Skip to content

Conversation

@echeresh
Copy link
Contributor

@echeresh echeresh commented Dec 5, 2025

Port of #4423 + regression test.

The regression test covers two cases from MFDNN-14449 and MFDNN-13404. It's not super-straightforward though:

  • The test sets driver environment variables via CMake set_tests_properties + PROPERTIES ENVIRONMENT to trigger the bug
  • It's specific to f64 and PVC so I added an internal interface for GPU ISA query to target PVC only: dnnl_impl_gpu_intel_get_isa_name()

@echeresh echeresh requested a review from a team as a code owner December 5, 2025 21:32
@echeresh echeresh added the platform:gpu-intel Codeowner: @oneapi-src/onednn-gpu-intel label Dec 5, 2025
@echeresh echeresh requested a review from a team as a code owner December 5, 2025 21:32
@github-actions github-actions bot added component:tests Codeowner: @oneapi-src/onednn-arch third_party labels Dec 5, 2025
@echeresh echeresh changed the title Echeresh/fix pvcwa main third_party: ngen: restore PVC WAR bug workaround Dec 5, 2025
@echeresh echeresh force-pushed the echeresh/fix-pvcwa-main branch from 133fd08 to ca06071 Compare December 5, 2025 21:40
@echeresh
Copy link
Contributor Author

echeresh commented Dec 5, 2025

make test
set test_scope=NIGHTLY
disable test_device_cpu
disable benchdnn_all
enable benchdnn_conv
enable benchdnn_deconv
enable arch_gpu_xe-hpc
enable arch_gpu_xe-hpg-atsm
enable arch_gpu_xe-hpg-dg2
enable arch_gpu_xe-lp
enable arch_gpu_xe-lpg
enable arch_gpu_xe-lpg+
enable arch_gpu_xe2-hpg-bmg
enable arch_gpu_xe2-lpg
enable arch_gpu_xe3-lpg

@echeresh echeresh force-pushed the echeresh/fix-pvcwa-main branch 2 times, most recently from f735292 to 783441a Compare December 8, 2025 17:09
@echeresh
Copy link
Contributor Author

echeresh commented Dec 8, 2025

make test
set test_scope=NIGHTLY
disable test_device_cpu
disable benchdnn_all
enable benchdnn_conv
enable benchdnn_deconv
enable arch_gpu_xe-hpc
enable arch_gpu_xe-hpg-atsm
enable arch_gpu_xe-hpg-dg2
enable arch_gpu_xe-lp
enable arch_gpu_xe-lpg
enable arch_gpu_xe-lpg+
enable arch_gpu_xe2-hpg-bmg
enable arch_gpu_xe2-lpg
enable arch_gpu_xe3-lpg

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

component:tests Codeowner: @oneapi-src/onednn-arch platform:gpu-intel Codeowner: @oneapi-src/onednn-gpu-intel third_party

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants