Frame Rate Limiting/Streaming with Lightboost?

[K3D]

Few questions:
1. In-Game Frame Limiter or NVIDIA Inspector Profile Frame Limiter for lower input lag?
2. Lightboost while streaming at 60 FPS to Twitch.TV? How will this work?

[Chief Blur Buster]

Few questions:
1. In-Game Frame Limiter or NVIDIA Inspector Profile Frame Limiter for lower input lag?

In Game fps limiters are best for lowest lag. Driver based frame rate limiters often increase input lag.

2. Lightboost while streaming at 60 FPS to Twitch.TV? How will this work?

They are independent of each other so you can use both. LightBoost is a monitor feature that works independently of the GPU. The local frame rate can be higher than the twitch streamed frame rate regardless of LightBoost or otherwise.

[K3D]

Few questions:
1. In-Game Frame Limiter or NVIDIA Inspector Profile Frame Limiter for lower input lag?

In Game fps limiters are best for lowest lag. Driver based frame rate limiters often increase input lag.

2. Lightboost while streaming at 60 FPS to Twitch.TV? How will this work?

They are independent of each other so you can use both. LightBoost is a monitor feature that works independently of the GPU. The local frame rate can be higher than the twitch streamed frame rate regardless of LightBoost or otherwise.

The in-game frame limiter has three settings:
High FPS Option 1 - 110-130 FPS fluctuations
High FPS Option 2 - 115-150 FPS fluctuations
Benchmark - 300-400 FPS fluctuations

Which one would you personally go with? How do fluctuations below and above 120 FPS affect Lightboost?

Also, would you go with Displayport over DVI for the VG248QE? Any recommended companies for Displayport cables?

[Chief Blur Buster]

The in-game frame limiter has three settings:
High FPS Option 1 - 110-130 FPS fluctuations
High FPS Option 2 - 115-150 FPS fluctuations
Benchmark - 300-400 FPS fluctuations

Which one would you personally go with? How do fluctuations below and above 120 FPS affect Lightboost?

Also, would you go with Displayport over DVI for the VG248QE? Any recommended companies for Displayport cables?

I have no particular preference of DVI and DisplayPort, though DisplayPort provides an official route to higher bandwidth that DVI currently doesn't. But that currently doesn't matter at 1080p 144Hz(and under) at this time.

For the frame rate fluctuations:
Short Answer: Try all the common frame rate settings and see what looks/feels the best.
Long Answer: If you prefer VSYNC OFF, having VSYNC OFF at 300-400fps will likely look smoother than VSYNC OFF 110-130fps. This is because of VSYNC OFF microstutters, and microstutter vibration amplitude is equal to the step between frames (or 1/fps of the total movement distance along motion vector over a one second period). (see microstutter mathematics in Area 51). Thus, even 400fps@120Hz looks visually smoother than 200fps@120Hz, due to less microstutters (less aliasing / beatfrequency effects between framerate and refreshrate). There are exceptions -- where it may not always be the case; e.g. if framepacing is much worse at 400fps than it is at <200fps, or if there's more input lag at higher framerates (with certain game engines; especially multithreaded, where the keyboard/mouse input reading thread gets starved when the rendering thread is all maxed out). So it's important to test it out, to see how it looks/feels.

[RealNC]

Hm, is there a way to calculate the "best" framerate cap with vsync off for a certain monitor refresh rate? With that I mean a framerate that would result in the least noticeable beatfrequency effects.

[Chief Blur Buster]

Hm, is there a way to calculate the "best" framerate cap with vsync off for a certain monitor refresh rate? With that I mean a framerate that would result in the least noticeable beatfrequency effects.

There's really no "best", as it is often subjective. Some people are sensitive to the microstutters, while others are sensitive to the tearing, so it's often a matter of personal preference.

fps_max 3fps less than Hz = beat frequency of 3 microstutters per second, faster-rolling tearline moving downwards
fps_max 2fps less than Hz = beat frequency of 2 microstutters per second, fast-rolling tearline moving downwards
fps_max 1fps less than Hz = beat frequency of 1 microstutters per second, slow-rolling tearline moving downwards
fps_max same as Hz = stationary/vibrating/slow moving tearline effect (non-stationary = timing imprecisions)
fps_max 1fps more than Hz = beat frequency of 1 microstutters per second, slow-rolling tearline moving upwards
fps_max 2fps more than Hz = beat frequency of 2 microstutters per second, fast-rolling tearline moving upwards
fps_max 3fps more than Hz = beat frequency of 3 microstutters per second, faster-rolling tearline moving upwards
etc.

fps_max much further away from Hz (e.g. fps_max 117 at 144Hz) will just microstutter continually (rapidly-vibrating edge effect), so if you are prioritizing on smoothness of VSYNC OFF without the beat-frequency side effects, then it is often best to uncap the framerate, since 400fps@144Hz visibly has less microstutters than 200fps@144Hz, since the microstutter vibration amplitude is equal to the step distance between frames during refresh-rate-unsynchronized situations (e.g. VSYNC OFF or triple buffering). Since the movement step between frames are only half as much at 400fps than at 200fps, the microstutter edge-vibration effect is half the amplitude (microstutter math). We're of course, excluding other sources of microstutters (e.g. mouse limitations, engine limitations). However, on very fast systems with very fast GPU's, the microstutter difference between 200fps and 400fps is quite clearly noticeable with smooth pans - e.g. keyboard left/right pans at close-distances from high-detail objects. This is why some VSYNC OFF afficanados sometimes like to cap at ultra high frame rates, if the GPU horsepower permits. Some configurations did get more input lag if you uncap the framerate (e.g. starving the game-input reading thread, when the graphics rendering thread is running in overdrive), but that does not always happen (engine-dependant/config dependant), and you also always cap somewhere below becoming CPU-limited.

For simplicity, and GPU power limitations, frame rate caps a few fps away from Hz, have been common among CS:GO players. (e.g. fps_max 142 or 143 during 144Hz).

You could also do game-engine-based adaptive VSYNC (not driver-based) with just-in-time rendering, to achieve similar latency to VSYNC OFF. Basically fps_max equalling Hz, with some intelligent logic to steer the tearline algorithmically towards the top edge or bottom edge of the screen (aka VSYNC OFF with tear-line steering towards edge of screen). Problem is, it's so extremely difficult to do with slightly variable frame rates with equal lag of VSYNC OFF. (Until GSYNC came along). Some interesting stuff in ezQuake: Just-in-Time VSYNC, for ultra-low-latency VSYNC ON algorithms that so few games seem to do.