Blur Busters Forums

Chief Blur Buster wrote:External frame rate limiters work, but internal framerate limiters (fps_max) will have less input lag. Most people will not notice, except for highly attuned competitive game players sensitive to input lag differentials of just a few miliseconds.

I'd like to see you do one of those red led button tests with Crysis 3 and then do 3 different input lag tests with engine, RSS and nvidia driver. I'm willing to bet driver will be laggy and RSS/engine will be almost the same.

It's true on laggy engines, that the differences will get lost in the 'noise'. Games like Crysis 3 has many tens of milliseconds of lag, and lots of lag variability, so internal vs external framerate capping (At least at 144Hz = 1/144sec = 6.9ms per frame), that few milliseconds can get lost in the noise of 80-millisecond button-to-pixels input lag in a high-lag game like Battlefield 4 or Crysis 3.

But, for low-lag engines such as Counterstrike and Quake Live, it makes a massive difference. During a 180 degree flicks where the whole screen pans in 1/2 second (4000 pixels per second), a single millisecond equals 4 pixels off-aiming, and 10 milliseconds equals 40 pixels off-aiming. In fast action gaming, mere milliseconds matter, even if you don't feel it except for "whoops, I seem to be aiming more poorly" effects. Especially if you've trained for a specific latency, and you're not used to the change in input latency. The more predictable lag of these engines, means you more easily get used to a specific lag behaviour, and if the lag changes, then in the paid/sponsored professional gaming leagues - aiming can be thrown off by mere milliseconds, until you re-train for the new lag. (Then again, one would rather not have any lag to begin with!)

I will eventually do lag tests (high speed video method, button-to-pixels) of the various methods over time.

A question a bit on the side here: Will capping the framerate low in any game cause any unwanted effects in G-SYNC mode? In Skyrim for instance, you have to cap it at 60 or less. And I would want to cap it to about 85 in Far Cry 3, because that game scorches my graphics cards if I use higher framerates.

Is there a way to keep it on BF4? Prior to getting my ROG Swift I capped it @ 60 fps since my monitors refresh rate was 60Hz. The way I was taught to do it was disable V-Sync in the game and create a user.cfg file with notepad and add a gametime.maxfps 60 or something along those lines. Don't remember the code, but after I got my monitor I deleted that file and the game ran fine with GSync the first time, but now when I launch BF4 it starts @ 60 fps and in the game settings it says it's on 2560x1440@144Hz.

The trick I learned to get it to kick out of that is by changing it to 2560x1440@120Hz, click change, then when it says if I want to keep those changes I tell it no and it flashes back and now I'm running @ true 144. Sometimes it fails and the display loses connection with a "No signal" message and I have to press the restart button on my PC.

Any suggestions?

From my experience with the ROG Swift, once I enabled G-sync it automatically set my MAX FPS to whatever hz I set the monitor to. It doesn't look like you need any of these applications to limit your FPS.

Turboman750 wrote:From my experience with the ROG Swift, once I enabled G-sync it automatically set my MAX FPS to whatever hz I set the monitor to. It doesn't look like you need any of these applications to limit your FPS.

When you hit the monitor's maxixmum refresh rate, g-sync falls back to double buffered v-sync. The problem is that when your framerate hits the max refresh, buffers fill up and wait for the next stage of the pipeline to open up. This means you get up to an extra frame of latency for each step down the pipeline you push the framerate limiter. If your framerate is limited by the game engine(more specifically, right before it checks for user input), then by the time each step of the pipeline has finished working on a frame, the next step is ready to start working on it.

Another thing to note is that Riva Statistics Server / MSI Afterburner is the most accurate frame limiter. It works within 0.1 fps for me while fps_max and driver limiting are generally off by about 3 fps @ 120 Hz.

I would like to chime back into this topic. I've used G-sync on both the ROG Swift and the new ACER 27 inch 1080p g-sync monitor. On both monitors it appears the g-sync driver automatically limits the GPU's FPS to "1" under the cap. So if I'm set to 144hz refresh rate, I don't exceed 143 FPS in game.

I haven't seen any input lag at max FPS that would indicate a need to use 3rd party software to limit your frame rate.

If I'm wrong has anybody conducted a test recently that disproves me?

It would seem really silly for Nvidia to implement this technology to replace v-sync and have the driver enable v-sync at max fps.

Turboman750 wrote:If I'm wrong has anybody conducted a test recently that disproves me?

It would seem really silly for Nvidia to implement this technology to replace v-sync and have the driver enable v-sync at max fps.

The tests have been done and indicate there is input lag.

Also, yes I notice the frame rate seems to be slightly lower than refresh, as you report, as if the drivers are using "frame rate limiter" to automatically cap slightly lower than refresh (though it does not do this with any other refresh other than 144hz, as far as I'm aware).

I also notice nvidia driver level frame rate capping introduces input lag in itself, in some cases, even if you run with no vsync or no gsync (the manual type forced from nvidia inspector. Which is also currently mostly broken, as far as I am aware, since it often disables after a certain time). Perhaps this is what the problem currently is?

fenderjaguar wrote:

Turboman750 wrote:If I'm wrong has anybody conducted a test recently that disproves me?

It would seem really silly for Nvidia to implement this technology to replace v-sync and have the driver enable v-sync at max fps.

The tests have been done and indicate there is input lag.

Also, yes I notice the frame rate seems to be slightly lower than refresh, as you report, as if the drivers are using "frame rate limiter" to automatically cap slightly lower than refresh (though it does not do this with any other refresh other than 144hz, as far as I'm aware).

I also notice nvidia driver level frame rate capping introduces input lag in itself, in some cases, even if you run with no vsync or no gsync (the manual type forced from nvidia inspector. Which is also currently mostly broken, as far as I am aware, since it often disables after a certain time). Perhaps this is what the problem currently is?

Okay, I'm sure Nvidia will address this soon if they havn't already.

So in the mean time, what are those who are using G-sync recommending? 144 hz with Rivatuner capping at 120fps?

or are you capping FPS using the internal engine?

I noticed that the imput lag test was done at 144hz refresh rate and received input lag at 143 max fps, but not at 120 max fps. If i set my monitor to 120hz instead of 144hz, is there still input lag? Has this been tested?

Based off of the input lag test it seems the G-sync cap is somewhere between 120hz and 144hz, so if I set the monitor to 120hz, G-sync should be enabled 100% of the time.