Skip to content

[Fix] the bug in the trtllm-gen heurisitcf for MLA kernels. #6284

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Merged
merged 1 commit into from
Jul 24, 2025
Merged

[Fix] the bug in the trtllm-gen heurisitcf for MLA kernels. #6284

merged 1 commit into from
Jul 24, 2025

Conversation

PerkzZheng
Copy link
Collaborator

@PerkzZheng PerkzZheng commented Jul 23, 2025 •
edited by coderabbitai bot
Loading

this fixes the bug of still selecting low-latency (swapsMmaAb) MLA kernels when batch size is quite large under the high-throughput case (attention DP is used).

The same fix has been merged into flashinfer. see flashinfer-ai/flashinfer#1307.

Summary by CodeRabbit

Summary by CodeRabbit

  • New Features

    • Enhanced kernel selection to improve performance for certain sequence lengths and head configurations during MLA generation.

  • Refactor

    • Refined conditions for choosing low-latency kernels, optimizing processing efficiency under specific workload constraints.

@coderabbitai coderabbitai
Copy link
Contributor

coderabbitai bot commented Jul 23, 2025 •
edited
Loading

📝 Walkthrough

Walkthrough

The update modifies the kernel selection logic for MLA generation in the FMHA kernels. It replaces the integer-returning method for computing sequence length per CTA for KV with a boolean method that determines whether to use a specific kernel based on sequence length and CTA count. The kernel selection logic is updated accordingly.

Changes

File(s) Change Summary
cpp/tensorrt_llm/kernels/trtllmGenKernels/fmha/fmhaKernels.h Replaced computeSeqLenPerCtaKv method with useSwapsMmaAbMlaGenKernel, updating logic to return a boolean and modifying kernel selection accordingly.

Estimated code review effort

🎯 1 (Trivial) | ⏱️ ~5 minutes

Suggested reviewers

  • lfr-0531
  • yweng0828

Poem

In the land of kernels, swaps now decide,
Boolean logic leads where integers once lied.
CTAs and heads, with sequence in tow,
Guide the right kernel, efficiency to show.
A hop and a skip, this rabbit’s delight—
Code made more clever, and kernels more bright!
🐇✨

Note

⚡️ Unit Test Generation is now available in beta!

Learn more here, or try it out under "Finishing Touches" below.

📜 Recent review details

Configuration used: .coderabbit.yaml
Review profile: CHILL
Plan: Pro

📥 Commits

Reviewing files that changed from the base of the PR and between 200f01b and c8ef0c5.

📒 Files selected for processing (1)
  • cpp/tensorrt_llm/kernels/trtllmGenKernels/fmha/fmhaKernels.h (3 hunks)
🚧 Files skipped from review as they are similar to previous changes (1)
  • cpp/tensorrt_llm/kernels/trtllmGenKernels/fmha/fmhaKernels.h
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (1)
  • GitHub Check: Pre-commit Check
✨ Finishing Touches
  • 📝 Generate Docstrings
🧪 Generate unit tests
  • Create PR with unit tests
  • Post copyable unit tests in a comment
🪧 Tips

Chat

There are 3 ways to chat with CodeRabbit:

  • Review comments: Directly reply to a review comment made by CodeRabbit. Example:
    • I pushed a fix in commit <commit_id>, please review it.
    • Explain this complex logic.
    • Open a follow-up GitHub issue for this discussion.
  • Files and specific lines of code (under the "Files changed" tab): Tag @coderabbitai in a new review comment at the desired location with your query. Examples:
    • @coderabbitai explain this code block.
    • @coderabbitai modularize this function.
  • PR comments: Tag @coderabbitai in a new PR comment to ask questions about the PR branch. For the best results, please provide a very specific query, as very limited context is provided in this mode. Examples:
    • @coderabbitai gather interesting stats about this repository and render them as a table. Additionally, render a pie chart showing the language distribution in the codebase.
    • @coderabbitai read src/utils.ts and explain its main purpose.
    • @coderabbitai read the files in the src/scheduler package and generate a class diagram using mermaid and a README in the markdown format.
    • @coderabbitai help me debug CodeRabbit configuration file.

Support

Need help? Create a ticket on our support page for assistance with any issues or questions.

Note: Be mindful of the bot's finite context window. It's strongly recommended to break down tasks such as reading entire modules into smaller chunks. For a focused discussion, use review comments to chat about specific files and their changes, instead of using the PR comments.

