🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/157994

• 📄 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 or our office hours

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

❌ 4 New Failures, 5 Pending

As of commit 1d51259 with merge base bfc873d ():

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

Hi @sanchitintel , I'm trying to upgrade oneDNN to v3.9 for testing. since v3.9 is not officially released, I tried main branch commit first. And it wil fail in test_jit_llga_fuser.py::TestOpCPU::test_add_cpu_float32. Could you please take a look first? Thanks!

FAILED [0.0335s] test_jit_llga_fuser.py::TestOpCPU::test_add_cpu_float32 - AssertionError: Tensor-likes are not close!
2025-07-10T02:59:20.4033167Z 
2025-07-10T02:59:20.4033380Z Mismatched elements: 13696 / 25088 (54.6%)
2025-07-10T02:59:20.4034391Z Greatest absolute difference: 1.701464974517854e+38 at index (0, 0, 5, 12) (up to 1e-05 allowed)
2025-07-10T02:59:20.4035723Z Greatest relative difference: inf at index (0, 4, 25, 12) (up to 1.3e-06 allowed)
2025-07-10T02:59:20.4036446Z 
2025-07-10T02:59:20.4036941Z To execute this test, run the following from the base repo dir:
2025-07-10T02:59:20.4037928Z     python test/test_jit_llga_fuser.py TestOpCPU.test_add_cpu_float32

Does this fix uxlfoundation/oneDNN#3383 btw? ;-;

Does this fix uxlfoundation/oneDNN#3383 btw? ;-;

Unfortunately no. We recommend disabling IPO for oneDNN builds on Windows.

Hi @sanchitintel , I'm trying to upgrade oneDNN to v3.9 for testing. since v3.9 is not officially released, I tried main branch commit first. And it wil fail in test_jit_llga_fuser.py::TestOpCPU::test_add_cpu_float32. Could you please take a look first? Thanks!

FAILED [0.0335s] test_jit_llga_fuser.py::TestOpCPU::test_add_cpu_float32 - AssertionError: Tensor-likes are not close!
2025-07-10T02:59:20.4033167Z 
2025-07-10T02:59:20.4033380Z Mismatched elements: 13696 / 25088 (54.6%)
2025-07-10T02:59:20.4034391Z Greatest absolute difference: 1.701464974517854e+38 at index (0, 0, 5, 12) (up to 1e-05 allowed)
2025-07-10T02:59:20.4035723Z Greatest relative difference: inf at index (0, 4, 25, 12) (up to 1.3e-06 allowed)
2025-07-10T02:59:20.4036446Z 
2025-07-10T02:59:20.4036941Z To execute this test, run the following from the base repo dir:
2025-07-10T02:59:20.4037928Z     python test/test_jit_llga_fuser.py TestOpCPU.test_add_cpu_float32

UT python test/test_jit_llga_fuser.py TestOpCPU.test_add_cpu_float32 has been fixed.

Hi @vpirogov ,

For the bazel build failure of https://github.com/pytorch/pytorch/actions/runs/16334831332/job/46154536338, does #include <altivec.h> need to be inside #ifdef __MMA__ in src/cpu/ppc64/ppc64_gemm_reorder.cpp:20? And I'm not sure if we need to build files under mkl-dnn/src/cpu/ppc64/, can we just simply disable the build of ppc64?

@pytorchbot rebase

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

Successfully rebased yanbing/update_onednn_v3.9 onto refs/remotes/origin/viable/strict, please pull locally before adding more changes (for example, via git checkout yanbing/update_onednn_v3.9 && git pull --rebase)

Hi @milpuz01 , could you please help verify this CI failure https://github.com/pytorch/pytorch/actions/runs/16643249983/job/47097723925?pr=157994? Thanks!

Hi @milpuz01 , could you please help verify this CI failure https://github.com/pytorch/pytorch/actions/runs/16643249983/job/47097723925?pr=157994? Thanks!

Thanks for testing this @yanbing-j. Seems like ACL version should be bumped up to the latest one.

2025-07-31T07:59:30.9603124Z -- Found ACL: /ComputeLibrary
2025-07-31T07:59:31.0004741Z �[31mCMake Error at third_party/ideep/mkl-dnn/cmake/ACL.cmake:67 (message):
2025-07-31T07:59:31.0005536Z Detected ACL version 25.02, but minimum compatible is 52.0

I can test the build with a forked version of this changes after bumping ACL properly to verify.

@yanbing-j with this commit, the build is working fine. I'll open a draft PR to upgrade ACL with this version. The build is working fine now: https://github.com/pytorch/pytorch/actions/runs/16647200279/job/47110635776

@yanbing-j with this commit, the build is working fine. I'll open a draft PR to upgrade ACL with this version. The build is working fine now: https://github.com/pytorch/pytorch/actions/runs/16647200279/job/47110635776

Thanks @aditew01 ! Will the PR of upgrading ACL to the version be fast? Maybe I can involve the code change inside the current PR.

Maybe I can involve the code change inside the current PR.

@yanbing-j sounds good to me. Thanks!

Maybe I can involve the code change inside the current PR.

@yanbing-j sounds good to me. Thanks!

Okay. I have put this into the task list. Thanks!