1880487 - 13.73 - 13.61% perf_reftest_singletons bidi-resolution-1.html / perf_reftest_singletons bidi-resolution-1.html (Linux) regression on Tue February 6 2024

Status: RESOLVED WONTFIX

(Core :: Internationalization, defect)

Description

[Andrej (:aglavic)]

Perfherder has detected a talos performance regression from push 187caa38d3490df8b32dbeaf8cce22e6fdcc634a. 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)
14%	perf_reftest_singletons bidi-resolution-1.html	linux1804-64-shippable-qr	e10s fission stylo webrender	181.18 -> 206.07
14%	perf_reftest_singletons bidi-resolution-1.html	linux1804-64-shippable-qr	e10s fission stylo webrender	180.69 -> 205.29

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 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.

You can run these tests on try with ./mach try perf --alert 41482

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 1824671

Comment 2

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

Set release status flags based on info from the regressing bug 1824671

Comment 4

[Jonathan Kew [:jfkthame]]

The unicode-bidi update in bug 1881487 should help a bit here, although it won't fully mitigate the issue.

Comment 5

[Dianna Smith [:diannaS]]

ty! Looks like it helped on linux, only slightly on bug 1880484.
do we plan on uplifting bug 1881487 to 124 or let it ride the trains?
Any further mitigations here?

Comment 6

[Jonathan Kew [:jfkthame]]

For 124, I propose we should uplift bug 1881232, which will switch us back to the old bidi implementation for Beta/Release builds; that'll give us space to continue investigations on Nightly to see if we can close the remaining gap.

Comment 7

[Dianna Smith [:diannaS]]

thanks Jonathan! bug 1881232 has been uplifted and the perf regressions have been marked as disabled for 124

Comment 8

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

The severity field is not set for this bug.
:m_kato, could you have a look please?

For more information, please visit BugBot documentation.

Comment 9

[Makoto Kato [:m_kato]]

Actually, this is disabled on release channel

Comment 10

[Jonathan Kew [:jfkthame]]

I think we've done what we can here for the present, with the several optimizations that landed in the dependent bugs. Recent mozilla-central runs show perf_reftest_singletons back to mid-190s levels, which means we've regained somewhere towards half of the original regression. What's left is pretty much imperceptible in real-world scenarios (e.g. wikipedia page-load tests), and IMO we should accept it as a reasonable trade-off for being able to move forward with the migration away from ICU4C.

So this is partly-WFM but I'll close as WONTFIX given that we haven't entirely closed the gap. (Probably at least some of the remaining difference arises because unicode-bidi actually follows the Unicode bidi spec more precisely, and this sometimes results in creating more distinct level runs.)