1480693 - 2.71 - 16.8% rasterflood_svg / tp5o / tp5o responsiveness / tp5o_scroll / tp5o_webext / tp6_google (linux64) regression on push bfa19e6f2407 (Wed Aug 1 2018)

Status: RESOLVED INVALID

(Core :: Graphics: Layers, defect, P3)

Description

[Ionuț Goldan [:igoldan]]

Talos has detected a Firefox performance regression from push:

https://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=c918e687d594375e1964a68c9fceec8a08f61e59&tochange=bfa19e6f240792564d99c2f85c92354c0f1c84e4

As author of one of the patches included in that push, we need your help to address this regression.

Regressions:

 17%  rasterflood_svg linux64 pgo e10s stylo     10,469.49 -> 12,227.84
 11%  tp5o responsiveness linux64 opt e10s stylo 0.62 -> 0.69
 10%  rasterflood_svg linux64 opt e10s stylo     10,622.39 -> 11,678.62
  4%  tp5o linux64 pgo e10s stylo                121.70 -> 126.24
  4%  tp5o linux64 opt e10s stylo                134.60 -> 139.38
  3%  tp5o_webext linux64 opt e10s stylo         208.93 -> 215.66
  3%  tp6_google linux64 opt 1_thread e10s stylo 401.62 -> 412.75
  3%  tp5o_scroll linux64 pgo e10s stylo         0.57 -> 0.58

Improvements:

  7%  tsvg_static linux64 pgo e10s stylo     53.60 -> 49.86
  7%  tsvg_static linux64 opt e10s stylo     54.42 -> 50.70
  6%  tscrollx linux64 opt e10s stylo        0.77 -> 0.72
  6%  tscrollx linux64 pgo e10s stylo        0.75 -> 0.70
  5%  displaylist_mutate linux64 pgo e10s stylo2,198.28 -> 2,092.59
  4%  displaylist_mutate linux64 opt e10s stylo2,426.40 -> 2,320.62


You can find links to graphs and comparison views for each of the above tests at: https://treeherder.mozilla.org/perf.html#/alerts?id=14702

On the page above you can see an alert for each affected platform as well as a link to a graph showing the history of scores for this test. There is also a link to a treeherder page showing the Talos jobs in a pushlog format.

To learn more about the regressing test(s), please see: https://wiki.mozilla.org/Buildbot/Talos/Tests

For information on reproducing and debugging the regression, either on try or locally, see: https://wiki.mozilla.org/Buildbot/Talos/Running

*** Please let us know your plans within 3 business days, or the offending patch(es) will be backed out! ***

Our wiki page outlines the common responses and expectations: https://wiki.mozilla.org/Buildbot/Talos/RegressionBugsHandling

Comment 1

[Ionuț Goldan [:igoldan]]

I'll soon post the Gecko profiles.

Comment 2

[Ionuț Goldan [:igoldan]]

Here are the Gecko profiles, from Linux 64bit:


For rasterflood_svg:

before: https://perf-html.io/from-url/https%3A%2F%2Fqueue.taskcluster.net%2Fv1%2Ftask%2FK0QsaGaXQ9eR-xijnG7vFw%2Fruns%2F0%2Fartifacts%2Fpublic%2Ftest_info%2Fprofile_rasterflood_svg.zip

after: https://perf-html.io/from-url/https%3A%2F%2Fqueue.taskcluster.net%2Fv1%2Ftask%2FZTgy7TdCQLme4G-pwmJG0w%2Fruns%2F0%2Fartifacts%2Fpublic%2Ftest_info%2Fprofile_rasterflood_svg.zip


For tp5o & tp5o responsiveness:

before: https://perf-html.io/from-url/https%3A%2F%2Fqueue.taskcluster.net%2Fv1%2Ftask%2FSg53X8fZSNmfnAXvsPFsNw%2Fruns%2F0%2Fartifacts%2Fpublic%2Ftest_info%2Fprofile_tp5o.zip

after: https://perf-html.io/from-url/https%3A%2F%2Fqueue.taskcluster.net%2Fv1%2Ftask%2FH9hqZCR3SA68BRKA5Gte2w%2Fruns%2F0%2Fartifacts%2Fpublic%2Ftest_info%2Fprofile_tp5o.zip


For tp6_google:

before: https://perf-html.io/from-url/https%3A%2F%2Fqueue.taskcluster.net%2Fv1%2Ftask%2FerRq94RxQNKFG0rHQk0U6A%2Fruns%2F0%2Fartifacts%2Fpublic%2Ftest_info%2Fprofile_tp6_google.zip

after: https://perf-html.io/from-url/https%3A%2F%2Fqueue.taskcluster.net%2Fv1%2Ftask%2FUhMbanqZRHyhDKpn9oeeAA%2Fruns%2F0%2Fartifacts%2Fpublic%2Ftest_info%2Fprofile_tp6_google.zip

Comment 3

[Ryan Hunt [:rhunt]]

This is the inverse of bug 1476964, because the change that caused those improvements and regressions got backed out due to a high crash rate. So this isn't exactly improvements or regressions, it's just us going back to our previous behaviour.

Comment 4

[Ionuț Goldan [:igoldan]]

(In reply to Ryan Hunt [:rhunt] from comment #3)
> This is the inverse of bug 1476964, because the change that caused those
> improvements and regressions got backed out due to a high crash rate. So
> this isn't exactly improvements or regressions, it's just us going back to
> our previous behaviour.

Oh, thanks for mentioning this.