Blur Busters Forums

Posted: **29 Mar 2023, 08:19**

With common ingame framerate limiters such as NVCP or ingame, the FPS becomes more erratic the higher the FPS target.
With such inconsistent frame times you could have latency hits greater than what your trying to achieve with low latency.

With an erratic framecap you could be hit in a few ways:

If to low the drop in frames would cause latency, if to high it could add frames to the buffer if using V-sync or add GPU workload if your looking for that <95% GPU usage low latency.

To show this is frametimes:

You could get latency changes of in a range of 6ms @200FPS and possibly even more so under load.
This latency range will only increase as you increase the FPS target so assuming there is 1 frame latency in RTSS it becomes less of an issue the higher the FPS target.

Thoughts?

Posted: **29 Mar 2023, 09:13**

Tiberiusmoon wrote: ↑
29 Mar 2023, 08:19
With common ingame framerate limiters such as NVCP or ingame, the FPS becomes more erratic the higher the FPS target.
With such inconsistent frame times you could have latency hits greater than what your trying to achieve with low latency.

How steady an FPS limiter is isn't always just on the limiter itself, but, in part, on the capabilities of the system and the given game being run; if I ran identical tests on my system in the same game with the same limiters, we may see different (possibly even less) frametime variances in the NVCP limiter scenario.

I.E. the reason you're probably seeing more erratic frametimes with non-RTSS limiters the higher the framerate, is because your particular system is becoming less and less able to maintain a consistent average.

And when you say "NVCP" limiter, you mean the "Max Frame Rate," setting, and not the legacy limiting solution still available in Nvidia Inspector, correct?

Because if so, it may mean that sometime between the release of "Max Frame Rate" and now, Nvidia might have switched from limiting by frametime, and is instead setting an average framerate target (letting frametimes run free), which would result in less steady overall frametime performance, but potentially less latency.

For instance, in your frametime graph, nearly half the time the Nvidia limiter is allowing faster than 5ms (~200 FPS average) frametimes (~4-3ms), potentially resulting in faster frame delivery over RTSS, which is part of the reason most in-game limiters are lower latency.

Posted: **29 Mar 2023, 09:28**

jorimt wrote: ↑
29 Mar 2023, 09:13
How steady an FPS limiter is isn't always just on the limiter itself, but, in part, on the capabilities of the system and the given game being run; if I ran identical tests on my system in the same game with the same limiters, we may see different (possibly even less) frametime variances in the NVCP limiter scenario.

I.E. the reason you're probably seeing more erratic frametimes with non-RTSS limiters the higher the framerate, is because your particular system is becoming less and less able to maintain a consistent average.

And when you say "NVCP" limiter, you mean the "Max Frame Rate," setting, and not the legacy limiting solution still available in Nvidia Inspector, correct?

Because if so, it may mean that sometime between the release of "Max Frame Rate" and now, Nvidia might have switched from limiting by frametime, and is instead setting an average framerate target (letting frametimes run free), which would result in less steady overall frametime performance, but potentially less latency.

For instance, in your frametime graph, nearly half the time the Nvidia limiter is allowing faster than 5ms (~200 FPS average) frametimes (~4-3ms), potentially resulting in faster frame delivery over RTSS, which is part of the reason most in-game limiters are lower latency.

Even in non GPU bound situations?

Yeah max frame rate.
I mean sure a little less latency can be good but if its not consistent whats the point?

In regards to Blurbuster's low latency V-sync setup would this not add more frames to the buffer than is intended at higher FPS?

Posted: **29 Mar 2023, 10:00**

Tiberiusmoon wrote: ↑
29 Mar 2023, 09:28
Even in non GPU bound situations?

Yes

Tiberiusmoon wrote: ↑
29 Mar 2023, 09:28
Yeah max frame rate.
I mean sure a little less latency can be good but if its not consistent whats the point?

Assuming the in-game or NVCP limiter in question isn't causing visible recurring stutter, and we're talking about using G-SYNC or no-sync, then whether you use one of them or RTSS can be situational/up to preference.

