[Webkit-unassigned] [Bug 230749] New: Poor WebGL gl.POINTS performance in Safari 15

bugzilla-daemon at webkit.org bugzilla-daemon at webkit.org
Fri Sep 24 05:52:48 PDT 2021


https://bugs.webkit.org/show_bug.cgi?id=230749

            Bug ID: 230749
           Summary: Poor WebGL gl.POINTS performance in Safari 15
           Product: WebKit
           Version: Other
          Hardware: Unspecified
                OS: Unspecified
            Status: NEW
          Severity: Normal
          Priority: P2
         Component: WebGL
          Assignee: webkit-unassigned at lists.webkit.org
          Reporter: ashley at scirra.com
                CC: dino at apple.com, kbr at google.com, kkinnunen at apple.com

Safari 15 appears to have a significant performance regression rendering points in WebGL relative to Safari 14. Perhaps this is related to the new Metal backend.

Demo URL: https://downloads.scirra.com/labs/particles/

On an iPad Pro with Safari 15, this starts off very slow, around 12 FPS. There are only a few hundred point sprites being rendered, which should be well within the capability of the hardware. It seems to gradually warm up and end up rendering at either 30 FPS or 60 FPS. The CPU measurement (based on timing how long is spent in JS execution) looks very high.

Conversely, an iPhone 7 Plus with iOS 14.7.1, and a Pixel 3 running Chrome for Android, both run the same demo at a steady 60 FPS right from the start. Safari 15 clearly has unexpectedly bad performance in this case.

In our engine, the particles are drawn with gl.drawArrays(gl.POINTS, ...). The GLSL references both gl_PointSize and gl_PointCoord. We have not found any other performance issues with any of our other WebGL content. I speculate that this points to a performance regression specifically with point rendering in Safari 15's new Metal backend.

-- 
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/20210924/bfcd83f2/attachment-0001.htm>


More information about the webkit-unassigned mailing list