Fix wrong kernel selection for int32/int64 indices

[hnyls2002]

Bug introduced in #16273

CI breaks.

[hnyls2002]

https://github.com/sgl-project/sglang/actions/runs/20897156909/job/60042429830?pr=16820

[gemini-code-assist]

Summary of Changes

Hello @hnyls2002, 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 fully reverts the previously introduced feature 'Support JIT set kv cache' (#16273). The primary motivation for this revert is to address continuous integration (CI) failures that were occurring after the feature's introduction. The changes systematically remove all components related to the JIT KV-cache, including its custom CUDA kernel, Python bindings, benchmarking, testing, and its integration within the memory management system, aiming to restore stability to the CI pipeline.

Highlights

• JIT KV-Cache Feature Removal: The core Just-In-Time (JIT) KV-cache kernel (kvcache.cuh) and its Python wrapper (kvcache.py) have been completely removed, effectively undoing the 'Support JIT set kv cache' feature.
• Memory Pool Integration Reverted: The memory_pool.py file has been updated to remove all integration points for the JIT KV-cache, including the _set_kv_buffer_impl function, related imports, and direct assignment logic for K/V cache updates.
• Associated Files Deleted: Dedicated benchmark (bench_store_cache.py), test files (test_store_cache.py), and a utility file (utils.py) specifically created for the JIT KV-cache feature have been deleted.
• Norm Module Cleanup: The norm.py module and its corresponding C++ kernel (qknorm.cuh) have been adjusted. The qknorm.cuh file was removed, and norm.py now points to a more general norm.cuh and uses a renamed internal function _jit_norm_module.

🧠 New Feature in Public Preview: You can now enable Memory 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 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. ↩

[hnyls2002]

/rerun-stage stage-c-test-large-4-gpu

[gemini-code-assist]

Code Review

This pull request reverts the feature 'Support JIT set kv cache' due to CI breakages. The changes correctly remove the new JIT kernel for setting KV cache, including its implementation, benchmarks, and tests. The code in memory_pool.py that utilized this feature has been reverted to its previous implementation. The revert appears to be mostly complete. However, I've found a minor issue in a benchmark file that seems to be an incomplete part of the revert, which I've commented on.

[github-actions]

✅ Triggered stage-c-test-large-4-gpu to run independently (skipping dependencies).

[github-actions]

🔗 View workflow run

[DarkSharpness]

/rerun-stage stage-c-test-large-4-gpu

[github-actions]

✅ Triggered stage-c-test-large-4-gpu to run independently (skipping dependencies).

[github-actions]

🔗 View workflow run

[DarkSharpness]

/tag-and-rerun-ci

[DarkSharpness]

/rerun-failed-ci