Qualcomm AI Engine Direct - add pass for extra padding then maxpool2d #16534
+185
−11
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
Summary:
The padding value used in max_pool2d operations differs between PyTorch and QNN implementations.
PyTorch uses negative infinity, while QNN uses zero. To ensure consistent max_pool2d output across both frameworks,
we handle this by padding tensor with constant in advance then doing max_pool2d without constant padding.
Test plans:
python backends/qualcomm/tests/test_qnn_delegate.py TestQNNQuantizedOperator.test_qnn_backend_max_pool2d -b build-android -H ${HOST} -s ${SN} -m ${CHIPID}
python backends/qualcomm/tests/test_qnn_delegate.py TestQNNFloatingPointOperator.test_qnn_backend_max_pool2d -b build-android -H ${HOST} -s ${SN} -m ${CHIPID}
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/16534
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit f3c142d with merge base 47dc1de ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
meta-cla
bot
added
the
CLA Signed
Contributor
Author
|
@pytorchbot label "release notes: qualcomm" |
pytorch-bot
bot
added
the
release notes: qualcomm
Contributor
Author
|
Due to the padding value used in max_pool2d operations differs between PyTorch and QNN implementations. This pass will improve this situation. Please take a look. Thanks. |
cccclai
approved these changes
Jan 10, 2026
Contributor
|
Hey can you address the lintrunner error? |
Contributor
Contributor
Author
Sure. |
Contributor
Author
|
@pytorchbot label "release notes: qualcomm" |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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.
Qualcomm AI Engine Direct - add pass for extra padding then maxpool2d
Summary:
The padding value used in max_pool2d operations differs between PyTorch and QNN implementations. PyTorch uses negative infinity, while QNN uses zero. To ensure consistent max_pool2d output across both frameworks, we handle this by padding tensor with constant in advance then doing max_pool2d without constant padding. Note that for the quantization flow, we set quant_min as the padding value. If, at runtime, there is a value smaller than quant_min, it could result in an accuracy drop.
Test plans:
python backends/qualcomm/tests/test_qnn_delegate.py TestQNNQuantizedOperator.test_qnn_backend_max_pool2d -b build-android -H ${HOST} -s ${SN} -m ${CHIPID}
python backends/qualcomm/tests/test_qnn_delegate.py TestQNNFloatingPointOperator.test_qnn_backend_max_pool2d -b build-android -H ${HOST} -s ${SN} -m ${CHIPID}