Blur Buster's G-SYNC 101 Series Discussion

[jorimt]

So I stuck on the G-Sync ON + V-Sync OFF bottom tearing problem and after some googling found this thread.

For an explanation on exactly why it does this, see entry #2 of my Closing FAQ here:
https://blurbusters.com/gsync/gsync101- ... ttings/15/

My question is, why can't RTSS measure frametimes right after GPU has done rendering and only then delay displaying the frame? And if this is not possible than how do VR games and headsets achieve perfect frametime and why can't it be done the same in flat games?

Firstly, framerate limiters aren't really intended to prevent tearing (you have to adhere to the display's VBLANK instead), and secondly, RTSS can set whatever frametime target it wants, but it's up to the system to ultimately comply, and due to inherit instability (PC performance is still limited by physics, after all), the system can't always hit an exact target, so sometimes said target gets overshot, and only after RTSS intercepts the frame, at which point it's out of any software's control.

This is where G-SYNC + V-SYNC "On" comes in; the V-SYNC option within the G-SYNC range acts as compensation for these instances of frametime variance by adhering to the VBLANK on the display (not the system) side.

As for VR, I'm not familiar with it's ins and outs, and some others here may be able to answer more specifically on that, but I'm guessing that 1) it adheres to the VBLANK (or its equivalent) to prevent tearing, and 2) its games are tuned to hit a very strict and specific framerate/frametime target as to accomplish its lock (along with techniques such as ATW and ASW, of which I can't currently speak on, as I haven't studied them in the least), something that the majority of non-VR games usually don't prioritize (and usually don't have to).

That, and VR vs. monitors is basically apples vs. oranges.

[hmukos]

Thank you for answer, jorimt!

For an explanation on exactly why it does this, see entry #2 of my Closing FAQ here:
https://blurbusters.com/gsync/gsync101- ... ttings/15/

It's not the bottom tearing that worries me the most. I read your article and know that it can be fixed by V-Sync with no additional input delay. But doesn't bottom tear imply that there is too much of a frametime variance? Consider 120fps fixed with RTSS at 144hz G-Sync monitor. In order for tearline to appear the last two scanouts should happen in 1-2ms less than 1/144 seconds. That means frametime of around 5 ms where perfect frametime should be 1/120=8.33ms. Isn't that too incosistent or have I misunderstood something?

[jorimt]

But doesn't bottom tear imply that there is too much of a frametime variance? Consider 120fps fixed with RTSS at 144hz G-Sync monitor. In order for tearline to appear the last two scanouts should happen in 1-2ms less than 1/144 seconds. That means frametime of around 5 ms where perfect frametime should be 1/120=8.33ms. Isn't that too incosistent or have I misunderstood something?

It depends on what range the frametime variances are in. G-SYNC (on) + V-SYNC off will only tear when either A) they occur near the max refresh rate, due to some frames' render time exceeding the refresh rate (e.g. being "faster" than the delivery time of the display), or B) when a frametime spike abruptly occurs, which causes the system to not output a new frame for a frame or frames at a time.

With 120 FPS at 144Hz, so long as the latter doesn't occur (and you have a framerate limit in place to keep it at that average for the former), you won't see tearing with G-SYNC (on) + V-SYNC off, as the render time of a frame with an 120 FPS target (8.3ms per) is well within the display time of 144Hz (6.9ms), since G-SYNC will simply "pad" the VBLANK (the span between the previous and next frame) by 1.4ms (8.3ms - 6.9ms = 1.4ms; you can apply this to any frametime higher than the monitor's scanout rate), then deliver the (already rendered) frame in 6.9ms, repeat.

[hmukos]

With 120 FPS at 144Hz, so long as the latter doesn't occur (and you have a framerate limit in place to keep it at that average for the former), you won't see tearing with G-SYNC (on) + V-SYNC off, as the render time of a frame with an 120 FPS target (8.3ms per) is well within the display time of 144Hz (6.9ms), since G-SYNC will simply "pad" the VBLANK (the span between the previous and next frame) by 1.4ms (8.3ms - 6.9ms = 1.4ms; you can apply this to any frametime higher than the monitor's scanout rate), then deliver the (already rendered) frame in 6.9ms, repeat.

So if I have 144 hz monitor, limit game at 120fps using RTSS and there are no abrupt frametime spikes (MSI afterburner frametime graph is perfectly flat) then I should not see tearing in the bottom? Strange, because that is the exact setup where I got bottom tearing (hence the above messages). I won’t have access to my monitor for a couple of weeks. When I get to it I will try again and report here. Thank you for your help!

[jorimt]

I won’t have access to my monitor for a couple of weeks. When I get to it I will try again and report here. Thank you for your help!

Sure, feel free to update here when you do.

[BTRY B 529th FA BN]

TL:DR

Is there any explanation why using Ultra Low Latency Mode combined with G-Sync Compatible & V-sync ON limits the fps under the monitor's max refresh rate? e.g. a 240Hz monitor will be limited to 225fps

Thanks

[jorimt]

TL:DR

Is there any explanation why using Ultra Low Latency Mode combined with G-Sync Compatible & V-sync ON limits the fps under the monitor's max refresh rate? e.g. a 240Hz monitor will be limited to 225fps

Thanks

https://blurbusters.com/gsync/gsync101- ... ttings/14/:

Low Latency Mode* Settings:
*This setting is not currently supported in DX12 or Vulkan.

• If an in-game or config file FPS limiter is not available, RTSS is prohibited from running, a manual framerate limit is not required, and framerate exceeds refresh rate:
  Set “Low Latency Mode” to “Ultra” in the Nvidia Control Panel. When combined with G-SYNC + V-SYNC, this setting will automatically limit the framerate to ~59 FPS @60Hz, ~97 FPS @100Hz, ~116 FPS @120Hz, ~138 FPS @144Hz, ~224 FPS @240Hz, etc.
• If an in-game or config file FPS limiter, and/or RTSS FPS limiter is available, or Nvidia’s “Max Frame Rate” limiter is in use, and framerate does not always reach or exceed refresh rate:
  Set “Low Latency Mode” to “On.” Unlike “Ultra,” this will not automatically limit the framerate, but like “Ultra,” “On” (in supported games that do not already have an internal pre-rendered frames queue of “1”) will reduce the pre-rendered frames queue in GPU-bound situations where the framerate falls below the set (in-game, RTSS, or Nvidia “Max Frame Rate”) FPS limit.

[BTRY B 529th FA BN]

I should have edited my post to explain that I've seen that page, and that I've experienced it in-game, but I'm more looking for an explanation on why it does this rather than what happens with different parameters when it's in use.

Thanks for the great article, btw

[jorimt]

but I'm more looking for an explanation on why it does this rather than what happens with different parameters when it's in use.

In what respect, exactly? The auto FPS limit? For G-SYNC, so long as your FPS is limited by the auto cap with Ultra enabled, the pre-rendered frames queue is effectively 0 AND it's keeping the FPS within the G-SYNC range/refresh rate at all times (but an in-game or RTSS limiter + LLM "On" with G-SYNC will do the same).

As for why Nvidia specifically chose to automatically cap the FPS with G-SYNC + V-SYNC + Ultra, my guess is the for above reasons.

[hmukos]

Why is there such a huge variation between Min and Max input latency when VSYNC OFF + 2000 FPS @ 60hz? Shouldn't game check inputs and show at least something on screen every 0.5ms or so regardless of refresh rate? Or have I misunderstood what "first on-screen reaction" means?