1829288 - 3.26 - 3.1% tresize / tresize (Linux) regression on Mon April 17 2023

Status: RESOLVED FIXED

(Core :: Graphics: WebRender, defect)

Description

[Acasandrei Beatrice (needinfo me)]

Perfherder has detected a talos performance regression from push 1c58e8743220c7cab1ee952fa3d3a5ba046575cc. 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)
3%	tresize	linux1804-64-shippable-qr	e10s fission stylo webrender-sw	26.35 -> 27.21
3%	tresize	linux1804-64-shippable-qr	e10s fission stylo webrender-sw	26.27 -> 27.09

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) may be backed out in accordance with our regression policy.

If you need the profiling jobs you can trigger them yourself from treeherder job view or ask a sheriff to do that for you.

For more information on performance sheriffing please see our FAQ.

Comment 1

[BugBot [:suhaib / :marco/ :calixte]]

Set release status flags based on info from the regressing bug 1823578

Comment 2

[Ashley Hale [:ahale]]

This patch is part of a major refactor patches for clipmask handling and are expected to yield performance gains, so this regression is surprising, it could be that the SWGL performance is worse on this case and we need to look into optimizing in that case.

The author of this patch will be out of office for the next week so I don't expect any action to be taken by the graphics team in that time range.

While I did review the patch in question I am not sure I am the best suited to figure out why performance has regressed on this new approach, so I will bring this up in our weekly triage meeting in a few days, I am not sure if that is an adequate timeline for dealing with a regression though.

Comment 3

[Glenn Watson [:gw]]

This should be temporarily resolved when bug #1830691 lands. I'll work on a proper fix for this before re-enabling the new clip-mask path.