Add rocm720 base image

[akao-amd]

For both gfx942/gfx950 images, we add a new version of base image based on latest ROCm v7.1.x.

Motivation

As ROCm v7.1.x is around the corner now, integrating SGLang onto the new software stack is critical for internal/external early evaluation and adoption.

Modifications

1. Add a base image.
2. This specific image relies on python 3.12 while the build process of tilelang/llvm requires python 3.10. Add some pass to specifically turns off python 3.12 during llvm build.

Accuracy Tests

N/A

Benchmarking and Profiling

N/A

Checklist

N/A

[gemini-code-assist]

Summary of Changes

Hello @akao-amd, 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 integrates the upcoming ROCm v7.1.x software stack into the SGLang Docker build system. This update is critical for facilitating early evaluation and adoption of SGLang with the latest ROCm version. It primarily involves adding new base images for gfx942 and gfx950 GPUs and implementing conditional Python version switching to resolve build compatibility issues for components like TileLang and LLVM.

Highlights

• ROCm 7.1.x Base Images: New Docker base images for ROCm v7.1.x have been added for both gfx942 and gfx950 architectures, enabling SGLang integration with the latest ROCm software stack.
• Python Version Compatibility: Logic has been introduced to manage Python versions during the build process. Specifically, for the gfx950-rocm710 image, the default Python is temporarily switched from 3.12 to 3.10 for TileLang/LLVM builds and then reverted to 3.12 afterwards to ensure compatibility.
• SGLang Router Update: References and build paths for the Rust-based component have been updated from 'sgl-model-gateway' to 'sgl-router'.

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 support for ROCm 7.1.0 base images for both gfx942 and gfx950 architectures. A key part of this change is a workaround in the Dockerfile to temporarily switch from Python 3.12 to 3.10 during the TileLang/LLVM build process, as required, and then switch back. The changes look reasonable and address the goal of integrating with the new ROCm stack.

I have two suggestions for improvement:

1. Correct a minor typo in an example build command to ensure it uses the correct image tag.
2. Refactor the Python version switching logic to merge two RUN commands into one. This will improve the Docker image by reducing the number of layers and keeping related logic together.

[akao-amd]

Change Log:

• Update to ROCm 7.2.
• For gfx950 only.

[akao-amd]

Closing in favor of #17799