Page 1 of 1

Test ufo animation time graph with firefox not complete flat?

Posted: 02 Sep 2025, 12:08
by wiseude
After uninstalling/installing firefox I've noticed this test is not as smooth as it used to be.The only extension I'm using is ublock origin.
Firefox
Image

Chrome
Image

I even tried to refresh firefox 1 more time but it;s staying like this.It's acting as if hardware acceleration is disabled when it's actually enabled (on firefox)

Or is this normal for firefox?

Re: Test ufo animation time graph with firefox not complete flat?

Posted: 14 Sep 2025, 18:27
by Chief Blur Buster
wiseude wrote:
02 Sep 2025, 12:08
Or is this normal for firefox?
It is generally normal on FireFox because FireFox use bigger Meltdown/Spectre javascript clock fuzzing, 1ms (FireFox) versus 0.1ms (Chrome) -- a security method to make things difficult for hackers. So those 1ms spikes are normal on FireFox and Safari.

That being said, there's a workaround with a higher-security TestUFO that unlocks a more precise clock, in terms of HTML5 APIs available to web developers.

The upcoming TestUFO 2.2 upgrades the clock source (the best possible security sandbox JavaScript can give permits a higher precision clock source) -- with 5 microsecond on Chrome and 20 microsecond on FireFox/Safari. This will hide that 1ms noise.

Re: Test ufo animation time graph with firefox not complete flat?

Posted: 21 Sep 2025, 10:16
by wiseude
Chief Blur Buster wrote:
14 Sep 2025, 18:27
wiseude wrote:
02 Sep 2025, 12:08
Or is this normal for firefox?
It is generally normal on FireFox because FireFox use bigger Meltdown/Spectre javascript clock fuzzing, 1ms (FireFox) versus 0.1ms (Chrome) -- a security method to make things difficult for hackers. So those 1ms spikes are normal on FireFox and Safari.

That being said, there's a workaround with a higher-security TestUFO that unlocks a more precise clock, in terms of HTML5 APIs available to web developers.

The upcoming TestUFO 2.2 upgrades the clock source (the best possible security sandbox JavaScript can give permits a higher precision clock source) -- with 5 microsecond on Chrome and 20 microsecond on FireFox/Safari. This will hide that 1ms noise.
Out of curiosity.How accurate is the visual frame skipping indicator?The magenta box on the bottom right I mean.

Because when I have it enabled to see if there's any frame skipping I do notice some graphic flickering going on constantly(some subtle and some not).Yet the graph still stays within 1,-1ms
Image

Is this normal or should the magenta box stay a full on magenta without any graphical flickering in the magenta box?This is on firefox.I've also tried it on edge.Same thing.

Re: Test ufo animation time graph with firefox not complete flat?

Posted: 21 Sep 2025, 13:37
by Chief Blur Buster
wiseude wrote:
21 Sep 2025, 10:16
Out of curiosity.How accurate is the visual frame skipping indicator?The magenta box on the bottom right I mean.
It's very accurate once you know how to use it. It's a benchmark for browser and OS compositor performance. Sometimes OS compositors will frameskip even when the browser think it's not, so that's why it was added. It commonly happens with multimonitor systems, for example.

Ignore the regular flicker (that's designed to do that), and only focus on the erratic flickers in the VIsual Frameskipping Indicator.

1. If red spikes syncs with the erratic flickers (and the bottom part turned yellow/red), the browser successfully self-detected the frame skip.

2. If no red spikes occured (and the bottom part stayed green) while you saw an erratic flicker, the browser did not self-detect the frame skip.

Item 2 is a bigger cause for concern than item 1.

Re: Test ufo animation time graph with firefox not complete flat?

Posted: 22 Sep 2025, 10:18
by wiseude
Chief Blur Buster wrote:
21 Sep 2025, 13:37
wiseude wrote:
21 Sep 2025, 10:16
Out of curiosity.How accurate is the visual frame skipping indicator?The magenta box on the bottom right I mean.
It's very accurate once you know how to use it. It's a benchmark for browser and OS compositor performance. Sometimes OS compositors will frameskip even when the browser think it's not, so that's why it was added. It commonly happens with multimonitor systems, for example.

Ignore the regular flicker (that's designed to do that), and only focus on the erratic flickers in the VIsual Frameskipping Indicator.

1. If red spikes syncs with the erratic flickers (and the bottom part turned yellow/red), the browser successfully self-detected the frame skip.

2. If no red spikes occured (and the bottom part stayed green) while you saw an erratic flicker, the browser did not self-detect the frame skip.

Item 2 is a bigger cause for concern than item 1.
So I should be focusing on the graph itself then instead of the magenta box?
Green spikes=good (Basically what it looks like in the picture I showed)
Yellow/red spikes=bad
Even if the Magenta box flickers?

Re: Test ufo animation time graph with firefox not complete flat?

Posted: 23 Sep 2025, 22:00
by Chief Blur Buster
wiseude wrote:
22 Sep 2025, 10:18
So I should be focusing on the graph itself then instead of the magenta box?
Green spikes=good (Basically what it looks like in the picture I showed)
Yellow/red spikes=bad
Even if the Magenta box flickers?
To be clear:

Steady flicker + no red spikes = good

Concurrent (spikes + erratic flicker) = reliable browser detection
EXAMPLE: Background software (and over a hundred other causes)

Erratic flickers but only green graph = bad and browser couldn't detect it, harder to diagnose.
EXAMPLE: Display frameskipping (and over a hundred other causes)

Steady flicker is normal. And things like display-side frameskipping can cause "Erratic flickers without red spikes". Red spikes will often concurrently cause erratic flickers.

Be noted, there's many thousands of possible causes, many not a cause for concern (e.g. simple browser performance) and others cause for concern (e.g. too much background software). Thusly, I cannot tell you the cause, except possible examples. However, common cause is background software, and sometimes some of them is bad (certain brands of RGB software adds a lot of spikes). Another common cause is different-Hz multimonitor.

www.testufo.com/animationtimegraph is simply a tool to help you troubleshoot your system. Browsers are MUCH more sensitive to stutters than games, so if you can de-stutter the browser, you've peeled off a layer of the stutter onion (The many causes of stutters).