Issue metadata
Sign in to add a comment
|
java_heap:allocated_objects_size_avg is very noisy |
||||||||||||||||||||||||
Issue descriptionjava_heap:allocated_objects_size_avg is noisy enough that we are getting spurious regressions. For eample here is a recent system_health.memory_mobile N5X log https://luci-logdog.appspot.com/v/?s=chrome%2Fbb%2Fchromium.perf%2FAndroid_Nexus5X_Perf__1_%2F4443%2F%2B%2Frecipes%2Fsteps%2Fsystem_health.memory_mobile%2F0%2Flogs%2Fjson.output%2F0 and these are the three numbers for load_news@@memory:chrome:browser_process:reported_by_chrome:java_heap:allocated_objects_size_avg: 19698992 22397328 18281744 Here are a graphs showing a spurious regression for load_news java_heap:allocated_objects_size_avg: https://chromeperf.appspot.com/report?sid=50feba5c236dce85d8e0c960b5ecd4bfb7f738dedabbaef35685afc35318b096&rev=447645 And two traces: Before: https://console.developers.google.com/m/cloudstorage/b/chrome-telemetry-output/o/trace-file-id_14-2017-02-01_15-09-14-20890.html After: https://console.developers.google.com/m/cloudstorage/b/chrome-telemetry-output/o/trace-file-id_14-2017-02-01_23-39-42-9454.html
,
Aug 22 2017
From the analysis in 733894, and confirmed e.g. here: https://chromeperf.appspot.com/report?sid=74750f70661d6a71660fddc2ccb0147fe2f2989b86e6bb668a8369335ee55080 Noise in this metric is no longer a concern. The main metric to focus now is by_os:system_memory:java_heap. |
|||||||||||||||||||||||||
►
Sign in to add a comment |
|||||||||||||||||||||||||
Comment 1 by benhenry@chromium.org
, Aug 7 2017Owner: ----