Add AMD GPU support for DeepSeek-V3.1 (MI300X/MI325X/MI355X + AITER)

[GoldenGrapeGentleman]

Summary

This PR adds AMD GPU support for DeepSeek-V3.1 on MI300X/MI325X/MI355X GPUs.

Changes

• Step 1: uv-based vLLM ROCm installation guide
• Step 2: vLLM server launch command with AITER and AITER_MOE enabled
• Step 3: Benchmark script

Hardware Tested

Hardware	Status
8x AMD MI300X + AITER	✅ Verified
8x AMD MI355X + AITER	✅ Verified

Related

Closes the AMD GPU support gap originally started in #158.

[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 significantly enhances the DeepSeek-V3.1 documentation by integrating full support for AMD GPUs, specifically the MI300X, MI325X, and MI355X series. It provides users with a clear, step-by-step guide for setting up the vLLM environment, launching the server with optimized configurations for AMD hardware, and performing benchmarks, thereby broadening the accessibility and performance capabilities of DeepSeek-V3.1 on AMD platforms.

Highlights

• AMD GPU Support: Added comprehensive support for DeepSeek-V3.1 models on AMD MI300X/MI325X/MI355X GPUs.
• vLLM Installation Guide: Provided a uv-based installation guide for vLLM with ROCm backend, including specific environment requirements.
• vLLM Server Configuration: Included commands to launch the vLLM server with AITER and AITER_MOE enabled, along with tensor parallelism and expert parallelism settings.
• Benchmarking Script: Added a script to benchmark the DeepSeek-V3.1 model running on the vLLM server.

🧠 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

• DeepSeek/DeepSeek-V3_1.md
  • Added a new top-level section titled "AMD GPU Support" to the document.
  • Included detailed instructions for installing vLLM with the ROCm backend using uv.
  • Provided a sample command to start the vLLM server, configuring it for AITER, AITER_MOE, tensor parallelism, and expert parallelism.
  • Added a benchmark script to test the performance of the DeepSeek-V3.1 model on the vLLM server.

Activity

• No specific activity (comments, reviews) has been recorded for this pull request yet.

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 adds valuable documentation for running DeepSeek-V3.1 on AMD GPUs. The instructions are clear and cover installation, server launch, and benchmarking. I've found one issue with the benchmark command which is incorrect and will prevent users from running it. I've suggested a fix for it. I also noticed that other documentation files in the repository use the same incorrect vllm bench serve command. It would be beneficial to correct these in a separate effort to maintain consistency and accuracy across all documentation. Otherwise, the changes in this PR look good.

[gemini-code-assist]

The vllm bench serve command appears to be incorrect. The correct command to run the vLLM benchmark is vllm benchmark. The arguments provided are valid for the vllm benchmark command.

Suggested change

	vllm bench serve \
	--model deepseek-ai/DeepSeek-V3.1 \
	--dataset-name random \
	--random-input-len 8192 \
	--random-output-len 1024 \
	--request-rate 10000 \
	--num-prompts 16 \
	--ignore-eos \
	--trust-remote-code
	vllm benchmark \
	--model deepseek-ai/DeepSeek-V3.1 \
	--dataset-name random \
	--random-input-len 8192 \
	--random-output-len 1024 \
	--request-rate 10000 \
	--num-prompts 16 \
	--ignore-eos \
	--trust-remote-code