Page 2 of 4

Re: How many frames of lag are there with the following

Posted: 04 Sep 2017, 12:26
by drmcninja
Also, when I refer to the feel of the mouse, I don't mean the click latency-type of feel. But the movement of the cursor across the screen.

Re: How many frames of lag are there with the following

Posted: 04 Sep 2017, 13:00
by jorimt
@drmcninja
drmcninja wrote:Yeah, I've read it.

That's a strong placebo effect, considering my accuracy % is different between the two consistently. Could it just be down to better mouse feel due to more input from the mouse at the higher framerate? Even though they're both the same refresh rate, the game engine is taking mouse input twice as often with the other one (240fps vs. 120fps). I would have thought Fast Sync's buffer would make the mouse feel worse (since it's throwing out half the frames, right?), but it feels better.

In other words, the mouse at 240fps is different from 120fps. So with Fast-Sync, the game engine is doing the former (mouse input 240 frames a second), but the GPU is cutting it down to appear like the latter (120 fps hits the monitor screen). But the 120fps delivered to the display with Fast-Sync seem qualitatively different from the 120fps delivered to the display when the game engine is limited to no more than 120fps.

Just as how there's a difference between 240fps/120Hz and 120fps/120Hz (with no refresh sync). The mouse under Fast-Sync feels more like the former than the latter. Like somewhere inbetween the two.
You're posts are starting to go a little all over the place, so before we continue, what scenario are you specifically talking about now? Are you now only talking about the difference at 120Hz between Fast Sync at 120 FPS and Fast Sync at 240 FPS? Because that would require an entirely different explanation.

My last post was specifically addressing these comments:
Then I moved to 120Hz with Fast-Sync ON at 240-260fps (along with some computer upgrades so FPS wouldn't fluctuate), and that was amazing all around (on my laptop I use 120Hz, 118fps G-Sync On, V-Sync On, it also feels good, but not as good as Fast-Sync). 120Hz with Fast-Sync felt almost as good as uncapped/300fps, or just as good. The mouse felt like it had noticeably less input lag than 120Hz/118fps.
I thought you were saying that Fast Sync @120Hz vs. G-SYNC @120Hz/118 FPS felt more responsive?

Re: How many frames of lag are there with the following

Posted: 04 Sep 2017, 13:11
by RealNC
drmcninja wrote:I had to ask because I can actually notice a difference between V-Sync ON, 240.012Hz refresh rate and 240fps in-game cap versus V-Sync OFF, 300fps in-game cap.
In-game limiters are not accurate enough to give you 0.012FPS precision. You're going to have way more input lag compared to 300FPS since the frame limiter can't completely prevent vsync lag.

You would need to use RTSS for this, or cap well below 240FPS with the in-game limiter (like 237-ish) to make this work.

Re: How many frames of lag are there with the following

Posted: 04 Sep 2017, 13:22
by drmcninja
RealNC wrote:
drmcninja wrote:I had to ask because I can actually notice a difference between V-Sync ON, 240.012Hz refresh rate and 240fps in-game cap versus V-Sync OFF, 300fps in-game cap.
In-game limiters are not accurate enough to give you 0.012FPS precision. You're going to have way more input lag compared to 300FPS since the frame limiter can't completely prevent vsync lag.

You would need to use RTSS for this, or cap well below 240FPS with the in-game limiter (like 237-ish) to make this work.
Yup, I just installed RTSS as you posted this. I set it to 240.004 since my refresh rate seems consistently at 240.010 to 240.012. Does that seem reasonable? I'm not sure how to use RTSS but the in-game FPS thing is now steady at 240 with no alterations. I thought RTSS was supposed to have it's own FPS display but nothing appears while it's running.

Is it also better to enable V-Sync from Nvidia Control Panel or from within game?

Also, didn't RTSS itself add some latency? If it's a complicated answer, you don't have to go into detail, lol. I kind of glossed over all the RTSS/Frame Inspector sections of the articles until now.

Re: How many frames of lag are there with the following

Posted: 04 Sep 2017, 13:31
by drmcninja
jorimt wrote:You're posts are starting to go a little all over the place, so before we continue, what scenario are you specifically talking about now? Are you now only talking about the difference at 120Hz between Fast Sync at 120 FPS and Fast Sync at 240 FPS? Because that would require an entirely different explanation.

