Optimal settings for 60fps locked gaming on 240Hz display

[SpaghettiViking]

I've recently started playing a fighting game called Soul Calibur VI competitively on PC.

This game is framerate capped at 60fps by the game engine itself (UE4).

I currently have the following two monitors:

* Alienware AW2518HF: www.amazon.com/Alienware-25-Gaming-Monitor-Height-Adjustable/dp/B0733YCKM5

* LG 32MU59-B: www.amazon.com/LG-Electronics-32MU59-B-Screen-Monitor/dp/B075RPVGBX

Currently I enjoy playing the game at 4K on the LG monitor at 60Hz, 60FPS. I've found that with V-Sync off, my 2080ti graphics card causes a lot of tearing because it can render well above 60FPS. I started using FastSync instead of V-Sync, which works reasonably well, with occasional stutters.

Given that the game is engine-locked to 60FPS and I cannot utilize the VRR tech, is there any point in moving the game over to the 240 Hz 1080p monitor in hopes for lower input lag? Is there any appreciable difference between FastSync and V-Sync for this scenario? Should I use neither one?

For reference, I am using an EVGA RTX 2080ti XC for graphics, so hitting a particular framerate isn't an issue unless we're talking about 100+ FPS at 4K Ultra.


Cheers!

[Chief Blur Buster]

As of year 2018:

For PC: YES
Yes, you turn on VRR and simply let the game cap itself.
You will get less lag 60fps@240Hz fixed-framerate VRR.

For Consoles: NO
No. Current 240Hz monitors have higher input lag for 60Hz consoles because panel scanout is not synced to cable scanout. Most 240Hz panels require high-velocity scanout, which forces the monitor to buffer slow-scanrate signals for fast-scanrate panels. It requires something called Quick Frame Transport (QFT) (technically, ultra-large blanking intervals) to transmit a 60Hz refresh cycle at 240Hz-speeds. Or the ability to use high-Hz (e.g. 120Hz or 240Hz) while using 60fps.

Lag Differences To Expect
Excluding LCD GtG -- and assuming synchronized realtime panel scanout to signal scanout -- you will get as much as ~12ms less input lag (the difference of 240Hz and 60Hz) for screen top edge and ~6ms less input lag for screen center (the difference of 240Hz and 60Hz, divided by two) than a good TN 60fps@60Hz display using FastSync. That's because of scanout latency of common bog-standard 60Hz signals, the screen paints the 60fps frame slowly in 1/60sec from top to bottom. Even if the monitor realtime displayed the frame, the cable is still transmitting the frame, top to bottom, so the bottom row of pixels is displayed 16.7ms after the top row of pixels! So yes, 240Hz reduces the lag of 60Hz. And yes, fixed-framerate 60fps GSYNC has less lag on 240Hz monitors -- than FastSync on 60Hz monitors. You will not get the stutter-eliminating benefits of VRR but you will definitely get the lag-eliminating benefits of VRR. You *can* do FastSync 60fps at 240Hz and it will be much lower lag than FastSync 60fps at 60Hz. But if you're upgrading to a 240Hz monitor on a big-budget RTX series card, you might as well make sure you have GSYNC too -- GSYNC actually adds a few other benefits like good overdrive tuning and motion blur reduction capability (including ULMB 60Hz hack compatibility).

Being a Slave To A Monitor's Scheduled Refresh Cycle
FastSync is still beholden to the fixed refresh cycle schedule, so if a refresh cycle is missed, the frame is delayed to next refresh cycle. This does not happen with GSYNC, so you still get less lag. Assuming a frame is presented at a random time interval anytime within the current refresh cycle, that means a random latency delay of [0...refresh-cycle time] until that frame gets displayed. GSYNC displays refresh cycles instantly on software demand (refresh cycle timing are asynchronous and software-triggered when within VRR range), eliminating the quantization lag of converting amorphous frame presentation time to a monitor's fixed refresh cycle interval time. VRR tech bypasses that, and thus still benefits reducing lag even if you have a fixed framerate.

