904164 - Regression in eideiticker clock test on GN 8-10

Status: RESOLVED INCOMPLETE

(Firefox for Android Graveyard :: Toolbar, defect)

Description

[Wesley Johnston (:wesj)]

Looks like the clock test has regressed on galaxy nexus devices as of 8-10:

http://eideticker.wrla.ch/#/samsung-gn/clock/fps

All the skia-gl stuff going on makes me suspicious.

Comment 1

[James Willcox (:snorp) ([e-mail]) (he/him)]

The clock test is canvas, so it probably is due to SkiaGL. This isn't entirely unexpected, but I am a little surprised by the severity of the regression.

Comment 2

[James Willcox (:snorp) ([e-mail]) (he/him)]

Looking at the clock test, it calls getContext('2d') on every frame. I'm relatively certain that this causes us to create a new DrawTarget which is relatively expensive with SkiaGL. Can we change this so it just gets the context once?

Comment 3

[William Lachance (:wlach)]

Attachment: clockdiff.patch

Hmm, it seems that's not the problem.

Here's a patch that changes the clock test to not call getContext on every iteration. But even with it applied, the results I'm getting from the clock test are not changing (still at around 28.5 fps).

If you want to test this yourself, I did up a variant of this test which calculates fps via a frame counter:

http://people.mozilla.com/~wlachance/clock-fps.html

Comment 4

[BMO Automation]

We have completed our launch of our new Firefox on Android. The development of the new versions use GitHub for issue tracking. If the bug report still reproduces in a current version of [Firefox on Android nightly](https://play.google.com/store/apps/details?id=org.mozilla.fenix) an issue can be reported at the [Fenix GitHub project](https://github.com/mozilla-mobile/fenix/). If you want to discuss your report please use [Mozilla's chat](https://wiki.mozilla.org/Matrix#Connect_to_Matrix) server https://chat.mozilla.org and join the [#fenix](https://chat.mozilla.org/#/room/#fenix:mozilla.org) channel.