Skip to content

xe: softmax: restrict new kernel from PR2525 for only xe_hpc #3339

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Merged
merged 1 commit into from
May 30, 2025
Merged

xe: softmax: restrict new kernel from PR2525 for only xe_hpc #3339

merged 1 commit into from
May 30, 2025

Conversation

pv-pterab-s
Copy link
Contributor

@pv-pterab-s pv-pterab-s commented May 27, 2025 •
edited
Loading

The new softmax reusable kernel shows regressions on ATS-M, DG2, BMG, Xe-LPG+ (https://jira.devtools.intel.com/browse/MFDNN-13734). This PR restricts use of the new kernel to Xe-HPC to enhance performance while avoiding regressions on other platforms. The restriction is temporary while modifications for the former architectures are readied.

@github-actions github-actions bot added the platform:gpu-intel Codeowner: @oneapi-src/onednn-gpu-intel label May 27, 2025
@karturov
Copy link
Contributor

make test perf-gpu
set primitive=softmax

@pv-pterab-s pv-pterab-s force-pushed the pryorgal/softmax-accel-only-pvc branch from 530cdb3 to 7bb99fd Compare May 27, 2025 18:56
@pv-pterab-s
Copy link
Contributor Author

make test perf-gpu
disable os_win
set primitive=softmax

@pv-pterab-s pv-pterab-s force-pushed the pryorgal/softmax-accel-only-pvc branch from 7bb99fd to b9c8943 Compare May 27, 2025 22:23
@github-actions github-actions bot added documentation A request to change/fix/improve the documentation. Codeowner: @oneapi-src/onednn-doc platform:cpu-x64 Intel64/AMD64 processors. Codeowner: @oneapi-src/onednn-cpu-x64 platform:gpu-nvidia Codeowner: @oneapi-src/onednn-gpu-nvidia devops Github automation platform:gpu-generic Codeowner: @oneapi-src/onednn-gpu-generic governance component:api Codeowner: @oneapi-src/onednn-arch component:graph-api Codeowner: @oneapi-src/onednn-graph component:tests Codeowner: @oneapi-src/onednn-arch component:build component:common third_party labels May 27, 2025
@pv-pterab-s pv-pterab-s force-pushed the pryorgal/softmax-accel-only-pvc branch from b9c8943 to 73d3c38 Compare May 27, 2025 22:25
@github-actions github-actions bot removed documentation A request to change/fix/improve the documentation. Codeowner: @oneapi-src/onednn-doc platform:cpu-x64 Intel64/AMD64 processors. Codeowner: @oneapi-src/onednn-cpu-x64 platform:gpu-nvidia Codeowner: @oneapi-src/onednn-gpu-nvidia devops Github automation platform:gpu-generic Codeowner: @oneapi-src/onednn-gpu-generic governance component:api Codeowner: @oneapi-src/onednn-arch component:graph-api Codeowner: @oneapi-src/onednn-graph component:tests Codeowner: @oneapi-src/onednn-arch component:build component:common third_party labels May 27, 2025
@pv-pterab-s pv-pterab-s force-pushed the pryorgal/softmax-accel-only-pvc branch from 73d3c38 to 19cc34d Compare May 28, 2025 00:01
@pv-pterab-s
Copy link
Contributor Author

make test perf-gpu
disable os_win
set primitive=softmax

@pv-pterab-s pv-pterab-s marked this pull request as ready for review May 28, 2025 09:45
@pv-pterab-s pv-pterab-s requested a review from a team as a code owner May 28, 2025 09:45
@pv-pterab-s pv-pterab-s requested review from umar456 and syurkevi May 28, 2025 09:46
@pv-pterab-s pv-pterab-s force-pushed the pryorgal/softmax-accel-only-pvc branch from 19cc34d to fb9749f Compare May 28, 2025 14:11
@pv-pterab-s
Copy link
Contributor Author

make test linters

@pv-pterab-s pv-pterab-s marked this pull request as draft May 28, 2025 19:13
@pv-pterab-s pv-pterab-s force-pushed the pryorgal/softmax-accel-only-pvc branch from fb9749f to 4fe1e29 Compare May 28, 2025 19:15
@pv-pterab-s
Copy link
Contributor Author

make test perf-gpu
set primitive=softmax

@pv-pterab-s
Copy link
Contributor Author

make test
disable test_device_cpu
disable build_cpu_runtime_omp
disable build_cpu_runtime_sycl
disable build_cpu_runtime_tbb
disable benchdnn_all
enable benchdnn_softmax
enable test_device_gpu
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

1 similar comment
@pv-pterab-s
Copy link
Contributor Author

make test
disable test_device_cpu
disable build_cpu_runtime_omp
disable build_cpu_runtime_sycl
disable build_cpu_runtime_tbb
disable benchdnn_all
enable benchdnn_softmax
enable test_device_gpu
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

@pv-pterab-s
Copy link
Contributor Author

make test
disable test_device_cpu
disable build_cpu_runtime_omp
disable build_cpu_runtime_sycl
disable build_cpu_runtime_tbb
disable compiler_icx-oss
disable benchdnn_all
enable benchdnn_softmax
enable test_device_gpu
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

@pv-pterab-s pv-pterab-s marked this pull request as ready for review May 30, 2025 21:15
@pv-pterab-s pv-pterab-s merged commit b68d92e into main May 30, 2025
9 of 11 checks passed
@pv-pterab-s pv-pterab-s deleted the pryorgal/softmax-accel-only-pvc branch May 30, 2025 21:21
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@syurkevi syurkevi syurkevi approved these changes

@atkassen atkassen atkassen approved these changes

@umar456 umar456 Awaiting requested review from umar456

Assignees
No one assigned
Labels
platform:gpu-intel Codeowner: @oneapi-src/onednn-gpu-intel
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
4 participants
@pv-pterab-s @karturov @syurkevi @atkassen