[FP8] FP8 for SwishLayerNorm #157574
Closed
[FP8] FP8 for SwishLayerNorm #157574
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/157574
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit 153bf89 with merge base 19ae5af ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
|
This pull request was exported from Phabricator. Differential Revision: D76531303 |
frank-wei
approved these changes
Jul 3, 2025
|
LGTM but let's make sure all tests are passed. |
Summary:
Add a pass use_triton_fp8_swish_replace_normal_swish to replace _triton_swish_rms_norm with its counterpart that supports fp8 triton_swish_rms_norm, and turn on fp8 during inference.
Test Plan:
```
buck2 run mode/opt mode/inplace -c fbcode.platform010_cuda_version=12.4 -c fbcode.nvcc_arch=h100 caffe2/torch/fb/model_transform/experimental/benchmark:mts_gpu_benchmark -- --lower-backend=AOT_INDUCTOR --model-snapshot-id=899072727_0 --node-replacement-dict="{}" --gpu-trace --add-passes=use_triton_fp8_swish_replace_normal_swish
```
The perf improvement on the 100x model with this pass is roughly ~7%, details are recorded [here](https://docs.google.com/document/d/1eIV_OTQyQcf_DlEDxwycTwhyGxT5OJkLzs8cPL6EMYc/edit?tab=t.0)
Rollback Plan:
Reviewed By: frank-wei
Differential Revision: D76531303
oniononion36
force-pushed
the
export-D76531303
branch
from
545809e to
153bf89
Compare
|
This pull request was exported from Phabricator. Differential Revision: D76531303 |
|
@pytorchbot merge (Initiating merge automatically since Phabricator Diff has merged) |
Merge startedYour change will be merged once all checks pass (ETA 0-4 Hours). Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Summary: Add a pass use_triton_fp8_swish_replace_normal_swish to replace _triton_swish_rms_norm with its counterpart that supports fp8 triton_swish_rms_norm, and turn on fp8 during inference.
Test Plan:
The perf improvement on the 100x model with this pass is roughly ~7%, details are recorded here
Rollback Plan:
Reviewed By: frank-wei
Differential Revision: D76531303
cc @ezyang @SherlockNoMad @EikanWang @jgong5 @wenzhe-nrv