Closed Bug 1754908 Opened 3 years ago Closed 3 years ago

DDG map panning and zooming has bad rendering performances on Firefox on Android

Categories

(Core :: Performance: Responsiveness, defect)

Firefox 98
Unspecified
Android
defect

Tracking

()

RESOLVED DUPLICATE of bug 1753471
Webcompat Priority P1
Performance Impact low
Tracking Status
firefox98 + fixed
firefox99 + fixed

People

(Reporter: karlcow, Unassigned)

References

()

Details

(Keywords: perf:responsiveness)

[Tracking Requested - why for this release]: DDG is an important web property more likely to be used by Firefox users.

​​### Basic information

Steps to Reproduce:

  1. Go to https://duckduckgo.com/?q=corse+&t=fpas&ia=about&iaxm=maps

Expected Results:
Smooth panning and zooming.

Actual Results:
When panning & zooming in Firefox, visual glitches can occur after giving only a few times of user inputs: tiles load in the wrong location, pinch-to-zoom snaps to one finger, creating a rubber band-like effect so the user is moved continents away. The experience is almost unusable.


Performance recording (profile)

Profile URL:
https://share.firefox.dev/3oDsTnS

System configuration:

Tested with:
Browser / Version: Firefox Nightly 99.0a1 (🦎 99.0a1-20220209095640)
Operating System: Google Pixel 5 (Android 12) - 1080 x 2340 pixels, 19.5:9 ratio (~432 ppi density), Samsung Galaxy S8 (Android 9) - 1440 x 2960 pixels, 18.5:9 ratio (~570 ppi density)

More information

Note: The issue is not reproducible on Chrome.

The profile shows recurrent TOO_MUCH_MALLOC

Tracking for our releases in flight as it affects a major site.

I can reproduce the janky behavior / worse-performance-than-Chrome when panning. I'm not seeing the visual glitches, but we should probably consider those a distinct bug from the perf issues anyway.

The profile shows lots of pretty-long requestAnimationFrame callbacks (100ms or more), and I think that slows down paints (we have to service the requestAnimationFrame before we paint, IIRC). e.g. here's a 96ms one:
https://share.firefox.dev/369Tnaf

It looks like it's spending a lot of time doing set HTMLImageElement.src... tnikkel, maybe you could take a look here?

Flags: needinfo?(tnikkel)
Whiteboard: [qf:p3:responsiveness]

I think the profile might be hitting bug 1753471. It was just fixed a day ago. Can you get a new profile with current nightly?

Flags: needinfo?(tnikkel) → needinfo?(kdubost)

That seems to be working indeed. There is no glitch on my side.
https://share.firefox.dev/3IfQfrg

Flags: needinfo?(kdubost) → needinfo?(tnikkel)

Okay, if it's fixed should we just resolve this then?

Flags: needinfo?(tnikkel)
Status: NEW → RESOLVED
Closed: 3 years ago
Resolution: --- → FIXED
Resolution: FIXED → DUPLICATE
Performance Impact: --- → P3
Whiteboard: [qf:p3:responsiveness]

The performance on Android has improved slightly, but it is not noticeable by much on a midrange device

The Windows performance has actually degraded, having a high system load and spinning fans. The performance is terrible and the graphical corruption is also awful. Half of the screen is filled with blocks that are misplaced, like a mosaic.

Please see https://bugzilla.mozilla.org/show_bug.cgi?id=1758987 for rendering corruption part of bug

Flags: needinfo?(kdubost)
Flags: needinfo?(kdubost)
See Also: → 1758987
Component: Performance: General → Performance: Responsiveness
You need to log in before you can comment on or make changes to this bug.