1604119 - 8.42% about_preferences_basic (windows10-64-shippable) regression on push ea818e25e302a2412f5fcf740727dbb42c4b3aa4 (Wed December 4 2019)

Status: RESOLVED WONTFIX

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

Description

[Florin Strugariu [:Bebe]]

Talos has detected a Firefox performance regression from push:

https://hg.mozilla.org/integration/autoland/pushloghtml?changeset=ea818e25e302a2412f5fcf740727dbb42c4b3aa4

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

Regressions:

8% about_preferences_basic windows10-64-shippable opt e10s stylo 134.24 -> 145.55

You can find links to graphs and comparison views for each of the above tests at: https://treeherder.mozilla.org/perf.html#/alerts?id=24261

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/TestEngineering/Performance/Talos

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

Comment 1

[Florin Strugariu [:Bebe]]

please see for more information.
https://bugzilla.mozilla.org/show_bug.cgi?id=1601950#c4

From the list of bugs we have in the regression range I think this is the one that caused the regression.
jmuizelaar can you please take a look and maybe confirm if your bug could have generated the regression.

Comment 2

[Glenn Watson [:gw]]

It seems that the configurations affected are not -qr configurations? This change only affects WebRender (and thus -qr configs), so I don't think it's relevant to this regression?

Comment 3

[Sotaro Ikeda [:sotaro]]

Yea, the change affect only to WebRender. It should not affect to non webrender test.

Comment 4

[Ryan VanderMeulen [:RyanVM]]

Is this a dupe of bug 1601950?

Comment 5

[Florin Strugariu [:Bebe]]

I think this is a different issue:
See:
https://bugzilla.mozilla.org/show_bug.cgi?id=1601950#c4

(In reply to :Gijs (he/him) from comment #4)

I'm a bit confused because

(In reply to Florin Strugariu [:Bebe] (needinfo me) from comment #0)

8% about_preferences_basic windows10-64-shippable opt e10s stylo 134.24 -> 145.55

this first regression, in the graph link is https://treeherder.mozilla.org/perf.html#/graphs?highlightAlerts=1&selected=1921202,604967,200.1247345861836,236.5533199818633,993357074&series=autoland,1921202,1,1&timerange=1209600&zoom=1575416176011,1575428918848,113.73193839866607,176.0067376248982 (zoomed in). There's a lot of noise in the graph, but actually, the beginning and end of that interval get me this pushlog:

https://hg.mozilla.org/integration/autoland/pushloghtml?fromchange=1a5918cfe128992841682258cdc35c71c785bd01&tochange=ea818e25e302a2412f5fcf740727dbb42c4b3aa4

which was 1-2 hours before the regression that happened on the QR builds ( https://treeherder.mozilla.org/perf.html#/graphs?highlightAlerts=1&selected=1924035,605061,770.2694962623855,146.88073430694192,993473599&series=autoland,1924035,1,1&series=mozilla-inbound,1948550,1,1&timerange=1209600&zoom=1575412311738,1575444931875,126.68653590463158,157.27599559914404 and https://treeherder.mozilla.org/perf.html#/graphs?highlightAlerts=1&selected=1924035,605061,569.128138773506,77.58729211133762,993473599&series=autoland,1924035,1,1&series=mozilla-inbound,1948550,1,1&timerange=1209600&zoom=1575408583723,1575462639949,127.41026724257418,149.31541249972992 ). The QR builds definitely point to bug 1593060 and don't have nearly as much noise. But I suspect the non-QR regression is a different regression on the same test.

Florin, do you want to set up a separate bug for that?

Comment 6

[Julien Cristau [:jcristau]]

Too late for 72 now.

Comment 7

[Ryan VanderMeulen [:RyanVM]]

This looks inactionable to me. Looking at the graphs, this is a really noisy test. The Win64 graph doesn't show much change in the last 60 days from what I can see, and the Win32 graph does show a change in baseline when this was filed which went away about a week ago. I'm having a hard time seeing any obvious regression on mozilla-beta either when Gecko 73 merged there last week.

Feel free to reopen the bug if there's something I'm missing, though :-)