Tiberiusmoon wrote: ↑
29 Mar 2023, 09:28
In regards to Blurbuster's low latency V-sync setup would this not add more frames to the buffer than is intended at higher FPS?

Low latency standalone V-SYNC via a decimal-level RTSS limit is an exception; you need the stability of RTSS in that particular case.

And if you resort to standalone V-SYNC and don't use a decimal-level RTSS limit where the framerate can exceed the refresh rate, a limiter set above the refresh rate isn't going to do anything anyway, since V-SYNC will already throttle each frame to the max scanout time of the current physical refresh rate.

I was more specifically referring to G-SYNC + framerates within the refresh rate or no-sync + framerates above the refresh rate (what I assume your graphs are showing), where reduced latency via an in-game (or a non-frametime-limiting) limiter is concerned, in which case, frametime variances aren't always as important, so long as they remain within a certain threshold, and again, don't cause visible recurring stutter.

Posted: **29 Mar 2023, 10:37**

Ahh so aslong as it dose not cause cause stutter then the FPS can be a variable as it likes?
As this shows:

I will need to do more testing. . .

Posted: **29 Mar 2023, 11:46**

Tiberiusmoon wrote: ↑
29 Mar 2023, 10:37
Ahh so aslong as it dose not cause cause stutter then the FPS can be a variable as it likes?

Long story short, in a certain context, "yes."

Where frametime performance is concerned, the cost of more consistency typically comes down to higher absolute latency, and visa-versa. I.E. while normalization eliminates the extreme min and max values, that means it will also eliminate whatever latency reduction benefit those outliers might have allowed (in key scenarios) otherwise.

Basically, assuming we're not talking straight-out stutter vs. less stutter, there's no one-size-fits-all "correct" answer when it comes to frametime performance; it's almost entirely contextual and situational, and will depend on a variety of factors (system specs, game, current physical refresh rate, target framerate + achievable framerate/frametime performance, sync scenario, etc) and the ultimate end goal of your particular use-case (even from game-to-game).

But yes, typically, RTSS set at a limit your average framerate can be sustained at 99% of the time is going to achieve the most consistent average frametime/latency experience.

Posted: **29 Mar 2023, 12:20**

Sweet got it.
Thank you for the insight!

Posted: **08 May 2023, 10:38**

jorimt wrote: ↑
29 Mar 2023, 11:46

Tiberiusmoon wrote: ↑
29 Mar 2023, 10:37
Ahh so aslong as it dose not cause cause stutter then the FPS can be a variable as it likes?
But yes, typically, RTSS set at a limit your average framerate can be sustained at 99% of the time is going to achieve the most consistent average frametime/latency experience.

So RTSS will be the "best" Limiter nowadays? I only tested Limiters in CSGO and the NVCP Limiter provide me much more 0.1% fps than max_fps command at same value. Sry for aksing but i tryed RTSS only one time but compaired to other tools that monitor fps values t was very confusing and it doesnt shows me the values that way i wanted them. Therefore i never tryed it again

Posted: **10 May 2023, 08:52**

SachsenPowl wrote: ↑
08 May 2023, 10:38
So RTSS will be the "best" Limiter nowadays?

"Best," in the sense that it potentially has the most stable average frametime performance in the majority of scenarios, but this can vary depending on the game, and whether your average framerate can be consistently sustained at your set limit (whenever the framerate is below the set limit, limiters aren't doing anything).

Blur Busters Forums

Do you think less erratic FPS overcomes RTSS latency hit?

Do you think less erratic FPS overcomes RTSS latency hit?

Re: Do you think less erratic FPS overcomes RTSS latency hit?

Re: Do you think less erratic FPS overcomes RTSS latency hit?

Re: Do you think less erratic FPS overcomes RTSS latency hit?

Re: Do you think less erratic FPS overcomes RTSS latency hit?

Re: Do you think less erratic FPS overcomes RTSS latency hit?

Re: Do you think less erratic FPS overcomes RTSS latency hit?

Re: Do you think less erratic FPS overcomes RTSS latency hit?

Re: Do you think less erratic FPS overcomes RTSS latency hit?