Fix deepgemm masked grouped gemm jit compile #8679
Conversation
There was a problem hiding this comment.
Summary of Changes
Hello @ispobock, 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 addresses a critical bug in the JIT compilation process for masked grouped GEMM operations within the deepgemm library. The fix ensures that the correct number of groups is dynamically passed to the compilation configuration, resolving an issue where it was previously hardcoded and potentially leading to incorrect or suboptimal compilation.
Highlights
- Bug Fix: JIT Compilation: Corrected the
NUM_GROUPSparameter in the JIT compilation configuration for_compile_grouped_gemm_nt_f8f8bf16_masked_onefunction. Previously, it was hardcoded to1, but now it correctly uses the dynamicnum_groupsvariable passed to the function.
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 PR provides a critical fix for the DeepGEMM JIT compilation of masked grouped GEMM. By correctly using the num_groups parameter instead of a hardcoded value, it ensures the generated kernel is correct for configurations with more than one group.
| @@ -148,7 +148,7 @@ def _compile_grouped_gemm_nt_f8f8bf16_masked_one( | |||
| "NUM_MATH_THREADS_PER_GROUP": num_math_threads_per_group, | |||
| "N": n, | |||
| "K": k, | |||
| "NUM_GROUPS": 1, | |||
| "NUM_GROUPS": num_groups, | |||
There was a problem hiding this comment.
Using the num_groups parameter passed to the function is essential for correctness. Previously, NUM_GROUPS was hardcoded to 1, which would cause incorrect kernel compilation for masked grouped GEMM when num_groups was not 1.
Motivation
When using EP MoE with deepgemm, it will still re-compile masked grouped gemm kernels during serving time even though we precompiled all Ms in CUDA graph capture.
The num_groups of masked grouped gemm in generate config seems misconfigured.