Revert "[Bugfix] default set cuda_graph_sizes to max_num_seqs for v1 engine" #20128
Conversation
…engine (…" This reverts commit 9f0608f.
|
👋 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 🚀 |
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
There was a problem hiding this comment.
Summary of Changes
Hello @mgoin, 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 serves to revert a recent change that modified the default behavior of cuda_graph_sizes within the SchedulerConfig. The intent is to restore the previous default value of [512] for cuda_graph_sizes and remove the associated dynamic configuration logic, allowing for further discussion on the optimal default behavior.
Highlights
- Revert Previous Change: This pull request reverts a previous change (#20062) that modified the default behavior of
cuda_graph_sizes. - Default
cuda_graph_sizesRestored: The default value forcuda_graph_sizesinSchedulerConfighas been restored to[512]. - Removed Dynamic Configuration Logic: The logic within
SchedulerConfig.__post_init__that dynamically setcuda_graph_sizesto[self.max_num_seqs]when not explicitly provided has been removed.
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 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 reverts a previous bug fix and changes the default value of cuda_graph_sizes in vllm/config.py to [512]. The documentation and initialization logic have been updated accordingly. The removal of the conditional assignment in __post_init__ simplifies the code.
| cuda_graph_sizes: list[int] = field(default_factory=lambda: [512]) | ||
| """Cuda graph capture sizes, default is 512. | ||
| 1. if one value is provided, then the capture list would follow the | ||
| pattern: [1, 2, 4] + [i for i in range(8, cuda_graph_sizes + 1, 8)] | ||
| 3. more than one value (e.g. 1 2 128) is provided, then the capture list | ||
| 2. more than one value (e.g. 1 2 128) is provided, then the capture list | ||
| will follow the provided list.""" |
There was a problem hiding this comment.
The documentation should be updated to reflect the new default value and the behavior when no value is provided.
…engine" (vllm-project#20128) Signed-off-by: Will Eaton <[email protected]>
…engine" (vllm-project#20128) Signed-off-by: avigny <[email protected]>
Reverts #20062 as we would like to engage in further discussion on the "right" default behavior
See the thread in the original PR for initial discussion #20062 (comment)