Skip to content

Conversation

RUIJIEZHONG66166
Copy link
Contributor

@RUIJIEZHONG66166 RUIJIEZHONG66166 commented Aug 7, 2025

Enhance the op microbench test with device, test method and iter number ondemanded control.
disable_ut

@RUIJIEZHONG66166 RUIJIEZHONG66166 force-pushed the ruijie/set_different_time_measure branch from 13faaf7 to eda75dd Compare August 13, 2025 01:36
@RUIJIEZHONG66166 RUIJIEZHONG66166 force-pushed the ruijie/set_different_time_measure branch from 3a723b6 to 618e946 Compare August 14, 2025 06:47
@RUIJIEZHONG66166
Copy link
Contributor Author

RUIJIEZHONG66166 commented Aug 18, 2025

Ondemand test results: https://github.com/intel/torch-xpu-ops/actions/runs/16987046730
If Profile Change = "↓" or "↑", it means the regression or improvement exceed the threshold; if no Profile Change, it means the regression or improvement do not exceed the threshold.
Some outlier like case in scan.cumsum have regression because it compared with main branch results, and main branch's test scripts "scan.cumsum.py" has type issues which fixed by this PR. Will refresh the baseline after this pr merged.

@RUIJIEZHONG66166 RUIJIEZHONG66166 requested review from chuanqi129 and removed request for chuanqi129 August 18, 2025 08:37
@chuanqi129 chuanqi129 changed the title [Nightly] Set different time measure for microbench [CD] Set different time measure for microbench Aug 22, 2025
@RUIJIEZHONG66166 RUIJIEZHONG66166 requested review from chuanqi129 and removed request for chuanqi129 September 1, 2025 08:49
@RUIJIEZHONG66166 RUIJIEZHONG66166 requested review from chuanqi129 and removed request for chuanqi129 September 8, 2025 09:32
@RUIJIEZHONG66166
Copy link
Contributor Author

After rebasing, ondemand test results: https://github.com/intel/torch-xpu-ops/actions/runs/17570650066

@chuanqi129
Copy link
Contributor

After rebasing, ondemand test results: https://github.com/intel/torch-xpu-ops/actions/runs/17570650066

Hi @RUIJIEZHONG66166 , I have checked the on-demand test summary, there are a lot of regression/improvement cases, does it cause by test variance? Please hide improvement and mixed changes tables by default. And if we run 2 methods, let's mark those regression both on 2 metrics as regression cases

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants