[xpu][fix] Fix OneDNN deconv when using output_padding by LuFinch · Pull Request #169176 · pytorch/pytorch

LuFinch · 2025-11-27T09:41:11Z

Fixes intel/torch-xpu-ops#1898. The output padding was not used. This PR takes this parameter into account.

cc @jgong5 @mingfeima @XiaobingSuper @sanchitintel @ashokei @jingxu10 @jerryzh168 @aditew01 @voznesenskym @penguinwu @EikanWang @Guobing-Chen @zhuhaozhe @blzheng @wenzhe-nrv @jiayisunx @ipiszy @kadeng @muchulee8 @amjames @chauhang @aakhundov @coconutruben @jataylo

pytorch-bot · 2025-11-27T09:41:15Z

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/169176

📄 Preview Python docs built from this PR
📄 Preview C++ docs built from this PR
❓ Need help or want to give feedback on the CI? Visit the bot commands wiki

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 7cf7632 with merge base c6ae757 ():

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

trunk / linux-jammy-rocm-py3.10 / test (default, 4, 6, linux.rocm.gpu.gfx942.1) (gh) (similar failure)
test/inductor/test_native_matmul.py::TestTritonDotReduction::test_matmul_fp16

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

EikanWang · 2025-11-28T06:08:15Z

aten/src/ATen/native/mkldnn/xpu/detail/Deconv.cpp

  dnnl::memory::dims _stride = stride.vec();
-  dnnl::memory::dims _padding = padding.vec();
+  dnnl::memory::dims _padding_l = padding.vec();
+  dnnl::memory::dims _padding_r = padding_r(padding, dst_padding);


What does r(padding_r) represent? Right?

pytorch-bot · 2025-12-01T02:45:14Z

To add the ciflow label ciflow/xpu please first approve the workflows that are awaiting approval (scroll to the bottom of this page).

This helps ensure we don't trigger CI on this PR until it is actually authorized to do so. Please ping one of the reviewers if you do not have access to approve and run workflows.

LuFinch · 2025-12-03T01:54:48Z

@pytorchbot rebase

pytorchmergebot · 2025-12-03T01:56:23Z

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

pytorchmergebot · 2025-12-03T01:56:26Z

Successfully rebased lfq/fix_deconv onto refs/remotes/origin/viable/strict, please pull locally before adding more changes (for example, via git checkout lfq/fix_deconv && git pull --rebase)

LuFinch · 2025-12-04T05:04:54Z

@pytorchbot merge

pytorchmergebot · 2025-12-04T05:06:50Z

Merge failed

Reason: Approvers from one of the following sets are needed:

superuser (pytorch/metamates)
Core Reviewers (mruberry, lezcano, Skylion007, ngimel, peterbell10, ...)
Core Maintainers (soumith, gchanan, ezyang, dzhulgakov, malfet, ...)

Details for Dev Infra team

Raised by workflow job

Failing merge rule: Core Maintainers

LuFinch · 2025-12-04T13:41:44Z

@EikanWang Could you help merge? It needs approval

EikanWang · 2025-12-05T08:08:46Z

Hi @jansel , @desertfire , this PR intends to fix a deconv issue on XPU. And we also need to fix some Inductor test cases, respectively. Regarding the XPU part, it looks good to me. For Inductor test cases, it would be appreciated if you could review the related changes whenever you have a moment.

EikanWang · 2025-12-05T12:33:17Z

The CI failures are irrelevant to this PR.

LuFinch · 2025-12-08T01:49:49Z

@pytorchbot merge

pytorchmergebot · 2025-12-08T01:51:45Z

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

Fixes intel/torch-xpu-ops#1898. The output padding was not used. This PR takes this parameter into account. Pull Request resolved: pytorch#169176 Approved by: https://github.com/EikanWang, https://github.com/jansel

Fixes intel/torch-xpu-ops#1898. The output padding was not used. This PR takes this parameter into account. Pull Request resolved: #169176 Approved by: https://github.com/EikanWang, https://github.com/jansel

