KEMBAR78
[AOTI][Intel GPU] Add XPU quantization ops to AOT Inductor. by etaf · Pull Request #156572 · pytorch/pytorch · GitHub
Skip to content

Conversation

@etaf etaf requested review from EikanWang and gujinghui as code owners June 23, 2025 03:24
@pytorch-bot
Copy link

pytorch-bot bot commented Jun 23, 2025

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/156572

Note: Links to docs will display an error until the docs builds have been completed.

✅ You can merge normally! (1 Unrelated Failure)

As of commit d878b74 with merge base be16122 (image):

FLAKY - The following job failed but was likely due to flakiness present on trunk:

This comment was automatically generated by Dr. CI and updates every 15 minutes.

etaf added a commit that referenced this pull request Jun 23, 2025
@etaf etaf marked this pull request as draft June 23, 2025 03:24
@etaf etaf changed the title [AOTI][Intel GPU] Add XPU quantization ops to AOT Inductor. [WIP][AOTI][Intel GPU] Add XPU quantization ops to AOT Inductor. Jun 23, 2025
@github-actions
Copy link
Contributor

Attention! PyTorch one of the C-stable API file was changed

You MUST NOT change existing function declarations in this, as this header defines a stable C ABI. If you need to change the signature for a function, introduce a new v2 version of the function and modify code generation to target the new version of the function.


Caused by:

…uctor."

cc jgong5 mingfeima XiaobingSuper sanchitintel ashokei jingxu10 jerryzh168 voznesenskym penguinwu EikanWang Guobing-Chen zhuhaozhe blzheng wenzhe-nrv jiayisunx ipiszy chenyang78 kadeng muchulee8 amjames chauhang aakhundov

[ghstack-poisoned]
etaf added a commit that referenced this pull request Jun 23, 2025
…uctor."

cc jgong5 mingfeima XiaobingSuper sanchitintel ashokei jingxu10 jerryzh168 voznesenskym penguinwu EikanWang Guobing-Chen zhuhaozhe blzheng wenzhe-nrv jiayisunx ipiszy chenyang78 kadeng muchulee8 amjames chauhang aakhundov

[ghstack-poisoned]
etaf added a commit that referenced this pull request Jun 23, 2025
…uctor."

cc jgong5 mingfeima XiaobingSuper sanchitintel ashokei jingxu10 jerryzh168 voznesenskym penguinwu EikanWang Guobing-Chen zhuhaozhe blzheng wenzhe-nrv jiayisunx ipiszy chenyang78 kadeng muchulee8 amjames chauhang aakhundov

[ghstack-poisoned]
etaf added a commit that referenced this pull request Jun 23, 2025
@etaf etaf added the ciflow/xpu Run XPU CI tasks label Jun 23, 2025
…uctor."

cc jgong5 mingfeima XiaobingSuper sanchitintel ashokei jingxu10 jerryzh168 voznesenskym penguinwu EikanWang Guobing-Chen zhuhaozhe blzheng wenzhe-nrv jiayisunx ipiszy chenyang78 kadeng muchulee8 amjames chauhang aakhundov

[ghstack-poisoned]
etaf added a commit that referenced this pull request Jun 23, 2025
…uctor."

cc jgong5 mingfeima XiaobingSuper sanchitintel ashokei jingxu10 jerryzh168 voznesenskym penguinwu EikanWang Guobing-Chen zhuhaozhe blzheng wenzhe-nrv jiayisunx ipiszy chenyang78 kadeng muchulee8 amjames chauhang aakhundov

[ghstack-poisoned]
etaf added a commit that referenced this pull request Jun 23, 2025
@etaf etaf changed the title [WIP][AOTI][Intel GPU] Add XPU quantization ops to AOT Inductor. [AOTI][Intel GPU] Add XPU quantization ops to AOT Inductor. Jun 24, 2025
@etaf etaf marked this pull request as ready for review June 24, 2025 00:13
cc jgong5 mingfeima XiaobingSuper sanchitintel ashokei jingxu10 jerryzh168 voznesenskym penguinwu EikanWang Guobing-Chen zhuhaozhe blzheng wenzhe-nrv jiayisunx ipiszy chenyang78 kadeng muchulee8 amjames chauhang aakhundov

[ghstack-poisoned]
etaf added a commit that referenced this pull request Jun 24, 2025
@etaf etaf requested review from ZhiweiYan-96 and desertfire June 24, 2025 01:05
@etaf etaf reopened this Jul 22, 2025
@ZhiweiYan-96
Copy link
Collaborator

ZhiweiYan-96 commented Jul 23, 2025

@ZhiweiYan-96 I suggest to move the xpu quantization ops from aten/src/ATen/native/mkldnn/xpu/ to aten/src/ATen/native/quantized/xpu, which is the common design of cuda/cpu.

hi, @etaf , you may need consider change the cmake also? To assure the quantized ops are compiled.

Yes, but this PR does not move the cpp files.

oh, thanks. I prefer to keep the current file structure so as to aggregate the XPU oneDNN integrations in single directory. Currently, it is clean and easy to maintain. CPU-side need to add onednn-only operators like qliner_pointwise defined in library.cpp, while we only need to implement these ops. I think we currently can benefits from the easy structure and build system.

Copy link
Collaborator

Choose a reason for hiding this comment

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

@chuanqi129 , regarding this file, pls. help review it.

--shard "$1" "$NUM_TEST_SHARDS" \
--verbose

if [[ "${BUILD_ENVIRONMENT}" == *xpu* ]]; then
Copy link
Collaborator

Choose a reason for hiding this comment

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

Why we need add this test into function test_inductor_cpp_wrapper_shard()? This function only be tested with test config inductor_cpp_wrapper. We only use default test config in xpu test, could you please double confirm that those new test cases whether be tested with xpu ci job?

Copy link
Collaborator Author

Choose a reason for hiding this comment

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

No, but the reviewer asked us to add new test cases here.

@etaf
Copy link
Collaborator Author

etaf commented Aug 26, 2025

@desertfire Could you please kindly help review the stacked two PRs when convenient?

@xuhancn xuhancn requested a review from angelayi August 28, 2025 06:44
@xuhancn
Copy link
Collaborator

xuhancn commented Aug 28, 2025

@angelayi Could you please review it?

@etaf
Copy link
Collaborator Author

etaf commented Aug 28, 2025

@pytorchbot rebase

@xuhancn xuhancn added the ciflow/binaries Trigger all binary build and upload jobs on the PR label Aug 28, 2025
@xuhancn
Copy link
Collaborator

xuhancn commented Aug 28, 2025

@pytorchmergebot rebase

@xuhancn xuhancn self-assigned this Aug 28, 2025
@pytorchmergebot
Copy link
Collaborator

@pytorchbot started a rebase job onto refs/remotes/origin/viable/strict. Check the current status here

[ghstack-poisoned]
@pytorchmergebot
Copy link
Collaborator

Successfully rebased gh/etaf/138/orig onto refs/remotes/origin/viable/strict, please pull locally before adding more changes (for example, via ghstack checkout https://github.com/pytorch/pytorch/pull/156572)

pytorchmergebot pushed a commit that referenced this pull request Aug 28, 2025
@pytorchmergebot
Copy link
Collaborator

@pytorchbot started a rebase job onto refs/remotes/origin/viable/strict. Check the current status here

pytorchmergebot pushed a commit that referenced this pull request Aug 28, 2025
@pytorchmergebot
Copy link
Collaborator

Tried to rebase and push PR #156572, but it was already up to date. Try rebasing against main by issuing:
@pytorchbot rebase -b main

@etaf etaf added the ciflow/trunk Trigger trunk jobs on your pull request label Aug 28, 2025
-k 'xpu' \
--shard "$1" "$NUM_TEST_SHARDS" \
--verbose
fi
Copy link
Contributor

Choose a reason for hiding this comment

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

I'm unsure about this change but seems like CI is picking it up so looks ok?
cc @huydhn if you have more input 😅

@etaf
Copy link
Collaborator Author

etaf commented Aug 29, 2025

@pytorchbot merge

@pytorchmergebot
Copy link
Collaborator

Merge started

Your 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

Advanced Debugging
Check the merge workflow status
here

markc-614 pushed a commit to markc-614/pytorch that referenced this pull request Sep 17, 2025
@github-actions github-actions bot deleted the gh/etaf/138/head branch September 29, 2025 02:12
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

ciflow/binaries Trigger all binary build and upload jobs on the PR ciflow/inductor ciflow/trunk Trigger trunk jobs on your pull request ciflow/xpu Run XPU CI tasks Merged module: cpu CPU specific problem (e.g., perf, algorithm) module: inductor open source release notes: inductor (aoti)

Projects

Status: Done

Development

Successfully merging this pull request may close these issues.

10 participants