Add FP8 B200 best config to MiniMax-M2.5 recipe

[faradawn]

Summary

Adds FP8 serving config fo B200 GPUs to the MiniMax-M2.5 recipe

Reference: SemiAnalysisAI/InferenceX#757

[gemini-code-assist]

Summary of Changes

Hello, 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 enhances the MiniMax-M2.5 recipe by incorporating a new configuration tailored for FP8 serving on B200 GPUs. This update expands the hardware compatibility for deploying the model, providing optimized launch instructions for users leveraging B200 hardware and ensuring efficient model execution.

Highlights

• FP8 B200 Configuration Added: A new configuration has been introduced for launching the MiniMax-M2.5 model, specifically optimized for FP8 serving on 2x B200 GPUs. This configuration includes disabling FlashInfer MoE FP8 and DeepGEMM MoE.
• Expanded GPU Support: The documentation now explicitly lists 2x B200 GPUs as a supported option for launching the MiniMax-M2.5 model, alongside existing H200/H20 and A100/A800 configurations.

🧠 New Feature in Public Preview: You can now enable Memory to help Gemini Code Assist learn from your team's feedback. This makes future code reviews more consistent and personalized to your project's style. Click here to enable Memory in your admin console.

Changelog

• MiniMax/MiniMax-M2.md
  • Updated the list of supported GPUs to include 2x B200.
  • Added a new vllm serve command block specifically for 2x B200 GPUs, including environment variables to disable FlashInfer MoE FP8 and DeepGEMM MoE.

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 by creating a comment using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands on the current page.

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 pull request 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. If you're interested in giving your feedback about your experience with Gemini Code Assist for GitHub and other Google products, sign up here.

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

1. 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. ↩

[gemini-code-assist]

Code Review

This pull request effectively adds the FP8 B200 best configuration to the MiniMax-M2.5 recipe documentation. The new instructions are clear and provide valuable guidance for users leveraging B200 GPUs. The overall structure of the updated section is good, making it easy to follow the different deployment options.

[robertgshaw2-redhat]

instead of this, can we just update the default oracle in VLLM to select a different kernel for this case?

[robertgshaw2-redhat]

this is where the code is to update the priority order

• https://github.com/vllm-project/vllm/blob/main/vllm/model_executor/layers/fused_moe/oracle/fp8.py#L57

[faradawn]

Hi @robertgshaw2-redhat, thank you for the comment! Can you check if this PR addresses your comment: vllm-project/vllm#37056