[NVIDIA] Add Jetson Orin/Thor/Spark/GB300 Codegen

[johnnynunez]

Orin is 8.7
Thor is 11.0
Spark is 12.1
GB300 10.3

[gemini-code-assist]

Summary of Changes

Hello @johnnynunez, 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 support for NVIDIA Jetson Orin devices into the build system. It updates the CUDA compilation flags to include the compute capability specific to Orin, ensuring that kernels can be properly generated and executed on this hardware.

Highlights

• New Hardware Support: Added codegen support for NVIDIA Jetson Orin, specifically targeting its compute capability 8.7 (SM 8.7).
• Build Configuration Update: Modified sgl-kernel/CMakeLists.txt to include the necessary -gencode=arch=compute_87,code=sm_87 flag for CUDA compilation, enabling the kernel to be built for Orin devices.

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

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 CUDA codegen support for the NVIDIA Jetson Orin (compute capability 8.7) by appending the -gencode=arch=compute_87,code=sm_87 flag to SGL_FLASH_KERNEL_CUDA_FLAGS within the sgl-kernel/CMakeLists.txt file. This change is specifically for when FlashAttention3 (SGL_KERNEL_ENABLE_FA3) is enabled and Ampere architectures (SGL_FLASH_KERNEL_CUDA_ARCHS_AMPERE) are being targeted.

The modification is straightforward and appears correct for enabling Orin support in this context. The added line is consistent with the existing flags for other Ampere architectures like sm_80 and sm_86.

For improved clarity and future maintainability, consider expanding the pull request description to be more explicit about the change. For example: "This PR adds CUDA codegen support for NVIDIA Jetson Orin (compute capability 8.7) to the SGL FlashAttention3 kernel by including the sm_87 architecture in the CMake build flags."

[faradawn]

This PR adds CUDA codegen support for Jetson. Added support for -gencode=arch=compute_87,code=sm_87. Looks good to me.

[johnnynunez]

@zhyncs can you merge it?

[johnnynunez]

@faradawn @zhyncs this add orin support for sbsa and upcoming new devices with cuda 13

[johnnynunez]

@FlamingoPg i hope that it will be merged soon. New era of robotics start on Monday

[FlamingoPg]

LGTM, wait for CI

[johnnynunez]

@FlamingoPg FYI: Thor launch is today

[johnnynunez]

@faradawn closing in favor of #9495 could you merge it?