Sunday Chit Chat, Talk, Answer and Play Guitar (Stefano's Live Stream)

To me, vsync makes a HUGE difference in FFB and reaction time. Its always been the case, no matter what sim I drive, vsync off has always been the way to go when driving fast/competitively.

 

You guys realize how long consumer, belt driven wheel takes to react to your input.
This is not oprical gaming mouse with high polling frequency.
Kunos vsync implementation with triple buffering is as good as it gets.
Granted, nothing will beat G/Free Sync monitor, w/o one you need to compromise.
Would be interesting to see some stats on how used hardware (gpu, cpu) and monitor latency impacts ability to use vsync.

 

Guys, you´re talking different things. No one argues that Vsync OFF is the best when visual que is needed. Boris, and now I, confirms that whether there´s inputlag visible on screen or not, it does not change the FFB on the wheel. Sorry Boris for calling you a smoker

What I´m 100% sure of, is that it once DID change the FFB. My experience was it was less firm and lower in overall strenght. I could close my eyes and tell. Going back and forth between vsync on/off, it was like flipping a switch on the FFB wheel: race-mode and boat-mode. The later seems now gone or minimized to the insignificant. Visually for sharp handling, I still prefer vsync off on monitors. On the Rift it's impossible (locked by Oculus sdk), but I actually don´t see/feel delay , although there is some no doubt. It´s really good as is.

 

And I thought I'm the only one who experienced this. It really felt different in the past.[/QUOTE]

 

As a smoker i could've sweared you were 100% certain too a day ago that V-sync on/off changes FFB feel and delay.

Last several months I've been using V-sync out of necesitty and weren't even aware that visual lag comes with it. It's just that since some update (on GPU driver mainly I guess) I had so much tearing that it was unplayable with V-sync. Since Early Access I've been using the same hardware.

After watching Stefano's stream I tried again (after not so long ago) without V-sync and the game looks good again. I noticed I didn't had triple buffering on in GPU settings, and frame latency wasn't set at 1. I have to check which of the 2 made the biggest impact. FPS cap is now set at 83 and all looks smooth so far after a couple of multiplayer sessions.

I have been setting frame latency at 1 before but since it's not default I thought it wasn't necessary.


In short, I just wanted to share that after a while using V-sync I had a good experience again without it and using triple buffering and frame latency at 1 with FPS cap. To many it's like stating the obvious but it might be useful to noobs like me experiencing the same issue.

 

I know, embarrassingly enough I did not test again before making that statement, though it was based on tests done randomly the last 3 years.
Something has improved lately, to the point where I can finally leave it *sigh of relief*

 

great guitar-ing again stefano... btw i use V-sync. I prefer the smoothness like you and my eyes can't take the tearing. I wear glasses and it seems amplified by them.. Literally can't concentrate around corners. I dont get input lag at all on any of my sims.. maybe apart from DR that wheel just wobbles on screen!!

Great vid. good work as always kunos!.

 

If you have good reaction times you notice that Vsync causes a small delay.
Driving with Vsync means anticipating the corners, and in unpredictable situations more chances to lose control of the car.

More control vs eye candy.

 

I think all it goes to display.

I believe that almost all pc's will have (almost) the same delay.

The difference is the display.
In my case, I use a tv so lag is big.
Another user will use a 2ms monitor.
The monitor user can use vsync and might still have less input lag than me with the tv....

Στάλθηκε από το m2 note μου χρησιμοποιώντας Tapatalk

 

I just did a test setting frame latency to 0 and triple buffering off and didn't notice a significant difference. At least it's not like it recently used to be, unplayable without V-sync. I have FAA on, AF at 16 and AA at 4. Maybe it was a combination of all these settings that made it unplayable, or one of the many GPU options that was set wrong. Anywho, it works now for me without V-sync.

 

Negative here. Fairly smooth down the straights for me but huge stutter through the turns. Will wait for further optimization...it isn't even in the control panel, yet.
Of course others' experience is appreciated.

 

Well:

On the straights it is ok, but that probably is because the relative change of the frame is pretty low.
Look how bridges behave on straights, there should also be stuttering.

I'm back to Vsync, 60fps cap and frame latency to 1.

 

Let's not get upset here - it's a highly religious topic as the past has shown.

I will only add my experience: when I had maximum pre-rendered frames (or frame latency...whatever it's called) set to 0 I could feel a slight difference between vsync on and off. BUT I hotlapped both settings back to back and I couldn't beat my VSYNC time. When my car tried to spin on me I had a slight advantage without VSYNC in catching it. But I decided to go down the smoothness road. I strongly believe that VSYNC-off campers would match their lap times with VSYNC. Noone seems to dare try it, though.

But then someone made me aware that the max pre-rendered frames should be at 1, not 0. I assume that 0 makes it default to a much higher number (does anyone know?) because with 1 I definitely have less input lag, no doubt about it. I felt it instantly as I drove and almost lost the car for the first time I could catch it quite a bit better.

I assume I only have 16 ms lag now?!? And without VSYNC I get some 90 FPS so that would be 11 ms?!? I don't know exactly - this stuff is too complicated to tell universally for all systems. Only way would be to grab a high speed camera and record both physical and virtual wheel.

 

Sorry but I still don't understand what the max pre rendered frames setting does. Could anyone please explain?

What should be your setting if you're using gsync (assuming it makes a difference on these monitors yesterday).

I activated it yesterday but the difference was not obvious to me.

Thanks.

Envoyé de mon SM-N9005 en utilisant Tapatalk

 

Edit: Reminder for self: always read all posts -.-

 

Yeah, tried it. Works even with my GTX 660Ti.
Tried a couple of different settings, in AC and NVidia tools. None got a decent reduction of stuttering, I'm back to smooth VSync for now.
But look a few posts before yours.

 

Max frame at = 0 is actually 3 prerendered frames, as I remember. All this and how it work was covered long ago, I'm too lazy to find it atm. Probably is in the fanatec-controller bugs thread or so.

 

My nvidia driver setting (1) overrun AC's prerendered frames setting. There is no difference between 0 or 1.

 

Thanks all for your replies.

I have investigated this a bit and I found two sources most helpful.

1. this AC thread (among others). It's long but contains interesting information, including from Stefano. https://www.assettocorsa.net/forum/index.php?threads/fps-and-perceived-lag.1481/
2. This guide: http://www.displaylag.com/reduce-input-lag-in-pc-games-the-definitive-guide/#prettyPhoto. It tested pretty much every scenario you can have and measured input lag for each of them. It's based on Street Fighter 4 as the benchmark but this should be applicable to all games where input lag can be an issue (including racing simulators of course).

As for my Gsync question, I have only tried for a few minutes yesterday but I guess the reason I saw little to no difference between max prerendered frames at 0 or 1 was most likely that I'm using a Gsync monitor and AC runs at 90 fps+ on my rig so having slightly less input lag was barely noticeable regardless, at least to my (ageing) eyes and brain.


I copied the gist of the display lag article below if it can help some of you with their settings on non Gsync monitors. It's just a screenshot so obviously, if you want to watch the Youtube videos that are mentioned (or want AMD results), you need to read the whole article.

I hope this helps!

 

- MAXIMUM_FRAME_LATENCY at 0 means use driver default (maximum prerendered frames), if it's not set in driver, on Windows the default is 3.
1 is the lowest you can set it in driver. I also suspect that this setting is not used since Kunos put own "triple buffering" in place. You can use one or another, not both at the same time. Some confirmation would be nice.

- Triple Buffering is only used by OpenGL programs when using VSync. Setting it in driver should have no effect http://www.tweakguides.com/NVFORCE_8.html

- What Kunos call "triple buffering" is most likely program maintained rendering queue, slightly different but this terminology is often mixed.
http://www.anandtech.com/show/2794/2

- Fast sync is an interesting feature, if used properly, seems like for this to work you need to maintain 2x-3x fps of your monitor refresh rate (this is "forget it" in my situation)
http://www.tweakguides.com/NVFORCE_8.html