Closed Bug 528822 Opened 16 years ago Closed 16 years ago

Very slow painting, lots of memory traffic, on this testcase

Categories

(Core :: SVG, defect)

x86
macOS
defect
Not set
normal

Tracking

()

RESOLVED DUPLICATE of bug 523481

People

(Reporter: bzbarsky, Unassigned)

References

()

Details

Profiling the painting here, there's lots of argb32_image_mark_rgb32 called ultimately from _moz_cairo_paint_with_alpha, called from nsSVGUtils::PaintFrameWithEffects. There's also lots (16% of total time!) of __bzero called from _moz_cairo_quartz_surface_create, called ultimately from _moz_cairo_push_group_with_content called from nsSVGUtils::PaintFrameWithEffects. There's also 25% of the time spent in the OS vm_fault handler, presumably due to all the memory traffic that causes those __bzeros. Specifically, we're churning through about 6GB worth of cairo_quart_surfaces in the 10s of my profile. Any idea what's up here? The general symptoms look similar to bug 523298, but this is different code, for sure.
This testcase is actually just fine for me. It looks like it would have been fixed by bug 523481 (fixed October 24), is it possible that your build doesn't have that fix?
Possible, yes, depending on when I last build 1.9.2 branch (which is what I was looking in, I think). Let me rebuild and retest.
bug 528613 is about the same thing. I can't reproduce the 3.5.4 vs 3.5.5 claims there though.
Yep, updated build is fast.
Status: NEW → RESOLVED
Closed: 16 years ago
Resolution: --- → DUPLICATE
You need to log in before you can comment on or make changes to this bug.