My last post was specifically addressing these comments:
Then I moved to 120Hz with Fast-Sync ON at 240-260fps (along with some computer upgrades so FPS wouldn't fluctuate), and that was amazing all around (on my laptop I use 120Hz, 118fps G-Sync On, V-Sync On, it also feels good, but not as good as Fast-Sync). 120Hz with Fast-Sync felt almost as good as uncapped/300fps, or just as good. The mouse felt like it had noticeably less input lag than 120Hz/118fps.
I thought you were saying that Fast Sync @120Hz vs. G-SYNC @120Hz/118 FPS felt more responsive?
Yeah, Fast-Sync (240fps@120Hz) felt more responsive (not in click latency, but in mouse cursor movement) than G-Sync/V-Sync on 118fps@120Hz by a bit and by a lot over V-Sync 118fps@120Hz. My aim was better, my mouse felt like I really was at 240fps, not 120fps (though it's only displaying 120fps, the engine is doing 240fps).

I guess gamers call that 'Mouse Lag', not Input Lag. It's why CS or Overwatch feel better at 300fps or higher than, say, even 200fps.

Re: How many frames of lag are there with the following

Posted: 04 Sep 2017, 13:37
by drmcninja
I also see now what you mean by actual input latency/lag. While messing around with V-Sync, whenever I enable it (and I haven't yet properly tested 240.004fps @ 240.012Hz), I can aim exactly the same except the cursor movement feels "slower" and if I rapidly aim and click very quickly, the click does not register over the spot I pressed it. I only started doing that after I went to V-Sync Off 300 FPS and realized I could react very quickly to certain things.

Otherwise, with V-Sync On, I'd aim just the same, but have to wait to be sure, visually, that the crosshair was over the target before clicking. Even if only for the slightest split second. In like 80% of in-game scenarios, that was quick enough. I even hit plenty of twitch shots this way. And ironically, I think it helped improve my aiming technique/discipline in general...

I know I wasn't getting full V-Sync lag because the cursor movement didn't feel bad and there was a marked difference with V-Sync On, 300 FPS which was crazy laggy all over the place.

Re: How many frames of lag are there with the following

Posted: 04 Sep 2017, 14:01
by jorimt
drmcninja wrote:Yeah, Fast-Sync (240fps@120Hz) felt more responsive (not in click latency, but in mouse cursor movement) than G-Sync/V-Sync on 118fps@120Hz by a bit and by a lot over V-Sync 118fps@120Hz. My aim was better, my mouse felt like I really was at 240fps, not 120fps (though it's only displaying 120fps, the engine is doing 240fps).
Okay, then my previous comments stand.

And Fast Sync doesn't work like that; just because it's reporting 240 FPS at 120Hz, doesn't mean it's showing you 240 updates a second. You're still only seeing up to 120 updates per second (the rest are dropped, which can cause microstutter), and slower than G-SYNC in those specific scenarios. And since you can only react to what you see being output on screen, mouse input response isn't going to change either; it's still 120 updates per second either way.

E.g. G-SYNC should be/feel more responsive.

Some of this could be explained by conditioning; we adapt/get used to certain lag levels (low or high), and our aim, timing, and anticipation of on-screen reaction adjusts accordingly. I know when I was testing at 240Hz/2000+ FPS (V-SYNC OFF), the lag levels were so low, I was having a hard time syncing my clicks to the response on screen; I was so used to reacting to much more lag that I obviously subconsciously compensated for it over time.

Re: How many frames of lag are there with the following

Posted: 04 Sep 2017, 14:15
by drmcninja
jorimt wrote:
drmcninja wrote:Yeah, Fast-Sync (240fps@120Hz) felt more responsive (not in click latency, but in mouse cursor movement) than G-Sync/V-Sync on 118fps@120Hz by a bit and by a lot over V-Sync 118fps@120Hz. My aim was better, my mouse felt like I really was at 240fps, not 120fps (though it's only displaying 120fps, the engine is doing 240fps).
Okay, then my previous comments stand.

And Fast Sync doesn't work like that; just because it's reporting 240 FPS at 120Hz, doesn't mean it's showing you 240 updates a second. You're still only seeing up to 120 updates per second (the rest are dropped, which can cause microstutter), and slower than G-SYNC in those specific scenarios. And since you can only react to what you see being output on screen, mouse input response isn't going to change either; it's still 120 updates per second either way.

E.g. G-SYNC should be/feel more responsive.

Some of this could be explained by conditioning; we adapt/get used to certain lag levels (low or high), and our aim, timing, and anticipation of on-screen reaction adjusts accordingly. I know when I was testing at 240Hz/2000+ FPS (V-SYNC OFF), the lag levels were so low, I was having a hard time syncing my clicks to the response on screen; I was so used to reacting to much more lag that I obviously subconsciously compensated for it over time.
By mouse lag here I don't mean click latency, I mean the actual positions of the mouse cursor on the screen.

So if you have 10 frames @ 10 Hz, you would have something like this:

Frame 1:
|.....X.....|

Frame 2:
|........X..|

But what I'm saying is, with 20 frames @ 10 Hz, you have less mouse "lag", you don't have the same mouse positions on frames as above, you might have:

Frame 1:
|.......X..|

Frame 2:
|.........X|

Same number of frames, but different content in those frames with regards to mouse cursor position.

