1323991 - 4.1 - 58.27% tp5o_scroll / tscrollx (linux64) regression on push 6e02d48ef1226c6f8c7504c05dac113794d65f07 (Thu Dec 15 2016)

Status: RESOLVED FIXED

(Core :: Graphics, defect)

Description

[Joel Maher ( :jmaher ) (UTC -8)]

Talos has detected a Firefox performance regression from push 6e02d48ef1226c6f8c7504c05dac113794d65f07. As author of one of the patches included in that push, we need your help to address this regression.

Regressions:

 58%  tp5o_scroll summary linux64 opt e10s     2.81 -> 4.45
 58%  tp5o_scroll summary linux64 pgo e10s     2.78 -> 4.38
 56%  tp5o_scroll summary linux64 pgo          3.9 -> 6.08
 47%  tscrollx summary linux64 pgo             3.78 -> 5.56
 44%  tscrollx summary linux64 opt             4.12 -> 5.94
 33%  tp5o_scroll summary linux64 opt          4.99 -> 6.64
  7%  tscrollx summary linux64 opt e10s        3.71 -> 3.96
  4%  tscrollx summary linux64 pgo e10s        3.8 -> 3.95


You can find links to graphs and comparison views for each of the above tests at: https://treeherder.mozilla.org/perf.html#/alerts?id=4521

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

[Joel Maher ( :jmaher ) (UTC -8)]

In looking at this we have a lot of improvements and some regressions specifically in the scroll tests.  I know the GL changes typically trigger talos regressions, are these specific regressions expected?

Comment 2

[Lee Salzman [:lsalzman]]

(In reply to Joel Maher ( :jmaher) from comment #1)
> In looking at this we have a lot of improvements and some regressions
> specifically in the scroll tests.  I know the GL changes typically trigger
> talos regressions, are these specific regressions expected?

They are expected, yes. This is just a backout of a patch that caused the inverse of these changes at bug 1297004.

Since this was a nightly-only experiment, this movement in the numbers was just illusory, and not affecting any of the other branches than central.

Seeing as GL acceleration just causing continual breakage, and this just puts us back where we were before the experiment, we are going to accept any possible regression arising from this and don't believe it warrants further investigation.

Comment 3

[Joel Maher ( :jmaher ) (UTC -8)]

as this is a backout with no other steps to enable this or do more work, I am going to resolve this.