tune triton-fp8w8a8g64 and triton-fp8w8a8g128#1215
Conversation
Summary of ChangesHello, 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 introduces a set of new, pre-tuned kernel configurations for Triton's FP8 weight-only quantization feature. These configurations are designed to optimize the performance of scaled matrix multiplication operations on NVIDIA GeForce RTX 5090 GPUs running Triton 3.5.1, covering different activation quantization group sizes (64 and 128) and various matrix dimensions. The goal is to enhance the efficiency and speed of these specific computational workloads. Highlights
🧠 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. Changelog
Using Gemini Code AssistThe 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
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 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
|
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
There was a problem hiding this comment.
Code Review
This pull request adds autotuned Triton kernel configurations for triton-fp8w8a8g64 and triton-fp8w8a8g128 on NVIDIA GeForce RTX 5090 GPUs. The changes consist of several new JSON configuration files. My review focuses on the format and maintainability of these files. I've suggested sorting the keys in the JSON files numerically to improve readability. This change should be applied to all new configuration files.
| { | ||
| "1": { | ||
| "BLOCK_K": 128, | ||
| "BLOCK_M": 8, | ||
| "BLOCK_N": 64, | ||
| "GROUP_M": 8, | ||
| "num_stages": 3, | ||
| "num_warps": 4 | ||
| }, | ||
| "100": { | ||
| "BLOCK_K": 64, | ||
| "BLOCK_M": 32, | ||
| "BLOCK_N": 128, | ||
| "GROUP_M": 8, | ||
| "num_stages": 3, | ||
| "num_warps": 8 | ||
| }, | ||
| "1024": { | ||
| "BLOCK_K": 128, | ||
| "BLOCK_M": 64, | ||
| "BLOCK_N": 128, | ||
| "GROUP_M": 8, | ||
| "num_stages": 3, | ||
| "num_warps": 4 | ||
| }, | ||
| "128": { | ||
| "BLOCK_K": 128, | ||
| "BLOCK_M": 64, | ||
| "BLOCK_N": 64, | ||
| "GROUP_M": 8, | ||
| "num_stages": 3, | ||
| "num_warps": 4 | ||
| }, | ||
| "16": { | ||
| "BLOCK_K": 128, | ||
| "BLOCK_M": 16, | ||
| "BLOCK_N": 64, | ||
| "GROUP_M": 8, | ||
| "num_stages": 3, | ||
| "num_warps": 4 | ||
| }, | ||
| "256": { | ||
| "BLOCK_K": 128, | ||
| "BLOCK_M": 64, | ||
| "BLOCK_N": 128, | ||
| "GROUP_M": 8, | ||
| "num_stages": 3, | ||
| "num_warps": 8 | ||
| }, | ||
| "32": { | ||
| "BLOCK_K": 128, | ||
| "BLOCK_M": 16, | ||
| "BLOCK_N": 64, | ||
| "GROUP_M": 8, | ||
| "num_stages": 4, | ||
| "num_warps": 4 | ||
| }, | ||
| "64": { | ||
| "BLOCK_K": 128, | ||
| "BLOCK_M": 32, | ||
| "BLOCK_N": 64, | ||
| "GROUP_M": 8, | ||
| "num_stages": 4, | ||
| "num_warps": 8 | ||
| }, | ||
| "8": { | ||
| "BLOCK_K": 128, | ||
| "BLOCK_M": 8, | ||
| "BLOCK_N": 64, | ||
| "GROUP_M": 8, | ||
| "num_stages": 4, | ||
| "num_warps": 4 | ||
| } | ||
| } No newline at end of file |
There was a problem hiding this comment.
The keys in this JSON file are sorted alphabetically as strings (e.g., "1", "100", "1024", "128", ...), which makes it difficult to read and find configurations for a specific M value. For better readability and maintainability, please sort the keys numerically. This comment applies to all new JSON configuration files in this pull request.
{
"1": {
"BLOCK_K": 128,
"BLOCK_M": 8,
"BLOCK_N": 64,
"GROUP_M": 8,
"num_stages": 3,
"num_warps": 4
},
"8": {
"BLOCK_K": 128,
"BLOCK_M": 8,
"BLOCK_N": 64,
"GROUP_M": 8,
"num_stages": 4,
"num_warps": 4
},
"16": {
"BLOCK_K": 128,
"BLOCK_M": 16,
"BLOCK_N": 64,
"GROUP_M": 8,
"num_stages": 3,
"num_warps": 4
},
"32": {
"BLOCK_K": 128,
"BLOCK_M": 16,
"BLOCK_N": 64,
"GROUP_M": 8,
"num_stages": 4,
"num_warps": 4
},
"64": {
"BLOCK_K": 128,
"BLOCK_M": 32,
"BLOCK_N": 64,
"GROUP_M": 8,
"num_stages": 4,
"num_warps": 8
},
"100": {
"BLOCK_K": 64,
"BLOCK_M": 32,
"BLOCK_N": 128,
"GROUP_M": 8,
"num_stages": 3,
"num_warps": 8
},
"128": {
"BLOCK_K": 128,
"BLOCK_M": 64,
"BLOCK_N": 64,
"GROUP_M": 8,
"num_stages": 3,
"num_warps": 4
},
"256": {
"BLOCK_K": 128,
"BLOCK_M": 64,
"BLOCK_N": 128,
"GROUP_M": 8,
"num_stages": 3,
"num_warps": 8
},
"1024": {
"BLOCK_K": 128,
"BLOCK_M": 64,
"BLOCK_N": 128,
"GROUP_M": 8,
"num_stages": 3,
"num_warps": 4
}
}
2 participants
No description provided.