CodeRabbit Commands (Invoked using PR comments)

  • @coderabbitai pause to pause the reviews on a PR.
  • @coderabbitai resume to resume the paused reviews.
  • @coderabbitai review to trigger an incremental review. This is useful when automatic reviews are disabled for the repository.
  • @coderabbitai full review to do a full review from scratch and review all the files again.
  • @coderabbitai summary to regenerate the summary of the PR.
  • @coderabbitai generate docstrings to generate docstrings for this PR.
  • @coderabbitai generate sequence diagram to generate a sequence diagram of the changes in this PR.
  • @coderabbitai generate unit tests to generate unit tests for this PR.
  • @coderabbitai resolve resolve all the CodeRabbit review comments.
  • @coderabbitai configuration to show the current CodeRabbit configuration for the repository.
  • @coderabbitai help to get help.

Other keywords and placeholders

  • Add @coderabbitai ignore anywhere in the PR description to prevent this PR from being reviewed.
  • Add @coderabbitai summary to generate the high-level summary at a specific location in the PR description.
  • Add @coderabbitai or @coderabbitai title anywhere in the PR title to generate the title automatically.

Documentation and Community

  • Visit our Documentation for detailed information on how to use CodeRabbit.
  • Join our Discord Community to get help, request features, and share feedback.
  • Follow us on X/Twitter for updates and announcements.

@PerkzZheng
Copy link
Collaborator Author

/bot run

@tensorrt-cicd
Copy link
Collaborator

PR_Github #12677 [ run ] triggered by Bot

@PerkzZheng PerkzZheng requested review from Tom-Zheng and yuxianq July 23, 2025 09:05
Tom-Zheng
Tom-Zheng approved these changes Jul 23, 2025
View reviewed changes
Copy link
Collaborator

@Tom-Zheng Tom-Zheng left a comment

Choose a reason for hiding this comment

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

LGTM

@tensorrt-cicd
Copy link
Collaborator

PR_Github #12677 [ run ] completed with state SUCCESS
/LLM/main/L0_MergeRequest_PR pipeline #9428 completed with status: 'FAILURE'

@PerkzZheng PerkzZheng force-pushed the user/perkzz/fix-trtllm-gen-heurisitc branch from 3ba1097 to 200f01b Compare July 24, 2025 02:21
@PerkzZheng
Copy link
Collaborator Author

/bot run

@coderabbitai coderabbitai bot requested a review from lfr-0531 July 24, 2025 02:22
@tensorrt-cicd
Copy link
Collaborator

PR_Github #12775 [ run ] triggered by Bot

@tensorrt-cicd
Copy link
Collaborator

PR_Github #12775 [ run ] completed with state FAILURE

@PerkzZheng PerkzZheng force-pushed the user/perkzz/fix-trtllm-gen-heurisitc branch from 200f01b to c8ef0c5 Compare July 24, 2025 11:28
@PerkzZheng
Copy link
Collaborator Author

/bot run

@tensorrt-cicd
Copy link
Collaborator

PR_Github #12845 [ run ] triggered by Bot

@tensorrt-cicd
Copy link
Collaborator

PR_Github #12845 [ run ] completed with state SUCCESS
/LLM/main/L0_MergeRequest_PR pipeline #9575 completed with status: 'SUCCESS'

@PerkzZheng PerkzZheng merged commit 706f421 into NVIDIA:main Jul 24, 2025
3 checks passed
NVShreyas pushed a commit to NVShreyas/TensorRT-LLM that referenced this pull request Jul 28, 2025
@PerkzZheng @NVShreyas
[Fix] the bug in the trtllm-gen heurisitcf for MLA kernels. (NVIDIA#6284
2a3ff38 
)

Signed-off-by: Perkz Zheng <[email protected]>
Signed-off-by: Shreyas Misra <[email protected]>
Ransiki pushed a commit to Ransiki/TensorRT-LLM that referenced this pull request Jul 29, 2025
@PerkzZheng @Ransiki
[Fix] the bug in the trtllm-gen heurisitcf for MLA kernels. (NVIDIA#6284
a7e5dd3 
)

Signed-off-by: Perkz Zheng <[email protected]>
Signed-off-by: Ransiki Zhang <[email protected]>
lancelly pushed a commit to lancelly/TensorRT-LLM that referenced this pull request Aug 6, 2025
@PerkzZheng @lancelly
[Fix] the bug in the trtllm-gen heurisitcf for MLA kernels. (NVIDIA#6284
0da7f2a 
)

Signed-off-by: Perkz Zheng <[email protected]>
Signed-off-by: Lanyu Liao <[email protected]>
@coderabbitai coderabbitai bot mentioned this pull request Aug 14, 2025
Merged
@coderabbitai coderabbitai bot mentioned this pull request Aug 24, 2025
Merged
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@yuxianq yuxianq yuxianq approved these changes

@lfr-0531 lfr-0531 Awaiting requested review from lfr-0531

+1 more reviewer

@Tom-Zheng Tom-Zheng Tom-Zheng approved these changes

Reviewers whose approvals may not affect merge requirements
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
4 participants
@PerkzZheng @tensorrt-cicd @Tom-Zheng @yuxianq