[Bugfix] Don't attempt to use triton if no driver is active #19561
Conversation
|
👋 Hi! Thank you for contributing to the vLLM project. 💬 Join our developer Slack at https://slack.vllm.ai to discuss your PR in #pr-reviews, coordinate on features in #feat- channels, or join special interest groups in #sig- channels. Just a reminder: PRs would not trigger full CI run by default. Instead, it would only run Once the PR is approved and ready to go, your PR reviewer(s) can run CI to test the changes comprehensively before merging. To run CI, PR reviewers can either: Add 🚀 |
Signed-off-by: Konrad Zawora <kzawora@habana.ai>
There was a problem hiding this comment.
Summary of Changes
Hello @kzawora-intel, 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 fixes a bug where vLLM would attempt to initialize and use Triton kernels even on systems lacking a compatible GPU driver, leading to runtime errors. The change introduces a more robust check for Triton availability by verifying the presence of an active driver, ensuring Triton is only enabled when the underlying hardware supports it.
Highlights
- Triton Compatibility Check: Added a check to explicitly disable Triton usage within vLLM if the Triton package is installed but no active Triton driver is detected. This prevents errors on platforms without compatible GPU hardware.
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 is currently in preview and 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 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 configureGemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩
58b97a8 to
7230f14
Compare
There was a problem hiding this comment.
Code Review
This pull request correctly addresses an issue where vLLM could attempt to use an installed Triton package even if no compatible GPU driver is active, leading to runtime errors. The fix involves checking for the number of active Triton drivers and disabling Triton if a unique active driver is not found. My review suggests enhancing this new check by adding comprehensive error handling (using try-except blocks) to gracefully manage potential issues with the Triton installation itself, and by incorporating more specific logging to inform users why Triton might be disabled by this new logic. These changes would improve the robustness and diagnosability of the Triton usability check.
Signed-off-by: Konrad Zawora <kzawora@habana.ai>
Signed-off-by: Konrad Zawora <kzawora@habana.ai>
56dbd8a to
2d9be44
Compare
|
@simon-mo @mgoin @WoosukKwon @robertgshaw2-redhat , please help to review |
…ject#19561) Signed-off-by: minpeter <kali2005611@gmail.com>
…ject#19561) Signed-off-by: Yang Wang <elainewy@meta.com>
…ject#19561) Signed-off-by: avigny <47987522+avigny@users.noreply.github.com>
Platforms can have triton package installed by external packages (e.g. xgrammar), even if it's unsupported by them and not specified in vLLM requirements. This patch adds additional triton check and prevents non-GPU platforms from autotuning triton flash attention kernels when the package is installed, but incompatible. Example error solved by this PR (importing MLACommonImpl in out-of-tree attention backend, incorrectly attempting to use triton flash attention, because triton package was installed by external dependency):
CC @simon-mo @youkaichao @xuechendi