[GHA] Uplift Linux GPU RT version to 25.09.32961.5 #17498

bb-sycl · 2025-03-18T03:14:44Z

Scheduled drivers uplift

sycl/test-e2e/Matrix/SG32/joint_matrix_annotated_ptr.cpp

sycl/test-e2e/Matrix/SG32/joint_matrix_colA_rowB_colC.cpp

sycl/test-e2e/Matrix/SG32/joint_matrix_int8_rowmajorA_rowmajorB.cpp

sycl/test-e2e/Matrix/SG32/joint_matrix_out_bounds.cpp

sycl/test-e2e/Matrix/SG32/joint_matrix_prefetch.cpp

sycl/test-e2e/Matrix/joint_matrix_16bit_colmajorA_colmajorB.cpp

sycl/test-e2e/Matrix/joint_matrix_colA_rowB_colC.cpp

sycl/test-e2e/Matrix/joint_matrix_out_bounds_colmajor.cpp

sycl/test-e2e/Matrix/joint_matrix_transposeAB.cpp

YuriPlyakhin

LGTM, but I'm wondering how these tests pass on BMG and did not pass on PVC.
The implementation/drivers should be the same for both platforms,...

sarnex · 2025-03-18T16:58:59Z

No idea and not touching it :P

YuriPlyakhin · 2025-03-18T19:06:28Z

No idea and not touching it :P

I guess, it is because they got into unsupported list, due to several backends are being tested on PVC.

YuriPlyakhin · 2025-03-18T19:10:10Z

@sarnex , I see you have some problems with conditions for Matrix test. I'm pretty sure, if you just remove xfail, the test should pass. So, it might be easier way. Up to you.

Co-authored-by: Yury Plyakhin <[email protected]>

Signed-off-by: Sarnie, Nick <[email protected]>

sarnex · 2025-03-19T16:48:04Z

@YuriPlyakhin Do you mean just the XPASSing test SYCL :: Matrix/SG32/get_coordinate_ops.cpp or all of them?

YuriPlyakhin · 2025-03-19T18:09:06Z

@YuriPlyakhin Do you mean just the XPASSing test SYCL :: Matrix/SG32/get_coordinate_ops.cpp or all of them?

I mean, the best is to try removing XFAIL from all that you touched in this PR and that are passing on BMG and see how it goes. My bet is all should just pass on any GPU, where matrix and specific tested feature is supported.
But at least, if you have problems with condition of coordinate test and don't want to touch others, you can try removing xfail completely from this test.

Signed-off-by: Sarnie, Nick <[email protected]>

sarnex · 2025-03-19T18:37:28Z

Hopefully latest commit is what you meant

Signed-off-by: Sarnie, Nick <[email protected]>

sarnex · 2025-03-21T14:24:17Z

@YuriPlyakhin Some tests failed when removing the XFAIL, so I XFAILed those only in the newly failing cases. Hopefully these changes are okay, please take a final look before I merge :)

Postcommit passed

sarnex · 2025-03-21T14:34:13Z

Actually I'm merging this now because it conflicts with Andrei's PR and I want to beat him :P

If there's any feedback, happy to address post commit

YuriPlyakhin · 2025-03-21T18:48:53Z

@YuriPlyakhin Some tests failed when removing the XFAIL, so I XFAILed those only in the newly failing cases. Hopefully these changes are okay, please take a final look before I merge :)

Postcommit passed

Thank you @sarnex , your changes make sense to me.

This is from #17498 Patch-by: Sarnie, Nick <[email protected]> Co-authored-by: Yury Plyakhin <[email protected]>

bb-sycl requested a review from a team as a code owner March 18, 2025 03:14

bb-sycl temporarily deployed to WindowsCILock March 18, 2025 03:14 — with GitHub Actions Inactive

bb-sycl temporarily deployed to WindowsCILock March 18, 2025 03:27 — with GitHub Actions Inactive

sarnex self-assigned this Mar 18, 2025

sarnex force-pushed the ci/update_gpu_driver-linux-25.09.32961.5 branch from 49b94bd to e26bb0f Compare March 18, 2025 14:52

sarnex requested a review from a team as a code owner March 18, 2025 14:52

sarnex temporarily deployed to WindowsCILock March 18, 2025 14:53 — with GitHub Actions Inactive

sarnex had a problem deploying to WindowsCILock March 18, 2025 15:37 — with GitHub Actions Error