[Doc] fix: Update README for cu126 sgl-kernel compile problem #8665
Conversation
There was a problem hiding this comment.
Summary of Changes
Hello @Hongbosherlock, 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!
I've updated the sgl-kernel README to include essential guidance for compiling FlashAttention on Hopper GPU architectures. This change addresses a known compilation issue by providing specific CUDA toolchain version recommendations and a practical script to ensure the correct ptxas version is used, aiming to resolve issues like the one reported in #8097.
Highlights
- Documentation Update: I've added a new section to the
sgl-kernel/README.mdspecifically for compiling FlashAttention on Hopper GPU Architecture (sm90). - FlashAttention Compilation Guidance: The new documentation provides explicit recommendations for
nvcc(version 12.6) andptxas(version 12.8) versions, and includes a bash script to facilitate updatingptxasto the required version.
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 or fill out our survey to provide feedback.
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
-
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. ↩
There was a problem hiding this comment.
Code Review
This pull request updates the sgl-kernel README with instructions on how to compile FlashAttention on Hopper GPUs, which involves updating ptxas. The added documentation is helpful, but the provided shell script has a few issues related to permissions, cleanup, and robustness.
|
Fix Lint @Hongbosherlock |
|
Over all LGTM. Can we add some eye-catching symbols to remind users to pay attention to cu126 logic? Like |
all done |
|
Only change the doc, will not wait for PR |
|
Will fix on dockerfile later XD |
Motivation
A temporary fix for: #8097
Modifications
Update README with cuda 12.6 compilation tips on Hopper;
FlashAttention on hopper needs
nvcc 12.6andptxas 12.8refer to: https://github.com/Dao-AILab/flash-attention/blob/v2.8.2/hopper/setup.py#L379
Dockerfile may require future modification to upgrade
ptxas.Accuracy Test
Benchmark & Profiling
Checklist