github-actions[bot] commented on issue #33670:
URL: https://github.com/apache/beam/issues/33670#issuecomment-2646624492
Performance change found in the
test:
`pytorch_image_classification_benchmarks-resnet152-GPU-mean_inference_batch_latency_micro_secs`
for the metric: `mean_inference_batch_latency_micro_secs`.
For more information on how to triage the alerts, please look at
`Triage performance alert issues` section of the
[README](https://github.com/apache/beam/tree/master/sdks/python/apache_beam/testing/analyzers/README.md#triage-performance-alert-issues).
`Test description:` Pytorch image classification on 50k images of size 224 x
224 with resnet 152 with Tesla T4 GPU.
Test link -
https://github.com/apache/beam/blob/42d0a6e3564d8b9c5d912428a6de18fb22a13ac1/.test-infra/jenkins/job_InferenceBenchmarkTests_Python.groovy#L151).
Test dashboard -
http://metrics.beam.apache.org/d/ZpS8Uf44z/python-ml-runinference-benchmarks?from=now-90d&to=now&viewPanel=2
```
timestamp: Sun Feb 9 06:35:11 2025, metric_value: 152221.64
timestamp: Sat Feb 8 06:56:17 2025, metric_value: 161781.34
timestamp: Fri Feb 7 06:27:45 2025, metric_value: 146430.54 <---- Anomaly
timestamp: Thu Feb 6 06:28:36 2025, metric_value: 80862.86
timestamp: Wed Feb 5 06:38:04 2025, metric_value: 85889.79
timestamp: Tue Feb 4 06:22:29 2025, metric_value: 80285.64
timestamp: Mon Feb 3 06:14:52 2025, metric_value: 144679.03
timestamp: Sun Feb 2 06:25:12 2025, metric_value: 76917.03
timestamp: Sat Feb 1 06:19:08 2025, metric_value: 77450.92
timestamp: Fri Jan 31 06:15:15 2025, metric_value: 84229.17
timestamp: Wed Jan 29 06:34:31 2025, metric_value: 73585.53
timestamp: Tue Jan 28 06:32:22 2025, metric_value: 80284.24
timestamp: Mon Jan 27 06:15:54 2025, metric_value: 70148.31
```
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
