Gsync does not work for me

[Glide]

Well I have just swapped out three GPUs when doing some testing, and I would say that the rule for double-buffered V-Sync is generally true.

Here's a short framerate graph from Afterburner which shows this behavior:


At the points where the framerate drops to 30 FPS - which is not instant in this case - the GPU load may only be something like 60%.
So it's capping the performance to sync up with the refresh rate instead of allowing the framerate to stay totally unlocked.

If the framerate was unlocked with adaptive v-sync, triple-buffering, or g-sync, GPU load would be stuck at 99% and the framerate might be something like 35-55 FPS instead of a steady 30.

However it seems like there is some sort of prediction going on to provide some degree of hysteresis.
So you do occasionally get moments where the game is running in the 40-50 fps range.

I was using Half-Life 2 for most of the testing, but the handful of games I checked today all showed similar behavior.
I'm sure that I have seen some games where the graph did just look like a square wave, but that may have either been before I increased Afterburner's polling rate, or it may just be how some games implement V-Sync.

I thought this behavior might have been something which changed recently, or perhaps on newer GPUs due to NVIDIA enabling windowed G-Sync, but even rolling back to some drivers that were several years old (the reason I swapped out GPUs) or killing the DWM process (even though I was testing in Full-screen) I was still seeing this behavior.

It's anything but smooth when the games are running like this however.
At the point where it is locked to 30 or 60 it is as smooth as those framerates can be, but the transitions stutter worse than triple-buffering, since it is trying to stay at either 30 or 60.

If you use adaptive v-sync or triple-buffering and monitor both the framerate and GPU load, the average framerate will be much higher than this V-Sync behavior. And G-Sync should show exactly that as well.
Only G-Sync does this without the game stuttering, and without latency or tearing.

HARDOCP show this behavior with their adaptive v-sync article, comparing the three states of V-Sync. (on/off/adaptive)
Triple-buffering should have identical performance to the adaptive graphs, without the tearing.
G-Sync should have identical performance to the V-Sync off graphs, only without tearing or juddering because it's synchronizing the refresh rate to the framerate.

You can see that V-Sync on generally runs at either 30 or 60, but there are still points at which it hits values in-between for a brief moment.

[Lanarchy]

Ok, I've taken a video to show you.

First, a screenshot of my Nvidia Control Panel settings (as you can see, FIXED REFRESH and VSYNC ON NO ADAPTIVE):



And now the video:

phpBB [video]

BOOM ! No drop to 30 ! So, what now ?

[Joolsyz]

Not sure what you're trying to prove, from my experience many games are not double buffered now a days anyway as most have triple buffering. Pretty sure AC4: Black Flag was double buffered mind, give that a whirl if you have it.

[Lanarchy]

Joolsyz what I am trying to prove is that Gsync seems to not be doing anything more than what Vsync is already doing. If most games nowadays are not double-buffered, and Vsync eliminates tearing, then what else is Gsync supposed to be doing ?

If, using Vsync, I don't get tearing and frame drops to 30, then... what else is Gsync good for ? I have done tests in many games and the experience is the exact same whether I'm on Vsync or Gsync. And it is not because I'm not sensitive to stuff, because I am plus I asked my roommates to watch and they came to the same conclusion. There is no difference with Gsync or Vsync.

As for AC4, yeah I know that one does. So Gsync is only for AC4 ?

Don't get me wrong here, I would love to have the same WOW effect everyone seems to be having with Gsync. I just don't see it. If needed, I can make videos of 5 mins of gameplay in any game I have using both and you will see no difference at all.

[Joolsyz]

Out of interest do you mainly use a mouse for most games? As for AC4, G-sync doesn't even fix that game, it's a jittery mess with crappy frame times no matter how i play it.

[Glide]

BOOM ! No drop to 30 ! So, what now ?

Well I don't play Guild Wars 2, so I can't say what is going on.
It's possible that the game's "V-Sync" option is enabling triple buffering instead.
What happens if you disable the in-game option, but force V-Sync via the control panel?

Joolsyz what I am trying to prove is that Gsync seems to not be doing anything more than what Vsync is already doing. If most games nowadays are not double-buffered, and Vsync eliminates tearing, then what else is Gsync supposed to be doing ?

I think you're missing the point of G-Sync if all you thought it was meant to do was unlock the framerate.
The main advantage is that it synchronizes the refresh rate to the framerate, and does so without adding latency.

If you are using triple buffering to eliminate tearing in DirectX games, you have a minimum of three frames of latency. At say 40 FPS that would be 75ms.
With G-Sync you eliminate the tearing with only 1ms latency.

And framerates which are not a factor of the refresh rate cannot be displayed smoothly on a standard display.

The best test for this is to play an older game that you can easily hit 60+ FPS. Now cap the game to something in-between 30 and 60, like 55 FPS.
Use a controller, or map the keyboard to control the camera. Rotate the camera slowly at a fixed speed. (controllers with the sensitivity at the minimum are best for this)
Without G-Sync, this will constantly judder, because 55 doesn't fit evenly into 60.
With G-Sync, 55 FPS is displayed at 55Hz and should be completely smooth.

