[Webkit-unassigned] [Bug 248223] [GTK] Significant graphics performance regressions

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=248223

--- Comment #6 from kdwkleung at protonmail.com ---
(In reply to Michael Catanzaro from comment #1)
> So in summary, it's still way faster than GTK 3 was, but a huge regression
> from previous GTK 4 MotionMark results.
> 
> Would be useful to confirm that this regression was introduced by 2.39.1.
> You could test 2.38.2 by downgrading org.gnome.Platform and
> org.gnome.Epiphany.Devel to builds from, say, two days ago, following the
> instructions at
> https://github.com/flatpak/flatpak/wiki/Tips-&-Tricks#downgrading. That way
> we can be certain that the regression is due to 2.38.2 -> 2.39.1, and not
> something else that changed in the meantime, like a change in Epiphany or
> GTK itself.
> 
> BTW you mentioned on Matrix that you suspect the regression is caused by the
> switch to ANGLE for WebGL. I think that is mandatory now, but if not, maybe
> I could temporarily disable it to allow you to take 2.39.1 benchmarks with
> it turned off....

I run the benchmark pretty much every time WebKitGTK updates, and it’s still reporting near 500 in 2.38.1/2.

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.webkit.org/pipermail/webkit-unassigned/attachments/20221123/61dac435/attachment.htm>