1780390 - 5.02% displaylist_mutate (OSX) regression on Fri July 15 2022

Status: RESOLVED FIXED

(Core :: Graphics: WebRender, defect, P5)

Description

[Andrej (:aglavic)]

Perfherder has detected a talos performance regression from push 941414f045ab7b3d535a5d5776e845f34d808956. 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)
5%	displaylist_mutate	macosx1015-64-shippable-qr	e10s fission stylo webrender-sw	1,788.33 -> 1,878.09

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 1779387

Comment 2

[Glenn Watson [:gw] | PTO until 29th April]

This change is fixing a bug, but it's also a temporary fix only - once https://phabricator.services.mozilla.com/D151987 lands next week, this patch will be removed and replaced by the optimized path above. So let's wait and see what happens to dl_mutate once that lands.

Comment 3

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

Set release status flags based on info from the regressing bug 1779387

Comment 4

[Hsin-Yi Tsai (she/her) [:hsinyi]]

Hi Glenn,
Now that bug 1779952 landed and the comment showed improvements, can we say this regression is fixed, even the data here and there were on a slightly different options, i.e. webrender-sw v.s. webrender?

Comment 5

[Glenn Watson [:gw] | PTO until 29th April]

Yes, I think that's reasonable - I've looked at a profile of this test with the new code, and there's no obvious hot spots that are issues in the new clip-tree implementation now. I'll still be doing further clip optimization work that will improve this test in future too.

Comment 6

[Hsin-Yi Tsai (she/her) [:hsinyi]]

(In reply to Glenn Watson [:gw] from comment #5)

Yes, I think that's reasonable - I've looked at a profile of this test with the new code, and there's no obvious hot spots that are issues in the new clip-tree implementation now. I'll still be doing further clip optimization work that will improve this test in future too.

Hey Glenn, thanks for confirming! :) And ... sorry for another follow-up question - does it make sense to uplift bug 1779952 to Fx104? The 5% regression on Fx104 looks still obvious, but I am also wondering the risk of uplift that refactor patch. Thank you again!

Comment 7

[Glenn Watson [:gw] | PTO until 29th April]

I think it's OK to not uplift it - the regression time is only on a stress test case that's unlikely to be representative of any real world content.