1506851 - Firefox slow on HiDPI monitor until moved to a low res monitor (and back)

Status: RESOLVED WORKSFORME

(Core :: Widget: Cocoa, defect, P3)

Description

[Robin Whittleton]

User Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.13; rv:65.0) Gecko/20100101 Firefox/65.0

Steps to reproduce:

Start Firefox on a HiDPI monitor (in my specific case the laptop screen of a 15" Macbook Pro 2017, and usually a restart after the next version of Nightly).




Actual results:

Observe that the response of the browser is pretty sluggish (noticeable in chrome animations, scrolling of content etc).


Expected results:

Expected normal speed of the browser.

It’s possible to get this speed back by moving the window to a non-HiDPI screen (in my case a Dell monitor connected via DisplayLink). Once the window has been moved to this monitor then moving it back to the laptop screen retains the normal performance.

If it helps debugging then happy to supply performance logs etc. I’ve tested this with WebRender both on and off and got the same behaviour both times. I don’t have a regression range but it’s been happening for a few weeks at least. I did install new DisplayLink drivers a few weeks ago and upgraded to 10.13.6 at the same time but can’t remember whether there’s a link there.

Comment 1

[Adrian Florinescu [:aflorinescu]]

Robin, could you please confirm if the issue is reproducible when you only use the HiDPI? Or does it only happen when you have the secondary screen set-up?

Comment 2

[Robin Whittleton]

Just tested and yes, it’s fine when restarted while the DisplayLink screen is unplugged. It’s also fine if I restart the browser while the window is on the DisplayLink screen. The behaviour only appears if it’s restarted on the internal screen while the DisplayLink screen is attached.

The DisplayLink device is (according to system report) a Dell Universal Dock D6000. The display attached to it is a Dell EliteDisplay E232 which is reported as being 1920 x 1080. The DisplayLink drivers I’m using are the latest for High Sierra: 4.3.1.

Comment 3

[Adrian Florinescu [:aflorinescu]]

Unfortunately, I do not have the hardware around to test and confirm the issue, but I would guess that the problem is somewhere around the DisplayLink drivers, but not necessarily reduced to that. Also, I would guess again that this issue is reproducible with 63 and 64 as well, but it wouldn't hurt to check if you can provide additional support in investigating this.

It would probably also be interesting to see if you could test with an another browser, let's say Chrome and/or Safari, to see if the problem is reproducible?

Meanwhile, let's initially triage this issue to Core::Widget:Cocoa.

Comment 4

[Stephen A Pohl [:spohl]]

Could you run mozregression[1] to see if this is a regression in Firefox? If you have never run mozregression before, simply run these three commands in a Terminal window:

sudo easy_install pip
sudo pip2 install -U mozregression --ignore-installed
mozregression

A number of Firefox versions will open in succession to narrow down when this started occurring. Simply type "good" or "bad" in Terminal based on whether or not a build reproduces the bug.

[1] https://mozilla.github.io/mozregression/

Comment 5

[Robin Whittleton]

Apologies for late reply. Unfortunately my laptop has terminally died and I can’t connect to the system until I get a new one sorted. I’ll get back to. This as soon as it’s replaced.

Comment 6

[Stephen A Pohl [:spohl]]

Thanks for getting back to us, Robin. I'm closing this bug until we hear back from you.