G-Sync's 1ms Polling Rate: My Findings & Questions

Talk about NVIDIA G-SYNC, a variable refresh rate (VRR) technology. G-SYNC eliminates stutters, tearing, and reduces input lag. List of G-SYNC Monitors.
User avatar
RealNC
Site Admin
Posts: 3730
Joined: 24 Dec 2013, 18:32
Contact:

Re: G-Sync's 1ms Polling Rate: My Findings & Questions

Post by RealNC » 18 Mar 2017, 13:07

I'm getting bottom tearing with a single 980 Ti way before I hit the refresh cap. So I don't think it has to do with SLI.

GSync really prefers vsync on rather than off.
SteamGitHubStack Overflow
The views and opinions expressed in my posts are my own and do not necessarily reflect the official policy or position of Blur Busters.

User avatar
Chief Blur Buster
Site Admin
Posts: 11646
Joined: 05 Dec 2013, 15:44
Location: Toronto / Hamilton, Ontario, Canada
Contact:

Re: G-Sync's 1ms Polling Rate: My Findings & Questions

Post by Chief Blur Buster » 18 Mar 2017, 14:03

It would be great to have a dynamic VSYNC OFF/ON mode -- frame-specific.

Basically if raster is near the bottom (e.g. RasterStatus.ScanLine API) then mimic VSYNC ON by waiting 0.05ms (less than 1ms).

The closest the tearline is to the bottom, the less latency VSYNC ON (for that particular frame) would have.

LCD displays refresh from top to bottom (high speed video), and display data is transmitted from graphics card to monitor, one row of pixels at a time, in a top-to-bottom manner. The timing where the tearline occurs, determines how much latency saved relative to double-buffered (minimum possible queue depth) VSYNC ON.

A refresh cycle for GSYNC with a 144Hz cap, is 6.9ms from top edge to bottom edge.
--> So if a tearline occurs at the top edge, that's likely VSYNC OFF >6ms saved relative to VSYNC ON.
--> But if the tearline occurs near the bottom edge, that's VSYNC OFF saving less than 1ms relative to VSYNC ON.

So what could in theory happen, is tearing would only occur further above the bottom edge of the screen.

I'd think this could be a latency-optimized modified version of "Adaptive VSYNC". Turns VSYNC ON whenever the raster is near bottom, but turns VSYNC OFF when the raster is far away from bottom. This would cause the bottom-edge tearing to completely disappear, while tearing will still (rarely) occur throughout the screen at other times.

This could be a suggestion to NVIDIA, as an "Optimized VSYNC OFF" setting.
Head of Blur Busters - BlurBusters.com | TestUFO.com | Follow @BlurBusters on Twitter

Image
Forum Rules wrote:  1. Rule #1: Be Nice. This is published forum rule #1. Even To Newbies & People You Disagree With!
  2. Please report rule violations If you see a post that violates forum rules, then report the post.
  3. ALWAYS respect indie testers here. See how indies are bootstrapping Blur Busters research!

mminedune
Posts: 31
Joined: 01 Feb 2014, 08:12

Re: G-Sync's 1ms Polling Rate: My Findings & Questions

Post by mminedune » 19 Mar 2017, 11:45

RealNC wrote:I'm getting bottom tearing with a single 980 Ti way before I hit the refresh cap. So I don't think it has to do with SLI.

GSync really prefers vsync on rather than off.
Idk what to tell you I've been dealing with this forever now some games worst than other bf1 being the worst. Like when zooming in and out of map when spawning.

Now no tearing what so ever regardless my frames with vsync off.

Btw this is my thread

https://forums.geforce.com/default/topi ... king-100-/

If you look at my last vid i t got tearing in doom with SLI disabled. IDK is disabling SLI is same as just one GPU installed gpus installed. But I re ran doom with my 1080 ti no tearing at all.

Sparky
Posts: 682
Joined: 15 Jan 2014, 02:29

Re: G-Sync's 1ms Polling Rate: My Findings & Questions

Post by Sparky » 19 Mar 2017, 12:51

While I haven't heard any official confirmation, I believe the tearing while using gsync with vsync off and a framerate cap is caused by inconsistent frame times, where a single frame hitch can cause subsequent frame(s) to come in faster than the set framerate cap.

For example, you set a 100fps cap with a 144hz display. Normally you get one frame per 10ms, but say one frame hitches and takes 15ms. This gets displayed just fine, but now your very next frame comes in 5ms later. This causes a tear, because the refresh interval is 7ms. If you leave vsync on, that 5ms frame will get delayed by 2 ms, and you avoid the tear, with the third frame displayed on time.

It makes sense that SLI would make that type of tearing more common, especially with AFR. (overall frametime consistency, pipeline depth, and a delay on one frame won't delay rendering on the next frame).

mminedune
Posts: 31
Joined: 01 Feb 2014, 08:12

Re: G-Sync's 1ms Polling Rate: My Findings & Questions

Post by mminedune » 19 Mar 2017, 14:27

Battlefield 4 and more so 1 are prob worst games when it comes to frame times and it's fine now for me.

Also forgot to mention also increasingr dsr or any ingame downsamplet causes issues with gsync in SLI. With vsync off you get tearing vsync on can cause stutter.

single gpu with vsync off and frame cap i can downsample just fine.

User avatar
RealNC
Site Admin
Posts: 3730
Joined: 24 Dec 2013, 18:32
Contact:

Re: G-Sync's 1ms Polling Rate: My Findings & Questions

Post by RealNC » 19 Mar 2017, 16:28

It doesn't make sense to use gsync+vsync off to begin with, so it's a non-issue really.

If you want vsync off, just do that. There's no reason to use gsync. The only reason to use gsync is because you don't like vsync off.

Gsync + vsync on + frame cap. Just use that.
SteamGitHubStack Overflow
The views and opinions expressed in my posts are my own and do not necessarily reflect the official policy or position of Blur Busters.

mminedune
Posts: 31
Joined: 01 Feb 2014, 08:12

Re: G-Sync's 1ms Polling Rate: My Findings & Questions

Post by mminedune » 20 Mar 2017, 16:56

I never said i ran gsync+vsync off./ gsync on vsync off and frame cap... there is not reason to run vsync with gsync it does add slight lag and stutter/hitching when SLI frame times are off.

Ive been dealing with this since i got gsync couple years ago when i had 980 SLI.

Gsync sucks with SLI period.

mminedune
Posts: 31
Joined: 01 Feb 2014, 08:12

Re: G-Sync's 1ms Polling Rate: My Findings & Questions

Post by mminedune » 08 May 2017, 15:27

And just an update to my claim of not getting tearing on lower part of screen with single GPU low and behold setting in game frame cap causes tearing on lower part of screen with single GPU gsync on/vsync off.

Use nvidia profile inspector instead to set it at a driver level and set like 3 frames below refresh and you wont get any tearing on lower part of screen with gsync on/vsync off

Regardless tearing will happen on lower part of screen with SLI at higher frames because of AFR frame time inconsistencies and having vsync on can lead to stutter instead of tearing.

gsync on/vsync on is awful
Last edited by mminedune on 08 May 2017, 15:58, edited 2 times in total.

User avatar
RealNC
Site Admin
Posts: 3730
Joined: 24 Dec 2013, 18:32
Contact:

Re: G-Sync's 1ms Polling Rate: My Findings & Questions

Post by RealNC » 08 May 2017, 15:32

mminedune wrote:And just an update to my claim of not getting tearing on lower part of screen with single GPU low and behold setting in game frame cap causes tearing on lower part of screen with single GPU gsync on/vsync off.

Use nvidia profile inspector instead and set like 3 frames below refresh and you wont get any tearing on lower part of screen with gsync on/vsync off
On single GPU, you should use g-sync + vsync on + cap. It does not add input lag.

Using nvidia inspector can give you almost as much input lag as vsync.

The SLI thing, yeah. I've seen people report g-sync issues with SLI quite a bit. I don't think anyone found a solution to these problems.
SteamGitHubStack Overflow
The views and opinions expressed in my posts are my own and do not necessarily reflect the official policy or position of Blur Busters.

mminedune
Posts: 31
Joined: 01 Feb 2014, 08:12

Re: G-Sync's 1ms Polling Rate: My Findings & Questions

Post by mminedune » 08 May 2017, 15:40

vsync works with gsync it add frame buffer its much better without it you can think what you want i don't care ive been dealing with this forever now vsync /gsync is garbage no reason to have any frame buffer with gsync. Its ten times better with gsync on its own and capping frames

Post Reply