[Webkit-unassigned] [Bug 276959] New: powerPreference "high-performance" causes bad performance

bugzilla-daemon at webkit.org bugzilla-daemon at webkit.org
Tue Jul 23 08:37:34 PDT 2024


https://bugs.webkit.org/show_bug.cgi?id=276959

            Bug ID: 276959
           Summary: powerPreference "high-performance" causes bad
                    performance
           Product: WebKit
           Version: Safari 17
          Hardware: iPhone / iPad
                OS: iOS 17
            Status: NEW
          Severity: Normal
          Priority: P2
         Component: WebGL
          Assignee: webkit-unassigned at lists.webkit.org
          Reporter: kipashmobile at gmail.com
                CC: djg at apple.com, kbr at google.com, kkinnunen at apple.com

## Description
In a three.js scene containing skinned meshes, having `powerPreference` set to `"high-performance"` leads to 20fps. With `powerPreference` set to `"default"`, the performance is significantly better, capping at 120fps.

## Reproduction Steps
1. On iPhone 12, Safari 17.5.1:
2. Open this live build: https://weak-navy-leptoceratops.glitch.me
3. Observe the FPS counter showing 20fps.
4. Remix the Glitch project: https://glitch.com/edit/#!/remix/weak-navy-leptoceratops?path=main.js%3A1%3A0
5. Comment out `main.js at 33` -> won't use antialiasing.
6. Click "Preview" (bottom bar) -> 3 dots -> Copy link -> open on iPhone
7. Observe high framerates (120fps) with `powerPreference` set to `"high-performance"` and antialiasing off.
8. Comment out `main.js at 30` -> will use `"default"` power preference.

## Expected Result
- Performance for this scene should not be under 40fps on iPhone 12.
- Disabling antialiasing should not result in a 6x performance gain (likely more, frame rate is capped at 120fps).

## Actual Result
- `powerPreference` set to `"high-performance"` is causing worse performance.
- Antialiasing is surprisingly costly while `powerPreference` is set to `"high-performance"`.

## Reproduced In
- Reproduced on iPhone 12 in Safari (17.5.1) and Chrome (126.0.6478.153).

## Works In
- Not reproduced on Android Samsung S9+ in Chrome (126.0.6478.122) (90-110fps).
- Not reproduced on Windows 10 in Chrome (126.0.6478.182) (288fps), i7-10700KF, RTX 3070.

## Additional context:
Stats from Inspector.js on Windows:
- draw calls: 42
- triangle count: 3 095 064

-- 
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/20240723/faf92357/attachment-0001.htm>


More information about the webkit-unassigned mailing list