Frame Time Fluctuation and FPS Limiters

[r3lik]

In the G-SYNC 101: External FPS Limiter article, it says that it's best to use an in-game FPS limiter if that option exists. However, in a few games (e.g. BF V), the in-game limiter (144fps) causes frame time to fluctuate by quite a bit. E.g. 5ms to 10+ms. This also occurs when I use in-game commands to cap at a lower number (138fps).

Now when I use RTSS to cap at 138fps, I get a smooth frame-time, albeit a bit higher at 7.2ms.

Are in game fps limiters not always the better option? Which nvidia low-latency setting should be used in these situations?

Thanks!

[jorimt]

In the G-SYNC 101: External FPS Limiter article, it says that it's best to use an in-game FPS limiter if that option exists. However, in a few games (e.g. BF V), the in-game limiter (144fps) causes frame time to fluctuate by quite a bit. E.g. 5ms to 10+ms. This also occurs when I use in-game commands to cap at a lower number (138fps).

With G-SYNC, these minor frametime fluctuations aren't anything to worry about unless they're visibly impacting your perceived game performance (stutter, etc).

Part of how good in-game limiters further reduce input lag over external limiters is allowing the very frametime fluctuations you are seeing.

As far as BF goes, the Frostbite FPS limiter is a "good" one; I've used it myself (BF series, Mirror's Edge Catalyst, etc).

Are in game fps limiters not always the better option?

It can depend on the game.

The only time you should really worry about in-game limiter performance, is if the limiter is causing a perceivable increase in stutter (rare, but possible), and/or the average FPS (not frametime) drifts far above your actual set limit. A good example of this is the CoD Warzone limiter. It fluctuates rapidly, and tends to hover multiple frames above your set limit.

The best way to decide between limiters for a given game is to try each of them.

I can however say the limiters featured in games like Overwatch, and in engines such as Source, UE3, UE4, Frostbite, CryEngine, and (sometimes even) Unity are all typically viable alternatives over external limiters for the lowest input lag possible.

Which nvidia low-latency setting should be used in these situations?

Short answer is LLM only has impact on input lag in GPU-bound situations (e.g. when your framerate is both not limited by your FPS cap and is causing the GPU to max out), and only by about 1 frame.

For G-SYNC, it's safe to leave LLM "On." The only known difference between it and LLM "Ultra" with G-SYNC is Ultra auto-caps the FPS in supported games.

For more info, see the three below posts (this has been discussed a lot):
viewtopic.php?f=5&t=5903&start=30#p44825
viewtopic.php?f=10&t=7317&p=55513#p55509
viewtopic.php?f=10&t=7317&p=55513#p55513

[r3lik]

Thanks a lot for the detailed response.

BF V actually feels a lot smoother to me when I use RTSS to cap fps, which gives me a consistent frame time of 7.2ms. It stays at that even with a lot of action, whereas the in-game limiter will result in frame time spikes to ~11ms and things start to feel sluggish.

[jorimt]

BF V actually feels a lot smoother to me when I use RTSS to cap fps, which gives me a consistent frame time of 7.2ms. It stays at that even with a lot of action, whereas the in-game limiter will result in frame time spikes to ~11ms and things start to feel sluggish.

Sure, whatever works best for you. And again, limiter choice and adjoining performance can be game (and even system) specific.

As for RTSS, you can't go wrong with it; it's highly consistent, and very reliable across games (and, typically, systems).