1738928 - 21.22 - 6.23% perf_reftest_singletons bloom-basic-2.html / perf_reftest_singletons id-getter-2.html + 4 more (OSX) regression on Thu October 28 2021

Status: RESOLVED INVALID

(Core :: Graphics: CanvasWebGL, defect)

Description

[Andra Esanu (needinfo me)]

Perfherder has detected a talos performance regression from push 909433910d0d03229dd23cc430d08d6326b44a95. As author of one of the patches included in that push, we need your help to address this regression.

Regressions:

Ratio	Test	Platform	Options	Absolute values (old vs new)
21%	perf_reftest_singletons bloom-basic-2.html	macosx1015-64-shippable-qr	e10s stylo webrender	38.18 -> 46.29
21%	perf_reftest_singletons bloom-basic.html	macosx1015-64-shippable-qr	e10s stylo webrender	38.14 -> 46.23
20%	perf_reftest_singletons bloom-basic-2.html	macosx1015-64-shippable-qr	e10s fission stylo webrender	38.06 -> 45.50
18%	perf_reftest_singletons bloom-basic.html	macosx1015-64-shippable-qr	e10s fission stylo webrender	38.59 -> 45.62
15%	perf_reftest_singletons bloom-basic-2.html	macosx1015-64-shippable-qr	e10s fission stylo webrender	39.84 -> 45.92
6%	perf_reftest_singletons id-getter-2.html	macosx1015-64-shippable-qr	e10s fission stylo webrender	890.04 -> 945.52

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

[Kelsey Gilbert [:jgilbert]]

This was almost certainly not bug 1738125.
This is the second unlikely report I've gotten in the last week. How sure are we that the cset narrowing code is accurate?

Comment 2

[Kelsey Gilbert [:jgilbert]]

Seems spurious:
https://treeherder.mozilla.org/perfherder/graphs?timerange=1209600&series=autoland,3869898,1,1

Comment 3

[Gian-Carlo Pascutto [:gcp]]

Looking at the graphs, there's a bunch of results that are in a blob all over the place, and then there's a new baseline that's higher. No idea what happened during the "blob".