1677764 - 4.08 - 6.71% google-maps / google-maps fcp / youtube fcp (android-hw-g5-7-0-arm7-api-16-shippable) regression on push 16291a47acda8ffac58ea92a73f403b531640f11 (Fri November 13 2020)

Status: RESOLVED FIXED

(Core :: DOM: Core & HTML, defect)

Description

[Alexandru Ionescu (needinfo me) [:alexandrui]]

Perfherder has detected a browsertime performance regression from push 16291a47acda8ffac58ea92a73f403b531640f11. As author of one of the patches included in that push, we need your help to address this regression.

Regressions:

Ratio	Suite	Test	Platform	Options	Absolute values (old vs new)
7%	google-maps		android-hw-g5-7-0-arm7-api-16-shippable	warm webrender	886.13 -> 945.54
6%	google-maps	fcp	android-hw-g5-7-0-arm7-api-16-shippable	warm webrender	799.38 -> 849.79
4%	google-maps	fcp	android-hw-g5-7-0-arm7-api-16-shippable	warm webrender	815.48 -> 851.21
4%	youtube	fcp	android-hw-g5-7-0-arm7-api-16-shippable	warm webrender	704.29 -> 733.04

Improvements:

Ratio	Suite	Test	Platform	Options	Absolute values (old vs new)
19%	web-de	LastVisualChange	android-hw-p2-8-0-android-aarch64-shippable	warm webrender	921.38 -> 749.00

Details of the alert can be found in the alert summary, including links to graphs and comparisons for each of the affected tests. Please follow our guide to handling regression bugs and let us know your plans within 3 business days, or the offending patch(es) will be backed out in accordance with our regression policy.

For more information on performance sheriffing please see our FAQ.

Comment 1

[Steve Fink [:sfink] [:s:]]

Hm. If this is real, then it should be backed out, since it was intended as a refactoring. It does change the mechanism used for scheduling CC slices, and I don't fully understand the priority scheme involved there, so I could see how it's possible that this might change behavior. Especially on Android, where I would expect less idle time. (Note that I was trying to do this in a way that would not have any behavioral difference at all, other than creating one object instead of two.)

Comment 2

[Olli Pettay [:smaug][[e-mail]]]

Could you push a backout patch to try and see what the numbers look like?

Comment 3

[Steve Fink [:sfink] [:s:]]

I pushed, and confirmed the regression. I also found a bug; I'm testing the fix now.

I would like this patch to be backed out, on central and beta.

Comment 4

[Steve Fink [:sfink] [:s:]]

Bug 1662254 comment 32:

Backout by [e-mail]:
https://hg.mozilla.org/mozilla-central/rev/66a0f9b68c8d
Backed out changeset 16291a47acda for causing Bug 1677764. a=backout DONTBUILD

Comment 5

[Ryan VanderMeulen [:RyanVM]]

And Beta84:
https://hg.mozilla.org/releases/mozilla-beta/rev/814fd3f7d9d65147a4d432f3948679ab165f3dd8

Comment 6

[Florin Strugariu [:Bebe]]

Backout alert:
== Change summary for alert #27817 (as of Mon, 23 Nov 2020 09:22:59 GMT) ==

Improvements:

Ratio	Suite	Test	Platform	Options	Absolute values (old vs new)
16%	reddit		android-hw-g5-7-0-arm7-api-16-shippable	nocondprof warm webrender	921.47 -> 777.17
14%	reddit		android-hw-g5-7-0-arm7-api-16-shippable	nocondprof warm webrender	904.32 -> 781.86

For up to date results, see: https://treeherder.mozilla.org/perfherder/alerts?id=27817

Comment 7

[Florin Strugariu [:Bebe]]

== Change summary for alert #27795 (as of Sat, 21 Nov 2020 20:18:48 GMT) ==

Improvements:

Ratio	Suite	Test	Platform	Options	Absolute values (old vs new)
4%	raptor-sunspider-firefox		linux64-shippable		217.77 -> 209.19
3%	raptor-sunspider-firefox		linux64-shippable-qr	webrender	222.37 -> 216.21

For up to date results, see: https://treeherder.mozilla.org/perfherder/alerts?id=27795