Skip to content

[SKILL] fix(bench): Support model-specific DenoisingStage variants in…#21137

Merged
BBuf merged 1 commit intomainfrom
fix_skill_bench
Mar 23, 2026
Merged

[SKILL] fix(bench): Support model-specific DenoisingStage variants in…#21137
BBuf merged 1 commit intomainfrom
fix_skill_bench

Conversation

@BBuf
Copy link
Copy Markdown
Collaborator

@BBuf BBuf commented Mar 22, 2026

… latency parsing

fix(bench): Support model-specific DenoisingStage variants in latency parsing

Motivation

Extend the denoise latency extraction logic to accept model-specific variants of DenoisingStage (e.g., MOVADenoisingStage, HeliosChunkedDenoisingStage) in addition to the canonical name.

Modifications

Accuracy Tests

Benchmarking and Profiling

Checklist

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.

… latency parsing

fix(bench): Support model-specific DenoisingStage variants in latency parsing
@github-actions github-actions Bot added the diffusion SGLang Diffusion label Mar 22, 2026
@gemini-code-assist
Copy link
Copy Markdown
Contributor

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 benchmarking script by making its latency parsing more flexible. It updates the logic for identifying denoising stages to accommodate various model-specific implementations, thereby improving the accuracy and compatibility of latency measurements across different diffusion models.

Highlights

  • Latency Parsing Enhancement: Modified the bench_diffusion_denoise.py script to support model-specific variants of DenoisingStage (e.g., MOVADenoisingStage, HeliosChunkedDenoisingStage) when extracting denoise latency.
  • Broader Compatibility: The change ensures that the benchmarking script can accurately parse latency for a wider range of diffusion models that might use custom denoising stage names.

🧠 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.

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.

Copy link
Copy Markdown
Contributor

@gemini-code-assist gemini-code-assist Bot left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Code Review

This pull request extends the latency parsing logic to support model-specific variants of DenoisingStage, which is a good improvement for flexibility. I've added one suggestion to make the implementation more robust by handling cases where multiple matching stages might exist in the performance data.

Comment on lines 333 to 338
if (
step.get("name") == "DenoisingStage"
isinstance(step_name, str)
and "DenoisingStage" in step_name
and step.get("duration_ms") is not None
):
denoise_latency_s = float(step["duration_ms"]) / 1000.0
Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

medium

The current implementation iterates through all steps and will overwrite denoise_latency_s if multiple steps contain 'DenoisingStage' in their name. This could lead to incorrect latency reporting if, for example, there are multiple denoising-related stages. To ensure you capture the latency for the first relevant stage found and to improve efficiency, consider breaking out of the loop once a match is found.

Suggested change
if (
step.get("name") == "DenoisingStage"
isinstance(step_name, str)
and "DenoisingStage" in step_name
and step.get("duration_ms") is not None
):
denoise_latency_s = float(step["duration_ms"]) / 1000.0
if (
isinstance(step_name, str)
and "DenoisingStage" in step_name
and step.get("duration_ms") is not None
):
denoise_latency_s = float(step["duration_ms"]) / 1000.0
break

@BBuf
Copy link
Copy Markdown
Collaborator Author

BBuf commented Mar 22, 2026

/tag-and-rerun-ci

@BBuf BBuf merged commit a94d67d into main Mar 23, 2026
122 of 131 checks passed
@BBuf BBuf deleted the fix_skill_bench branch March 23, 2026 04:08
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

diffusion SGLang Diffusion run-ci

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant