Stutter Stutter or Smooth as Butter

[jimortality]

Brands Hatch especially, even in practice screen stutter. I think it's not just me and maybe down to the unreal engine? I've tried all different things so if you're smooth as Butter, please post your settings and how you're eliminating it. Thank you.

Acer Predator X34p
I 7-6700k
Nvidia 1080ti
16gb Ram a lam.
And not forgetting my brand new Master Cooler 212 Evo

 

[gyrtenudre]

Hmm that's interesting, never thought of that, but yes it should be coordinated by CPU. I guess it would then depend on the task priority one process has over the other (?)... But on the other hand, all forms of synchronization should be implemented in CPU code.

About the GPU load, I'm actually over 80% atm, and I have no issues besides the micro-stutters during the first outlap (if I go back to the pits and come out again it's smooth).

 

[gyrtenudre]

Guess I'm gonna test it! Probably in acc as with my hardware I can switch from CPU limit to gpu limit with just a few clicks

Oh, and something you might need if you test this: I had to turn on "Custom Direct3D support" in RTSS for scanline to work with ACC.

Edit: that was placebo. It still fails to detect the game at times. So, "Custom Direct3D support" make no difference

 

[Neilski]

Sadly you didn't test vsync + rtss at 0.01-0.3 fps below your true Hz.
Scanline sync still costs a lot of resources, doesn't it?

Doh, I totally forgot about the capping thing - long thread, short memory! Will have a play with that and see how the lag looks. I'm reminded that it still sorta makes no sense to me though

It never even crossed my mind that the scanline sync could use resources and I didn't spot any issues but will be more alert now. Indeed though, my 970 isn't stretched with the settings I use (as Martin spotted, I'm using AC for these tests ).

BTW I'm not aware about Scanline sync costing a lot of resources. At least with ScanFlush=1 it shouldn't, but even without it doesn't seem to. I use it pretty much in every sim nowadays.

Hmm, I can't find links to explain the flushing thing - is it called SyncFlush? I found one thread where a guy said "I can't find it" and some total loserknob (moderator on the Guru3D site) kept saying "if you can't find it, then it isn't for you" (i.e. too noobish)... ROFL. My installation of RTSS has been on my hard disk for a few years now (though it has been updated) and the word flush simply doesn't appear in the cfg files.

One other interesting thing to note about scanline sync: when using DSR, it uses the native resolution height, which is ofc lower if DSR > x1, and you can't push the tear line in the non-visible area.

Well that's just weird. I'd have thought that the DSR happened in a separate buffer somewhere and then the result got dumped into a buffer of native dimensions, so that the RTSS scanline stuff wouldn't even notice...

 

[Case_]

@Neilski Might be SyncFlush, sorry. I didn't check before posting... BTW it's not included in the profile by default, you have to add it manually.

 

[Neilski]

@Neilski Might be SyncFlush, sorry. I didn't check before posting... BTW it's not included in the profile by default, you have to add it manually.

No worries Martin - but how on earth do you figure out how to use it, e.g. what values to use? I can't find any kind of documentation on it, so am a bit confused.

 

[Case_]

@Neilski Some of it is explained in posts following the one this links to:

MSI AB / RTSS development news thread

Why not keep it in and just don't have it in the config file if users edit that directly. Because this: a lot of people just tend to ignore CFG entry...

forums.guru3d.com

 

[Case_]

BTW found the first real issue with the DX12 version. I present to you...the Lamborghini Huracán GT3:

Only does it in the game, not in showroom, only some liveries and only at the highest LOD level. May become invisible at lower LOD levels. Also sometimes shoots polygons

 

[maelstrom]

BTW found the first real issue with the DX12 version.

Just so that it's clear to anyone jumping into the thread, there isn't a DX12 version of ACC. It's a DX11 game and you're just forcing it to run in a version of DirectX that it wasn't developed or optimised for. See: https://www.assettocorsa.net/forum/...formance-discussion.60924/page-9#post-1125888

 

[RasmusP]

Just so that it's clear to anyone jumping into the thread, there isn't a DX12 version of ACC. It's a DX11 game and you're just forcing it to run in a version of DirectX that it wasn't developed or optimised for. See: https://www.assettocorsa.net/forum/...formance-discussion.60924/page-9#post-1125888

Thanks for underlining this clearly for everyone jumping in
It sounds a bit though like you wanna say that Martin didn't make it clear so I'll point that out too haha

See, I wasn't aware of it and it's not mentioned a lot, but you can run ACC in DX12. It's actually very simple - you just start the game using the "-dx12" switch.

Now from the little I've seen this discussed, the developers actively discouraged people from using this (and seemed very annoyed by it being even mentioned). Apparently the graphical components and functions are not all DX12 compatible and the performance boost is simply due to "just disabling a select and arbitrary number of incompatible graphical components and effects", to quote exactly, and is apparently "as scientific as closing your eyes and assuming the world has disappeared".

 

[maelstrom]

It sounds a bit though like you wanna say that Martin didn't make it clear

It was just the specific post above that I was referring to where he said "the DX12 version". For clarity, I was pointing out that there isn't one!

 

[Case_]

I'm not even saying anything...

(Oh, crap...well nevermind then.)

BTW, I quite literally quoted the post you linked. And I honestly still don't understand why they were so hostile to Koly when he mentioned it. Even though it's Koly, I would still like to know more about what the issues are, and, more importantly, if it really disables some graphics effects, why wouldn't they make a safer way to disable these effects directly into the game. It gives a huge performance boost in CPU-limited situations (which is basically all the time the ACC is running ) for an unnoticeable visual difference.

Meanwhile everyone including the devs just proceeds to angrily yell at him about how incredibly stupid what he's doing is. I guess getting ACC to perform 25% better is incredibly stupid...

Then again that's AC forums for you - a place that should be avoided if at all possible, because it will make you facepalm really hard really quickly.

 

[Case_]

BTW I did some input lag measurements on ACC yesterday. But as mentioned, it was only with a 60 fps camera, so not a great accuracy. However I tried to cover most scenarios talked about. I did multiple measurements for each.

Fullscreen optimizations on

fullscreen, vsync on – 183-200 ms
fullscreen, vsync off – 100-117 ms
borderless, vsync off – 117-133 ms
borderless, vsync on – 150 ms
borderless, vsync on, limit -0.01 Hz – 133-150 ms
fullscreen, vsync on, limit -0.01Hz – 133-150 ms
fullscreen, vsync off, negative scanline sync – 100-117 ms
borderless, vsync off, negative scanline sync – 133-150 ms

Fullscreen optimizations off

fullscreen, vsync on – 133-150 ms
fullscreen, vsync off – 100-117 ms
fullscreen, vsync on, limit -0.01Hz – 117-133 ms
fullscreen, vsync off, negative scanline sync – 100 ms
fullscreen, vsync off, ingame limit to 60 fps – 100 ms

I think the most important point here is that the input lag is indeed absolutely horrendous, even in best cases. I'm pretty sure that when I measured say AMS in the past, I was somewhere around 40-ish ms.

 

[Case_]

Regarding the input lag, after some more testing...it's even worse than what you see above, because it really seems that if you run the game close to a CPU bottleneck (which is very easy to do in ACC), then the input lag gets even higher. So you have high-ish framerate, but the driving itself gets awful. I'll be testing some potential ways of how to reduce it.

And as for the DX12 ver...forcefully demanded use that everyone rightfully frowns upon, the issue with the weird car might be related to the HLOD option - I haven't encountered the issue since I turned HLOD off, which for me doesn't seem to have any real impact on performance anyway. Which would indeed limit the DX12 issues to the disabled dashboard.

 

[Terry Rock]

Go get the 450.12 driver and watch smoothness like you've never seen before.

 

[maelstrom]

Go get the 450.12 driver and watch smoothness like you've never seen before.

What graphics card are you using it with? I thought that 450.12 was just a test driver for the upcoming major update to Windows. The latest official driver I can find on the GeForce website is 445.87.

 

[Terry Rock]

What graphics card are you using it with? I thought that 450.12 was just a test driver for the upcoming major update to Windows. The latest official driver I can find on the GeForce website is 445.87.

I am using an MSI GTX1080 Ti Duke Edition.
That driver is the absolute best I have ever used with any GeForce card.
I actually found it in the Nvidia GeForce driver forum at Guru3d.com.
It is even better than the later official 450.82 driver in my case.
I currently run every single game at or close to the maximum settings.
The performance is smoother than at any time previous...and it was great to begin with.
I do not go out of my way to check framerate, much as I look for all-out smoothness at every point on track.
The only time I look at frames is in VR... because I have the fpsVR app right there in my headset.

 

[RasmusP]

I am using the MSI GTX1080 Ti Duke Edition.
The driver is the absolute best I have ever used with any GeForce card.
I actually found it in the Nvidia GeForce driver forum at Guru3d.com
It is even better than the later official 450.82 driver in my case.
I can run every single game at or close to the maximum settings and the performance is smoother than at any time previous.
I do not go out of my way to check framerate, as much as I look for all-out smoothness at every point on track.
I'll look at frames in VR because I have the app there in my headset.

Sounds like it's time for a full backup and testing! With the sadly not so brilliant gsync on some drivers I didn't update in over a year. Every time I tried I went back to my current one.

But this sounds awesome! Thanks!

 

[maelstrom]

It is even better than the later official 450.82 driver in my case.

I'm puzzled (which is happening more frequently as I get older). According to the GeForce web site, the latest official driver I can download is 445.87, which was released on 15 Apr 20. The 450.82 driver only seems to be available as a DX12 developer preview for testing purposes.

 

[Terry Rock]

I'm puzzled (which is happening more frequently as I get older). According to the GeForce web site, the latest official driver I can download is 445.87, which was released on 15 Apr 20. The 450.82 driver only seems to be available as a DX12 developer preview for testing purposes.

Sorry....you are not confused.
I forgot I got the 450.82 from a tester.
In any event, the 450.12 for me anyway, is the better driver.
It just runs extremely smooth.

 

[RasmusP]

I finally did some tests. Grabbed the iPhone 8 from someone and did some 240 fps videos. Due to my tenkeyless keyboard with external usb-numblock I needed to map a number key for upshift.
I pushed the num-key and the 7-key as synced as I could. Basically slammed them to minimize the timing difference.

Results:
AC is completely synced comparing the cockpit-display, gears app and sidekick app. I have AC limited to 60 fps / 60 Hz. Gsync active.

Did multiple upshifts, here are the results. I went frame for frame in vegas and marked the last and first frame before/after the numpad-light and upshift.

7-9 ; 11-13 ; 16-18 ; 13-15 frames.
Average: 12.75 frames

Windows text editor is pretty identical at 100 Hz:
14-16 ; 9-11 ; 13-15 frames
Average: 13 frames

At 240 fps the calculation is pretty simple but I write it down anyway:
1/240 * frames = xx seconds

Time between camera frames is 4.2ms so accurately enough I'd say in relation to my results.

AC min = 29.2 ms
AC max = 75.0 ms
AC average = 53.1 ms

Windows min = 37,5 ms
Windows max = 62.5 ms
Windows avg = 54.2 ms

From my view I'd say the average is pretty much what it really is as my key hits were probably a bit de-synced.
I first wondered that the input lag wasn't lower at 100 Hz instead of 60 Hz but then I remembered that my monitor, according to tft central has an input lag of 3-4 ms.
So the lag would only become more consistent with a frame to frame time of 10ms instead of 16.67 ms.
This is visible in my test results but it's probably just my fingers...

Looking at my measurments, these roughly 7ms don't make a difference...

I'll try to measure ACC tomorrow! Didn't start it since months...