EXCLUSIVE: We measure G-SYNC Input Lag in GSYNC Preview #2!

[RealNC]

V-Sync behavior when Hz is maxed out? This doesn't bode well for 60Hz G-Sync monitors, does it? :-/ A test with "fps_max 144" would have been a good idea, I think.

[crun]

Great test, but results are worrisome.

In BF4 my FPS varies between 60-160 (depending on amount of players, what's happening on the screen etc).
I thought G-Sync will be a perfect solution for it, but if it is going to noticably increase the input lag when we reaching or exceeding G-Sync frame rate cap... not good.

[Chief Blur Buster]

Very interesting article Mark! I'm guessing that ULMB enabled at 85hz/85fps would have less motion blur than 120hz/120fps with no strobe of any kind enabled, no?

Yep.
The strobing appears to maintain approximately a 25%:75% duty cycle on:off as the refresh rate changes, so that's divides persistence by 4 (aka, www.testufo.com/photo becomes about 4x sharper), so you'd be getting:

120Hz non-strobed = 8.3ms persistence
85Hz non-strobed = 11.7ms persistence
120Hz ULMB = 1/4th of 120Hz non-strobed = ~2ms persistence
85Hz ULMB = 1/4th of 85Hz non-strobed = ~2.9ms persistence

Which translates to 2.9 pixels of motion blurring during 1000 pixels/second framerate==stroberate motion.

[Dustmuffins]

Fantastic work on this article! Can't wait to get my hands on one of these bad boys!

One thing though: I'm curious to see how gsync would perform in CS:GO at 142 fps, as this has been the point that vsync+frame rate limiting would produce the lowest input lag. I can notice a big difference between 142 and 143/144.

[Chief Blur Buster]

Fantastic work on this article! Can't wait to get my hands on one of these bad boys!

One thing though: I'm curious to see how gsync would perform in CS:GO at 142 fps, as this has been the point that vsync+frame rate limiting would produce the lowest input lag. I can notice a big difference between 142 and 143/144.

I imagine that decreasing to 142, or 141 would solve the problem. There must be a threshold somewhere where the GSYNC latency suddenly disappears.

Ideally it should not ever appear at 143 or 144fps, and should gradually fade-in. But the way that NVIDIA has currently done it, it appears to have a sudden emergence of latency at the frame rate cap. I imagine it's wholly possible that future iterations of the NVIDIA drivers, would completely solve this issue.

[nimbulan]

I imagine that decreasing to 142, or 141 would solve the problem. There must be a threshold somewhere where the GSYNC latency suddenly disappears.

Ideally it should not ever appear at 143 or 144fps, and should gradually fade-in. But the way that NVIDIA has currently done it, it appears to have a sudden emergence of latency at the frame rate cap. I imagine it's wholly possible that future iterations of the NVIDIA drivers, would completely solve this issue.

The nVidia control panel used to let you turn maximum pre-rendered frames down to 0, but now only allows a minimum of 1. I would think that bringing that option back would solve the problem, though it being configurable inside games would be very helpful as well. It's never going to happen though as people have been trying to get nVidia to allow color profile overrides in games for years and requests always seem to fall on deaf ears.

[Dustmuffins]

I imagine that decreasing to 142, or 141 would solve the problem. There must be a threshold somewhere where the GSYNC latency suddenly disappears.

Now if only we could find someone with a super cool setup to test this *nudge nudge* =P

I hypothesize that BF4 (and nearly all games for that matter) would behave just like CS:GO if you stared at the sky and cranked down the graphics to bring the framerate up to 144. Gsync seems to behave just like Vsync with respect to input lag at 144 and 143fps.

I'd be very curious to see if you get the decrease in input lag at 142, without the microstutters of vsync.

The nVidia control panel used to let you turn maximum pre-rendered frames down to 0, but now only allows a minimum of 1. I would think that bringing that option back would solve the problem

That sounds interesting, but my gut reaction would be that it probably wouldn't, as changing this value doesn't have an impact with Vsync. According to my well calibrated eyeballs, 144 and 143 fps are much lagger than 142 fps with the render ahead limit being 1 or 3. As expected, 142fps with render ahead limit set to 1 is the best option with Vsync.

[nimbulan]

The nVidia control panel used to let you turn maximum pre-rendered frames down to 0, but now only allows a minimum of 1. I would think that bringing that option back would solve the problem

That sounds interesting, but my gut reaction would be that it probably wouldn't, as changing this value doesn't have an impact with Vsync. According to my well calibrated eyeballs, 144 and 143 fps are much lagger than 142 fps with the render ahead limit being 1 or 3. As expected, 142fps with render ahead limit set to 1 is the best option with Vsync.

Turning that setting down to 1 had a significant impact on input lag in Guild Wars 2 on my system with vsync enabled. This is at framerate < refresh rate though.

[Dustmuffins]

As it should. What I mean is something like this:

RAL = Render ahead limit.
144 fps 1RAL = 40ms
143 fps 1RAL = 40ms
142 fps 1RAL = 25ms

144 fps 3RAL = 55ms
143 fps 3RAL = 55ms
142 fps 3RAL = 40ms

I'm just making these numbers up off of the top of my head... I'd imagine they're close to this.

As you can see, decreasing the render ahead limit does have a major impact, but it still doesn't affect the difference between 142 fps and 144. So I hypothosize that even with a render ahead limit of 0, the difference will still exist.

[Chief Blur Buster]

I think the fact that lag is still there at 143 is a deficiency somewhare along the chain, either in the game, the drivers, the G-SYNC poll algorithm, etc. It may be a poll collision (e.g. drivers polled the monitor, monitor still appeared busy refreshing, drivers suddenly waited much longer). I know NVIDIA said they are working on eliminating the poll (1ms), it's also possibly a manifestation of this that we're seeing. I might ask NVIDIA directly about this...