Arm backend: Avoid unnecessary lowering pass retraces#21138
Open
YufengShi-dudu wants to merge 1 commit into
Open
Arm backend: Avoid unnecessary lowering pass retraces#21138YufengShi-dudu wants to merge 1 commit into
YufengShi-dudu wants to merge 1 commit into
Conversation
Add cheap applicability checks before selected Arm passes enter the ExportPass retracing path. Skip InsertRescalePass for profiles without integer support and graphs without DQ nodes. Move symbolic shape and adaptive pool passes to ArmOpTargetedPass, and pre-scan RemoveGetItemPass producers. These checks preserve target-present transformations while avoiding retraces on non-applicable graphs. Lowering times are medians of three fresh-process exports with no warm-up. Regular models use TOSA FP with FP32; full Qwen uses BF16. | Model | Base (s) | Updated (s) | Speedup | |-------------------|---------:|------------:|--------:| | ResNet-18 | 3.633 | 3.519 | 3.16% | | MobileNetV3 | 11.063 | 10.100 | 8.70% | | DeiT-tiny | 33.323 | 31.137 | 6.56% | | Conformer | 13.121 | 12.204 | 6.98% | | Test Qwen text | 11.582 | 10.962 | 5.35% | | Test Qwen vision | 11.979 | 11.388 | 4.93% | | Full Qwen text | 155.825 | 146.824 | 5.78% | | Full Qwen vision | 126.671 | 120.153 | 5.15% | Change-Id: I3c539e023efb5d1e59a725e531c850a897758002 Signed-off-by: Yufeng Shi <yufeng.shi@arm.com>
YufengShi-dudu
added
partner: arm
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/21138
Note: Links to docs will display an error until the docs builds have been completed. ❗ 1 Active SEVsThere are 1 currently active SEVs. If your PR is affected, please view them below: ❌ 1 New Failure, 1 Unrelated FailureAs of commit 12d98d3 with merge base 5442036 ( NEW FAILURE - The following job has failed:
BROKEN TRUNK - The following job failed but were present on the merge base:👉 Rebase onto the `viable/strict` branch to avoid these failures
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
meta-cla
Bot
added
the
CLA Signed
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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.
Add cheap applicability checks before selected Arm passes enter the ExportPass retracing path. Skip InsertRescalePass for profiles without integer support and graphs without DQ nodes. Move symbolic shape and adaptive pool passes to ArmOpTargetedPass, and pre-scan RemoveGetItemPass producers.
These checks preserve target-present transformations while avoiding retraces on non-applicable graphs.
Lowering times are medians of three fresh-process exports with no warm-up. Regular models use TOSA FP with FP32; full Qwen uses BF16.
Change-Id: I3c539e023efb5d1e59a725e531c850a897758002
cc @digantdesai @freddan80 @per @zingo @oscarandersson8218 @mansnils @Sebastian-Larsson @robell @rascani