Fixes intel/torch-xpu-ops#1898. The output padding was not used. This PR takes this parameter into account. Pull Request resolved: pytorch#169176 Approved by: https://github.com/EikanWang, https://github.com/jansel

LuFinch requested review from EikanWang and gujinghui as code owners November 27, 2025 09:41

pytorch-bot bot added the module: cpu CPU specific problem (e.g., perf, algorithm) label Nov 27, 2025

LuFinch mentioned this pull request Nov 27, 2025

could not create a primitive descriptor for the deconvolution forward propagation primitive intel/torch-xpu-ops#1898

Closed

1 task

pytorchbot added the open source label Nov 27, 2025

EikanWang reviewed Nov 28, 2025

View reviewed changes

EikanWang approved these changes Dec 1, 2025

View reviewed changes

EikanWang added the ciflow/xpu Run XPU CI tasks label Dec 1, 2025

pytorch-bot bot removed the ciflow/xpu Run XPU CI tasks label Dec 1, 2025

EikanWang added ciflow/xpu Run XPU CI tasks topic: not user facing topic category labels Dec 1, 2025

LuFinch force-pushed the lfq/fix_deconv branch from 5696c24 to e4b4f82 Compare December 2, 2025 03:35

pytorch-bot bot added module: inductor and removed ciflow/xpu Run XPU CI tasks labels Dec 2, 2025

chuanqi129 added the ciflow/xpu Run XPU CI tasks label Dec 2, 2025

pytorch-bot bot removed the ciflow/xpu Run XPU CI tasks label Dec 2, 2025

chuanqi129 added the ciflow/xpu Run XPU CI tasks label Dec 2, 2025

LuFinch added 3 commits December 3, 2025 01:56

fix deconv when using output_padding

848b926

remove expected fail list

cb7bef2

fix lint

7cf7632

pytorchmergebot force-pushed the lfq/fix_deconv branch from 947193d to 7cf7632 Compare December 3, 2025 01:56

pytorch-bot bot removed the ciflow/xpu Run XPU CI tasks label Dec 3, 2025

pytorch-bot bot added the ciflow/trunk Trigger trunk jobs on your pull request label Dec 4, 2025

pytorchmergebot added the merging label Dec 4, 2025

pytorchmergebot removed the merging label Dec 4, 2025

LuFinch requested a review from EikanWang December 4, 2025 13:40

EikanWang requested review from desertfire and jansel December 5, 2025 06:13

EikanWang approved these changes Dec 5, 2025

View reviewed changes

jansel approved these changes Dec 6, 2025

View reviewed changes

pytorchmergebot added the merging label Dec 8, 2025

pytorchmergebot added the Merged label Dec 8, 2025

pytorchmergebot closed this in 2d1c0af Dec 8, 2025

pytorchmergebot removed the merging label Dec 8, 2025

LuFinch deleted the lfq/fix_deconv branch January 28, 2026 05:29

Conversation

LuFinch commented Nov 27, 2025 • edited by pytorch-bot bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

pytorch-bot bot commented Nov 27, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/169176

✅ You can merge normally! (1 Unrelated Failure)

Uh oh!

EikanWang Nov 28, 2025

Choose a reason for hiding this comment

Uh oh!

LuFinch Nov 28, 2025

Choose a reason for hiding this comment

Uh oh!

pytorch-bot bot commented Dec 1, 2025

Uh oh!

LuFinch commented Dec 3, 2025

Uh oh!

pytorchmergebot commented Dec 3, 2025

Uh oh!

pytorchmergebot commented Dec 3, 2025

Uh oh!

LuFinch commented Dec 4, 2025

Uh oh!

pytorchmergebot commented Dec 4, 2025

Merge failed

Uh oh!

LuFinch commented Dec 4, 2025

Uh oh!

EikanWang commented Dec 5, 2025

Uh oh!

EikanWang commented Dec 5, 2025

Uh oh!

LuFinch commented Dec 8, 2025

Uh oh!

pytorchmergebot commented Dec 8, 2025

Merge started

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

LuFinch commented Nov 27, 2025 •

edited by pytorch-bot bot

Loading

pytorch-bot bot commented Nov 27, 2025 •

edited

Loading