[Diffusion] [NPU] Enable profiler on NPU

[Makcum888e]

Motivation

Enable NPU profiling in diffusion models

sglang generate --model_path path --prompt prompt --profile

Checklist

• Format your code according to the Format code with pre-commit.
• Add unit tests according to the Run and add unit tests.
• Update documentation according to Write documentations.
• Provide accuracy and speed benchmark results according to Test the accuracy and Benchmark the speed.
• Follow the SGLang code style guidance.

Review Process

1. Ping Merge Oncalls to start the PR flow. See the PR Merge Process.
2. Get approvals from CODEOWNERS and other reviewers.
3. Trigger CI tests with comments or contact authorized users to do so.
   • /tag-run-ci-label, /rerun-failed-ci, /tag-and-rerun-ci
4. After green CI and required approvals, ask Merge Oncalls to merge.

[gemini-code-assist]

Summary of Changes

Hello @Makcum888e, 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 extends the existing profiling infrastructure to seamlessly support Neural Processing Units (NPUs) within diffusion models. By integrating torch_npu specific profiling tools, it enables detailed performance analysis of NPU operations, ensuring that NPU-accelerated workloads can be effectively monitored and optimized.

Highlights

• NPU Profiling Integration: The pull request introduces support for NPU (Neural Processing Unit) profiling by conditionally importing and applying torch_npu profiler functionalities when running on an NPU platform.
• Profiler Activity Extension: The Profiler class is updated to include torch_npu.profiler.ProfilerActivity.NPU in its list of monitored activities, enabling NPU-specific performance data collection.
• NPU Trace Handling: For NPU platforms, the on_trace_ready callback for the profiler is configured to use torch_npu.profiler.tensorboard_trace_handler, facilitating NPU trace visualization in TensorBoard.
• NPU Synchronization: A torch.npu.synchronize() call is added to ensure all NPU operations are completed before the profiler stops, guaranteeing accurate performance data.
• Conditional Chrome Trace Export: The export_chrome_trace method is now conditionally executed, skipping the export for NPU platforms, likely due to torch_npu handling its own trace export mechanism.

🧠 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.

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 enables NPU profiling for diffusion models by integrating torch_npu's profiler capabilities. The changes correctly add NPU activity, synchronize before stopping the profiler, and use an NPU-specific trace handler. My review includes two main points. First, the use of an internal _apply_patches API from torch_npu could be brittle; I recommend checking for a public alternative or adding a comment about its necessity. Second, there's a logic issue in the trace export function for the NPU case, which will lead to confusing warnings. I've provided a code suggestion to correct this behavior and ensure accurate logging for the user.

[gemini-code-assist]

Warning

You have reached your daily quota limit. Please wait up to 24 hours and I will start processing your requests again!

[ping1jing2]

/tag-and-rerun-ci

[ping1jing2]

I merged it since all CIs passed and only 20 lines of code change. please let me know if there are any issues.

[mickqian]

cc @BBuf
Please wait for me or @BBuf's approval next time