused by frequent conversions of Python objects to tensors. The update streamlines distributed operations, leading to more efficient KVCache management.

Highlights

• Performance Optimization: Replaced dist.all_gather_object with dist.all_reduce in cache_manager.py for improved performance in collective communication, specifically within the _group_aware_acquire_impl and get methods.
• Data Structure Change: Introduced a dedicated torch.Tensor (_coll_tensor) to facilitate efficient status and block count propagation across distributed processes using all_reduce.
• Logic Refinement: Refactored the logic for handling Status codes (OK, NOT_FOUND, ERROR) and block counts to be encoded as integer values within the _coll_tensor for all_reduce operations.
• Test Updates: Updated test cases in test_group_aware_cache_manager.py to reflect the new TokenListView usage and adjusted the number of processes for stress tests to 4.

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 in your pull request via creating an issue comment (i.e. comment on the pull request page) using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands.

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 issue comments and review comments.
Help	/gemini help	Displays a list of available commands.

Customization

To customize 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 counter productive. 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.

You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension.

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. ↩

[gemini-code-assist]

Code Review

This pull request aims to optimize collective communication by using all_reduce with tensors instead of all_gather with Python objects. The changes are in the right direction for performance. However, I've found a critical bug in the implementation within GroupAwareKVCacheManager.get where units (tokens vs. blocks) are mixed up, which could lead to incorrect behavior. I've also identified a high-severity issue regarding device placement for the collective communication tensor and a medium-severity issue about using magic numbers, which affects code maintainability. Please address these points to ensure the correctness and quality of the changes.