Update/Fix: ComfyUI-Zluda

[NeuralFault]

To provide more stable memory management and to bring SM's launch handling up to date with the source project's own launch script. In the past few months the main ComfyUI project added in changes that lead ComfyUI-Zluda to have certain launch arguments by default to prevent negative effects of the changes from ComfyUI in regards to Zluda and AMD GPUs.

Also updated the default cross attention method to be in-line with ComfyUI-Zluda project's default selection as using PyTorch cross attention was causing Cudnn frontend errors.

Changes:

• Updated the default value for the "Cross Attention Method" launch option to use --use-quad-cross-attention.
• Added four new boolean launch options:

  • "Disable Async Offload" (default: true, adds --disable-async-offload)

  • "Disable Pinned Memory" (default: true, adds --disable-pinned-memory)

    • Both of these are default in ComfyUI, disabled by default in ComfyUI-Zluda.

  • "Disable Smart Memory" (default: false, adds --disable-smart-memory)

  • "Disable Model/Node Caching" (default: false, adds --cache-none)

    • Added to launch options selections but disabled by default based upon recommendation from @patientx . For use in certain OOM situations beneficial for use with AMD GPUs.

[gemini-code-assist]

Summary of Changes

Hello @NeuralFault, 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 focuses on improving the stability and compatibility of ComfyUI-Zluda by updating its launch configurations. The changes address recent modifications in the main ComfyUI project, ensuring better memory management and preventing potential issues like Cudnn frontend errors, especially for users with AMD GPUs. The goal is to provide a more robust and up-to-date integration experience.

Highlights

• Cross Attention Method Default Update: The default value for the "Cross Attention Method" launch option has been updated to --use-quad-cross-attention. This change aims to prevent Cudnn frontend errors that were occurring with the previous default.
• Introduction of New Memory Management Launch Options: Four new boolean launch options have been added to enhance memory management and align with recent ComfyUI project changes. These include "Disable Async Offload" (default: true), "Disable Pinned Memory" (default: true), "Disable Smart Memory" (default: false), and "Disable Model/Node Caching" (default: false). These options are particularly beneficial for optimizing performance and stability on AMD GPUs.

🧠 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 updates the launch options for ComfyUI-Zluda to enhance memory management and align with upstream changes. The modifications correctly adjust the default "Cross Attention Method" and introduce four new boolean launch options. The implementation is solid, but I've provided one suggestion to refactor how the new options are added, which will improve code conciseness and maintainability.

[NeuralFault]

/gemini review

[gemini-code-assist]

Code Review

This pull request updates the launch options for ComfyUI-Zluda to enhance memory management and align with the upstream project. The changes correctly introduce new launch options and modify the default for cross-attention. My review includes suggestions to remove duplicated using statements and to simplify a LINQ expression for better readability and performance.

[gemini-code-assist]

This using statement is a duplicate. There are other duplicates on line 4 for System.Text.RegularExpressions and line 6 for Injectio.Attributes. Please remove all redundant using statements to improve code clarity.

[mohnjiles]

Thank you!!