[Lanarchy]

Well I don't play Guild Wars 2, so I can't say what is going on.
It's possible that the game's "V-Sync" option is enabling triple buffering instead.
What happens if you disable the in-game option, but force V-Sync via the control panel?

Same. I showed in the video that I was getting fps between 30 and 60 even before activating in-game Vsync.

If you are using triple buffering to eliminate tearing in DirectX games, you have a minimum of three frames of latency. At say 40 FPS that would be 75ms.
With G-Sync you eliminate the tearing with only 1ms latency.

And framerates which are not a factor of the refresh rate cannot be displayed smoothly on a standard display.

The best test for this is to play an older game that you can easily hit 60+ FPS. Now cap the game to something in-between 30 and 60, like 55 FPS.
Use a controller, or map the keyboard to control the camera. Rotate the camera slowly at a fixed speed. (controllers with the sensitivity at the minimum are best for this)
Without G-Sync, this will constantly judder, because 55 doesn't fit evenly into 60.
With G-Sync, 55 FPS is displayed at 55Hz and should be completely smooth.

Yes, I can confirm this. I have tried with Bioshock Infinite, locked the framerate to 50, and strafed. It was doing little skips all the time with Vsync, but with Gsync it was smooth. But, do I need to 50-lock games to see Gsync actually working ? I mean, people have been so amazed with Gsync, I don't think they all 50-locked their games to experience it. I know it's looking like I just want to hate on Gsync but quite the opposite, trust me. I want to feel the greatness, like everyone else. But I don't see it. Guild Wars 2 isn't the only one. I can do videos with Battlefield 4, Shadow of Mordor, World of Warcraft, Borderlands 2, Elder Scrolls Online, Final Fantasy XIV, Metro Last Light, Titanfall, and plenty other. In all of those games, I did the test. With Vsync, I'm getting framerates in between 30 and 60.

I wished my games were all dropping to 30 as soon as they can't maintain 60 with Vsync, then I would sing the praises of Gsync. But...they don't.

[Glide]

Yes, I can confirm this. I have tried with Bioshock Infinite, locked the framerate to 50, and strafed. It was doing little skips all the time with Vsync, but with Gsync it was smooth. But, do I need to 50-lock games to see Gsync actually working ?

No, the point is that G-Sync is always this smooth, even when the framerate is variable.
The reason I used the example of locking the framerate to 50 FPS is because:
a) Most people seem to think that games running at 50+ FPS should be smooth regardless of your display.
b) 50 FPS intentionally doesn't "fit" into 60Hz. (not a factor of the refresh rate)

Now if you repeated the same test at 60 FPS, or at 30 FPS, you would see that the motion would be perfectly smooth on the standard display as well. (well, as smooth as 30 FPS can be)
The point is that you get the same smoothness with a G-Sync display at any framerate. So it allows you to unlock the framerate without the game "skipping" (judder).

Now I don't know why so many games that you have tried don't seem to be locking the framerate to 30 or 60 FPS with standard V-Sync on a 60Hz monitor (or at least trying to - in my experience it's not perfect) but the point is that even if the framerate is unlocked like that, on a standard monitor it will constantly judder and you will have several frames of latency; which G-Sync eliminates.

Even if the framerates are identical between using G-Sync and using V-Sync/Triple-Buffering, they will judder on a regular display, and be totally smooth on a G-Sync display - at least to the point at which the framerate is so low that you no longer see it as being smooth - I'd never consider 25 FPS smooth, even on a G-Sync monitor for example. The minimum framerate that you still find to be smooth varies from person to person.

You mentioned that you play Borderlands 2 - I don't know about many of the other games, but I seem to remember that game running terribly in many areas if you had the draw distance turned up. (CPU limited, not GPU limited) That game should benefit massively from G-Sync. It was a juddery mess on my PC.

[Joolsyz]

Borderlands 2 for me is still not perfect with G-Sync no matter what the frame rate, not silky smooth where as if I use regular V-Sync it is. Must be the game engine i guess. I have a capable system too.

[Lanarchy]

I made 2 videos for Borderlands 2. One with Vsync On, and one with Gsync. I made extra sure Gsync was OFF when using Vsync, also confirming my PG278Q's led was white with Vsync, red with Gsync. For Vsync, I forced Vsync through Nvidia's control panel as well as in-game, as you will see in the video. I am 100% confident you will see there is no difference, not only framerates (which STILL aren't dropping to 30 !!) but on judder, and I know I don't show it on the video but I had draw distance at Ultra High for both videos. Still no worse judder on Vsync than Gsync.

phpBB [video]

phpBB [video]

2 games now I've shown you Gsync doesn't do anything different than Vsync. I can do it for a lot more games if you want. There is no 60 fps dropping to 30, no screen tearing, and no input lag in any of the videos I've shown you so far, using Vsync. And general visual and gameplay experience, I am sure, you can confirm look absolutely the same.

Again, I am not looking to discredit Gsync, I just want to see the light ! I'd be so happy to turn on Vsync and be like "Ewwwww" and then go back to Gsync and feeling better. As it is, I could switch to Vsync for a few days and I wouldn't even notice (as happened a few days ago during testing, I forgot to re-enable Gsync for an entire day. Couldn't even tell).