Pixels are transmitted one pixel row at a time, top-to-bottom, over the cable, so for anything that has no tearlines, bottom edge always has more lag than top edge. Assuming fastest possible refreshing (realtime refreshing of panel directly from cable scanout). VSYNC OFF can equallize top/bottom lag by essentially splicing new fragments of frames into current reresh cycle, but VSYNC OFF, obviously has those tearlines (where the new frameslice splices into the monitor's uninterrupted scanout). So let's exclude VSYNC OFF and focus on the lowest possible lag you can get with tearing-free sync technologies out-of-the-box:

By the math
Testing can vary (due to nebulous stuff like pixel response speeds and packetization behaviours on modern video cables) but mathematically "on the wire", this is what happens:

1/60sec = 16.7ms
1/240sec = 4.2ms

60Hz@60Hz VSYNC ON -- adds 2x-3x frame to screen bottom (33.3ms or 50ms lag)
60fps@60Hz FastSync -- adds 16.7ms lag + random [0ms ... 16.7ms] for screen bottom
60fps@240Hz FastSync -- adds 4.2ms lag + random [0ms ... 4.2ms] for screen bottom
60fps@240Hz GSYNC -- adds 4.2ms for screen bottom

Conclusion
If you're wanting the lowest possible PC-based 60fps lag while avoiding VSYNC OFF -- then I definitely do recommend 240Hz GSYNC for you. Alternatively, use 240Hz FastSync for 60fps if you rather save money and skip the GSYNC premium -- however, for framerates well below Hz -- FastSync 60fps is slightly more laggy than GSYNC 60fps (by an average of half max-Hz refresh cycle with).

[SpaghettiViking]

As of year 2018:
Snip

Thanks Chief!

That's really helpful stuff! That bit about how input lag is dynamic based on the vertical position of the reference pixel is new information to me. It makes good sense though.

Unfortunately I purchased my 240Hz monitor quite some time ago. The RTX card is a new addition to my rig.
If I knew then that I'd have the extra cash for the RTX card, I would have sprung for a GSYNC VRR instead of the FreeSync model.

For my Console experience, I'm pleased to use my 65" LG C8 OLED. It's not quite as competitive as the 60Hz monitors used for competitive console events, but it's gosh-darn pretty and the input lag is much lower than on big screen TVs from 5-10 years ago.

[Chief Blur Buster]

Thanks Chief!

That's really helpful stuff! That bit about how input lag is dynamic based on the vertical position of the reference pixel is new information to me. It makes good sense though.

Yep! Most sites only are familiar or measure screen center lag or full frame lag. I consider little nuances like this, even the asymmetry between cable scanrate and panel scanrate, and monitors using internal scanrate conversion to put a slowscan signal (e.g. 60Hz) on a fastscan panel.

Even Leo Bodnar Lag Tester has 3 squares (top, center, bottom) acknowledging spatial lag differences throughout the surface of the screen plane. But it goes way more complicated than that too, like the lag-differentials generated by differences in cable scanrate versus panel scanrate. Such asymmetries require internal scan converters (much like is mandatory for DLP projector and plasma displays) which creates interesting lag-distortion characteristics too)

I work with monitor manufacturers on the side (and more are contacting me as time goes on) so I'm quite familiar with the high-level input lag behaviours that spatially varies along the plane of the screen.

Eventually we'll have 1000Hz monitors and all of this will essentially not matter.

If you have not seen it yet -- see the Blur Busters Holiday 2017 special, Blur Busters Law And The Amazing Journey To Future 1000Hz Monitors. I am going to write a Holiday 2018 followup to this Holiday 2017 special. Although this is future stuff ("next decade gaming monitor stuff"), I found additional benefits of ultra-high-Hz display refresh rates that are not even written about in the Holiday 2017 special article.