1446863 - Poor rendering performance on Desmos graphing calculator

Status: NEW

(Core :: Graphics, defect, P3)

Description

[bgstandaert]

User Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.12; rv:59.0) Gecko/20100101 Firefox/59.0
Build ID: 20180315233128

Steps to reproduce:

* Go to https://www.desmos.com/calculator
* Click on the gear icon in the upper-right, and click the spiderweb-looking icon in the popup to change the grid mode to polar.
* Attempt to zoom in and out on the graph, either by scrolling or by clicking the + and - buttons.


Actual results:

Firefox renders the graph at ~5fps, making the animation feel extremely sluggish. In comparison, Chrome runs at ~30fps on the same hardware, and Safari is ~60fps.


Expected results:

Firefox should render the graph as least as fast as other browsers.

Other information:
* This is happening on an Intel Iris 6100 GPU; disabling hardware acceleration actually seems to help a bit (although it's still slower than other browsers).
* This is happening on a high-resolution display, it gets less noticeable as you make the window smaller.
* This reproduces on 58, 59, and the latest nightly.

Comment 1

[bgstandaert]

> * Click on the gear icon in the upper-right, and click the spiderweb-looking

Sorry, the settings icon is actually a wrench.

Comment 2

[Cosmin Muntean [:cmuntean], Ecosystem QA]

User Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:61.0) Gecko/20100101 Firefox/61.0
Firefox: 61.0a1, Build ID: 20180321220044

I have managed to reproduce the issue on Windows 10 x64 on a 4k monitor with high resolution display. I have also tested this issue on a Mac Book Pro with retina displayed with 13" and also on a IMac but the issue is not reproducible. 
Indeed if the browser window is resized the issue is less noticeable. 

I have reproduce the issue on latest Firefox (59.0.2) release and also on latest Nightly (61.0a1) build.

Comment 3

[Ryan Hunt [:rhunt]]

I don't have access to a retina display so I'm having a hard time reproducing this.

If possible could you record a performance profile so we can see what's going on here? [1]

[1] https://developer.mozilla.org/en-US/docs/Mozilla/Performance/Reporting_a_Performance_Problem

Comment 4

[bgstandaert]

Sure, here you go: https://perfht.ml/2EeHUCl. This was recorded while I zoomed in and out by scrolling on a trackpad.

Comment 5

[Bas Schouten (:bas.schouten)]

I can't reproduce a real issue on my Thinkpad with D2D on here, at least for me the plain graph, or with a simple plotted function performs at least on 30+ FPS, mconley is also seeing smooth performance on OS X. For the moment I don't see evidence there's a 'flow issue' here.

Comment 6

[bgstandaert]

> I can't reproduce a real issue on my Thinkpad with D2D on here, at least for me the plain graph, or with a simple plotted function performs at least on 30+ FPS, mconley is also seeing smooth performance on OS X. For the moment I don't see evidence there's a 'flow issue' here.

Interesting, what resolution are you testing at? This issue goes away completely for me if I run Firefox in low-resolution mode or reduce the size of the browser window, so I think it's only an issue on a high-resolution display. Also, just to confirm, are you trying it with the polar graphing mode (gear icon > spiderweb-looking button under "grid")? The cartesian mode seems to be unaffected by this.

Otherwise, I'm not really sure what could be different - is there anything else I should try?

Comment 7

[alexandre_perso]

(In reply to bgstandaert from comment #6)

I can't reproduce a real issue on my Thinkpad with D2D on here, at least for me the plain graph, or with a simple plotted function performs at least on 30+ FPS, mconley is also seeing smooth performance on OS X. For the moment I don't see evidence there's a 'flow issue' here.

Interesting, what resolution are you testing at? This issue goes away
completely for me if I run Firefox in low-resolution mode or reduce the size
of the browser window, so I think it's only an issue on a high-resolution
display. Also, just to confirm, are you trying it with the polar graphing
mode (gear icon > spiderweb-looking button under "grid")? The cartesian mode
seems to be unaffected by this.

Otherwise, I'm not really sure what could be different - is there anything
else I should try?

I just wanted to say that I am still getting this issue! Exact same thing, 4k monitor at full screen is unusable, gets progressively better the more you scale it down. Turns out desmos3D is also much much slower than its Chromium counterpart. I am running windows 10 with a 1070ti. I am using graphics acceleration and webrender is turned on for compositing.

Comment 8

[alexandre_perso]

(In reply to alexandre_perso from comment #7)

(In reply to bgstandaert from comment #6)

I can't reproduce a real issue on my Thinkpad with D2D on here, at least for me the plain graph, or with a simple plotted function performs at least on 30+ FPS, mconley is also seeing smooth performance on OS X. For the moment I don't see evidence there's a 'flow issue' here.

Interesting, what resolution are you testing at? This issue goes away
completely for me if I run Firefox in low-resolution mode or reduce the size
of the browser window, so I think it's only an issue on a high-resolution
display. Also, just to confirm, are you trying it with the polar graphing
mode (gear icon > spiderweb-looking button under "grid")? The cartesian mode
seems to be unaffected by this.

Otherwise, I'm not really sure what could be different - is there anything
else I should try?

I just wanted to say that I am still getting this issue! Exact same thing, 4k monitor at full screen is unusable, gets progressively better the more you scale it down. Turns out desmos3D is also much much slower than its Chromium counterpart. I am running windows 10 with a 1070ti. I am using graphics acceleration and webrender is turned on for compositing. (Tested firefox 124 and 126 nightly)