Skip to content

Conversation

@ckl117
Copy link
Collaborator

@ckl117 ckl117 commented Nov 16, 2025

Motivation

Before enabling PDL, there is a bubble between two kernels.
image

After enabling PDL, there is almost no bubbles between kernels.
image

Modifications

  • A new env FD_ENABLE_PDL, with a default value of 1, which means CUDA-PDL is enabled when sm>=90.
  • Modified the startup method of the kernel.
  • Changed some kernel names (when input's dtype is int).

Usage or Command

export FD_ENABLE_PDL=0 will disable CUDA-PDL.

Accuracy Tests

Append Attn Support CUDA-PDL

Checklist

  • Add at least a tag in the PR title.
    • Tag list: [[FDConfig],[APIServer],[Engine], [Scheduler], [PD Disaggregation], [Executor], [Graph Optimization], [Speculative Decoding], [RL], [Models], [Quantization], [Loader], [OP], [KVCache], [DataProcessor], [BugFix], [Docs], [CI], [Optimization], [Feature], [Benchmark], [Others], [XPU], [HPU], [GCU], [DCU], [Iluvatar], [Metax]]
    • You can add new tags based on the PR content, but the semantics must be clear.
  • Format your code, run pre-commit before commit.
  • Add unit tests. Please write the reason in this PR if no unit tests.
  • Provide accuracy results.
  • If the current PR is submitting to the release branch, make sure the PR has been submitted to the develop branch, then cherry-pick it to the release branch with the [Cherry-Pick] PR tag.

@paddle-bot
Copy link

paddle-bot bot commented Nov 16, 2025

Thanks for your contribution!

@ckl117 ckl117 changed the title [OP] Append Attn Support CUDA-PDL [Feature][OP] Append Attn Support CUDA-PDL Nov 17, 2025
@zhoutianzi666 zhoutianzi666 merged commit d58c1db into PaddlePaddle:develop Nov 17, 2025
14 of 16 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants