Issue metadata
Sign in to add a comment
|
14.7% - 28.9% regression in memory:chrome:all_processes:reported_by_chrome:v8:effective_size_avg at 515951:516016 |
||||||||||||||||||||||||
Issue descriptionThere was originally a regression at 515351:515651 (see point 515651 at https://chromeperf.appspot.com/group_report?keys=agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDg94uFzwsM), but this was fixed at 515852:515857. The same regression seems to have reappeared at 515951:516016. This regression has only been alerted on the VR perf tests, but the same regression is visible on non-VR tests, e.g. https://chromeperf.appspot.com/report?sid=918403aa4fae28e2d4f40c341d4a0828a3fa893bd8bb8fd7a8a366d49c4f9caa. At point 515681 on that graph, you can see the initial regression, which is then fixed at point 515857, then reappears at point 515960.
,
Nov 23 2017
Note: --lazy-handler-deserialization flips are on the range at Point ID: 518562 and Point ID: 515651. I don't think it's the main cause though, since improvement when lazy handlers were first switched on at Point ID: 515144 was much less than the regression at 515651.
,
Nov 23 2017
This is an accounting issue. Large table that was not accounted in the metric is now accounted. |
|||||||||||||||||||||||||
►
Sign in to add a comment |
|||||||||||||||||||||||||
Comment 1 by hablich@chromium.org
, Nov 23 2017Owner: u...@chromium.org
Status: Assigned (was: Untriaged)