chore: bump sgl-kernel 0.3.0 with torch 2.8.0 #8718
Conversation
There was a problem hiding this comment.
Summary of Changes
Hello @zhyncs, 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 made a routine update to the sgl-kernel library, advancing its version to 0.3.0. This change primarily involves upgrading the underlying PyTorch dependency from 2.7.1 to 2.8.0, which necessitated corresponding adjustments in the build scripts to ensure proper installation of the new PyTorch version. This update keeps the project aligned with recent PyTorch releases.
Highlights
- PyTorch Dependency Update: I have updated the
torchdependency across allpyproject.tomlfiles (including CPU and ROCm specific configurations) from version2.7.1to2.8.0. This ensures compatibility with the latest PyTorch features and bug fixes. - SGL-Kernel Version Bump: The
sgl-kernelproject version has been incremented from0.2.9to0.3.0in all relevantpyproject.tomlfiles and theversion.pyfile. This reflects the new release incorporating the updated PyTorch dependency. - Build Script Adjustments: The
build.shscript has been modified to correctly installtorch==2.8.0using specific test index URLs for different CUDA versions (cu128 and cu126). This ensures that the build process fetches the correct PyTorch binaries for the updated 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 sgl-kernel to version 0.3.0 and bumps the torch dependency to version 2.8.0. The changes are applied consistently across the pyproject.toml files and the version file. However, there is a potential issue in the build.sh script where a CUDA version is hardcoded, which could break the build for different CUDA environments. I've left a specific comment with a suggestion to address this.
| else | ||
| DOCKER_IMAGE="${BUILDER_NAME}:cuda${CUDA_VERSION}" | ||
| TORCH_INSTALL="pip install --no-cache-dir torch==2.7.1" | ||
| TORCH_INSTALL="pip install --no-cache-dir torch==2.8.0 --index-url https://download.pytorch.org/whl/test/cu126" |
There was a problem hiding this comment.
The PyTorch wheel index URL in the else block hardcodes the CUDA version to 12.6. This will likely cause the build to fail for any CUDA version other than 12.8. To ensure the script works correctly for various CUDA versions, you should construct the URL dynamically using the ${CUDA_VERSION} variable.
| TORCH_INSTALL="pip install --no-cache-dir torch==2.8.0 --index-url https://download.pytorch.org/whl/test/cu126" | |
| TORCH_INSTALL="pip install --no-cache-dir torch==2.8.0 --index-url https://download.pytorch.org/whl/test/cu${CUDA_VERSION//.}" |
|
torch 2.8.0 will be released at 08/06 |
Motivation
Modifications
Accuracy Test
Benchmark & Profiling
Checklist