[Webkit-unassigned] [Bug 90721] Don't destroy the decoded data of an image if WebKit is about to render the image.

bugzilla-daemon at webkit.org bugzilla-daemon at webkit.org
Fri Oct 5 14:38:42 PDT 2012


https://bugs.webkit.org/show_bug.cgi?id=90721





--- Comment #26 from Andreas Kling <kling at webkit.org>  2012-10-05 14:39:10 PST ---
(In reply to comment #25)
> (In reply to comment #24)
> > This change regressed Apple's Membuster benchmark by ~20% (80MB.)
> > 
> > Do you have any idea what may be causing this? Otherwise we should roll it out while we figure out the cause of the regression.
> 
> I'm sorry for this regression.
> 
> I don't know what you mean by "Apple's Membuster benchmark by ~20% (80MB.)"
> I'm guessing that some of Apple's tests have a lot of images on the viewport, because the above patch does not remove decoded data of images when they lay on the viewport.
> 
> I need more information to know exactly why this patch caused the regression.

You can see the problem by opening a bunch of pages in tabs in Safari, then sending the low-memory signal like so:

notifyutil -p org.WebKit.lowMemory

With your change, there's significantly more memory used now, as the images "visible" in background tabs don't appear to get evicted.

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.


More information about the webkit-unassigned mailing list