The fact 300fps @ 120Hz, no V-Sync/no G-Sync feels different from 100fps @ 120Hz, no V-Sync/no G-Sync is the same difference in type of feeling I get when using Fast Sync. Like the movement of the mouse cursor feels different, less "behind".

Click latency might be slower, but that's different and the difference between G-Sync+V-Sync and Fast-Sync is negligible when compared with all sync settings turned off and FPS uncapped.

I played with 300fps / 120Hz, no sync for about a year. Then switched to 118fps/120Hz V-Sync/G-Sync (and sometimes just V-Sync on Desktop) for a few months. Then when I went to 120 Hz Fast-Sync (after upgrading computer to handle 240+fps consistently), it felt more like 300 fps / 120Hz again in terms of familiarity, muscle memory, and how I aim.

Re: How many frames of lag are there with the following

Posted: 04 Sep 2017, 14:33
by jorimt
I know you were referencing mouse movement/sensor response, but what I said still applies; Fast Sync is going to show the same updates in the same order as G-SYNC with equivalent settings, but less consistently (dropped frames/microstutter), and thus with more latency than G-SYNC in the scenarios we've been discussing.

Barring possible G-SYNC setup issues on your specific system (not working as intended), I'm not sure what else to tell you.

Re: How many frames of lag are there with the following

Posted: 04 Sep 2017, 17:24
by drmcninja
You can see users on Blizzard's forum also reporting that Fast Sync feels better than V-Sync/G-Sync. I don't know how, but Fast-Sync has a different effect on the mouse position input than expected.
I know you were referencing mouse movement/sensor response, but what I said still applies; Fast Sync is going to show the same updates in the same order as G-SYNC with equivalent settings, but less consistently (dropped frames/microstutter), and thus with more latency than G-SYNC in the scenarios we've been discussing.
That's the crux of the matter. I expected the same thing, that taking 240 fps / 120 Hz and cutting out half the frames should give an experience that is more stuttery (with regards to mouse cursor position input) than straight 118 fps / 120 Hz G-Sync/V-Sync.

Best case scenario, I figured it would feel the same. Because 120 fps would naturally feel like 240 fps with half the frames cut out anyway. So the mouse cursor/crosshair position inputs should be virtually the same.

Except the opposite happened. The mouse cursor position felt more responsive.

Maybe because Fast-Sync allows the GPU to render at near full speed, which is processing mouse input faster (and the system is getting input from the mouse at 1000Hz), so the frames that are being delivered to the display, 120 times a second, are just a tiny bit more up to date than when the GPU is delivering frames to the display only as it renders them 120 times a second. For each displayed frame, the GPU is rendering twice at least (sometimes more on my system). That's half the normal frametime (8.33->4.17ms). The end effect is the mouse cursor/crosshair movement feels like it's less "slow", which is the same feeling you get (but to a much greater extent) when turning off all sync-ing technologies or going from like 60 fps to 300 fps.

So I ran Overwatch with Fast-Sync on an i7-7700k @ 5.0GHz, with DDR4-3600 RAM, and a GTX 1080 and I have Render Scale set to below 100%, which results in the FPS being locked to not 240, but somewhere around 250-270 and closer to mid-260s consistently. It felt like a vague mouse lag had suddenly been lifted when I got rid of the 118 fps cap.

But that's just V-Sync. So Fast-Sync feels worlds better than V-Sync under the refresh rate, not that big of a surprise.

On my laptop with 120 Hz / 118 fps, G-Sync+V-Sync, it felt better than the desktop that had no G-Sync, but it did not feel as good as the Desktop with Fast-Sync (I even moved my display further away to mimic the same size of the laptop and used the same mouse/keyboard on the laptop, putting my keyboard on top of the laptop itself to mimic conditions as much as possible). So I enabled Fast-Sync, same settings, and the laptop had no problem hitting 240-260fps at the same settings (i7-6700HQ, DDR4-2666, GTX 1070), and it again felt a bit better.

The end effect of having 240 mouse inputs a second being reflected in your GPU's processed frames seems to be an increased granularity/fineness to the crosshair movement.

I just tried 240.004fps (RTSS limited) @ 240.011Hz V-Sync on. This should give 1 frame lag. Then I turned off V-Sync (same frame cap though). I could tell the difference, and that's also just 4.17ms. (Edit: The overall cursor movement was almost indistinguishable, just click latency and sharp twitch movements felt different... and if you immediately shut off V-Sync, you could briefly feel that the cursor was now feeling a little faster, but then you got used to it after a minute)