Display frametime vs input latency.

[MT_]

Hi, few questions;

1. If a 120hz screen has a frame gap of 8.33ms, doesn't that mean when testing button to pixel latency, that the average response time solely for this aspect is actually 4.16ms?

2. And doesn't the same hold true for 1000hz mouse polling? The reaction can be anywhere from 0 to 1ms? So an average of 0.5ms response?

3. A bit unrelated, but if I run a game at 240fps on a 120hz screen, and lets assume the fps limiter used is incredibly accurate (tear lines pretty much on top or bottom (invisible) with a perfect 8.333ms frametime, is there still a benefit of using 240fps over 120fps?

Thank you.

[Sparky]

Depends on what you're measuring, because the whole screen doesn't update at the same instant. V-sync on vs off, or a single point on the screen vs anywhere on the screen.

You won't see that 4.16ms if you're talking about vsync off and are looking for a change anywhere on the screen. You will see it if you're looking at the top of the screen with v-sync on, or any single point with v-sync off. You'll see more than that if you look at a single point that's not the top of the screen with v-sync on.

It also won't be exactly 4.16ms, it will depend on how fast the scanout is vs blanking interval.

2: yes. Though that's only counting the usb polling interval latency, there are other sources.

3: the distance from the tearline down to your point of interest corresponds to the latency.

[MT_]

Thanks for the reply!

I'm measuring with 240fps camera with very first mouse movement (Even just a millimeter) vs any response on my screen, anywhere, doesn't matter what spot. In which case shouldn't the average that i would be observing 4.16ms?

In which case my tests seem to be fairly accurate;

My Monitor:
Pixel Response Time Element
3.45
Estimated Signal Processing Lag
0.35

+ 0.5ms average usb response = 4.80ms.

Assuming my assumption is right, 4.16ms average + 4.80 = 8.96ms.

My results so far have been around 8.5 to 9.7ms of input latency on 1709, and earlier testing showed 1809 in the range of 13~14ms.
I'm actually mostly testing different Windows versions and corresponding settings. (Game mode, etc)

Allegedly 170x builds reduce latency by ~5ms with game mode on (Before they changed up how game mode works), but so far I've been seeing no evidence of this and its basically just having 170x installed is enough vs other builds.

[Sparky]

With v-sync off you shouldn't be getting any input latency due to refresh rate, because you're not waiting for the next refresh to display the new frame. You will get latency due to framerate though. 240hz camera is also a little slow for this. 8 to 14ms is pretty good if we're talking about an actual game, but kind of slow if we're talking about a test program running at thousands of frames per second.