Skip to content

Fixed SDPA slow down and linear slow down#3700

Merged
peri044 merged 6 commits intomainfrom
broadcast-removal
Aug 4, 2025
Merged

Fixed SDPA slow down and linear slow down#3700
peri044 merged 6 commits intomainfrom
broadcast-removal

Conversation

@cehongwang
Copy link
Copy Markdown
Collaborator

@cehongwang cehongwang commented Jul 21, 2025

Description

Please include a summary of the change and which issue is fixed. Please also include relevant motivation and context. List any dependencies that are required for this change.

Fixes # (issue)

Type of change

Please delete options that are not relevant and/or add your own.

  • Bug fix (non-breaking change which fixes an issue)
  • New feature (non-breaking change which adds functionality)
  • Breaking change (fix or feature that would cause existing functionality to not work as expected)
  • This change requires a documentation update

Checklist:

  • My code follows the style guidelines of this project (You can use the linters)
  • I have performed a self-review of my own code
  • I have commented my code, particularly in hard-to-understand areas and hacks
  • I have made corresponding changes to the documentation
  • I have added tests to verify my fix or my feature
  • New and existing unit tests pass locally with my changes
  • I have added the relevant labels to my PR in so that relevant reviewers are notified

@meta-cla meta-cla bot added the cla signed label Jul 21, 2025
@github-actions github-actions bot added component: conversion Issues re: Conversion stage component: converters Issues re: Specific op converters component: api [Python] Issues re: Python API component: dynamo Issues relating to the `torch.compile` or `torch._dynamo.export` paths labels Jul 21, 2025
@github-actions github-actions bot requested a review from zewenli98 July 21, 2025 21:28
@github-actions github-actions bot added the component: lowering Issues re: The lowering / preprocessing passes label Jul 22, 2025
peri044
peri044 reviewed Jul 23, 2025
View reviewed changes
@cehongwang cehongwang marked this pull request as ready for review July 24, 2025 00:03
@cehongwang cehongwang changed the title skipped unnecessary broadcast Fixed SDPA slow down and linear slow down Jul 24, 2025
This was linked to issues Jul 24, 2025
🐛 [Bug] SDPA decomposition causing TorchTRT to be 2x slower than ONNX on SD3.5 #3682
Closed
🐛 [Bug] SDPA in Torch-TRT is slower than SDPA in ONNX as batch size (num_frames) grow for Wan2.1 #3695
Closed
🐛 [Bug] perf gap reduce on timm/ViT #3704
Closed
[Bug] MHA Kernels and Linear Kernels are slower in FLUX #3707
Closed
@zewenli98 zewenli98 mentioned this pull request Jul 24, 2025
Closed
peri044
peri044 reviewed Aug 1, 2025
View reviewed changes
py/torch_tensorrt/dynamo/conversion/aten_ops_converters.py
@@ -532,6 +532,7 @@ def aten_ops_gelu(


@dynamo_tensorrt_converter(torch.ops.aten.matmul, supports_dynamic_shapes=True)
Copy link
Copy Markdown
Collaborator

@peri044 peri044 Aug 1, 2025

Choose a reason for hiding this comment

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

Is this actually needed ? consider removing it if unnecessary

@peri044 peri044 merged commit 97b4073 into main Aug 4, 2025
113 of 114 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@peri044 peri044 peri044 left review comments

@zewenli98 zewenli98 zewenli98 left review comments

Assignees

No one assigned

Labels

cla signed component: api [Python] Issues re: Python API component: conversion Issues re: Conversion stage component: converters Issues re: Specific op converters component: dynamo Issues relating to the `torch.compile` or `torch._dynamo.export` paths component: lowering Issues re: The lowering / preprocessing passes

Projects

None yet

Milestone

No milestone

3 participants

@cehongwang @peri044 @zewenli98