Engine upgrade and DX12 coming to ETS2 and ATS!

[mmorselli]

(And VR is a very important topic for a flight simulation.)

And for a driving simulator too


But the main reason I would like a change in the graphics engine is not performance, but more effective antialiasing algorithms such as DLAA. I can fix the performance problem by buying more and more powerful computers, but I can't do anything for the awful Truck Simulator antialiasing

[Berniyh]

True, the antialiasing and also (further) improved lighting would be significant graphics update.
The latter especially if you're using the OpenGL backend which is still as bad as the very old DX9 backend in terms of lighting.
And if you're a Linux user like me, then for the native version that is all you have available, unfortunately.
(Well, you can use the Windows version via Proton, but that creates other problems. I'm currently trying to work around those.)

I can fix the performance problem by buying more and more powerful computers

Only to a limited amount. I recently upgraded from a Vega 64 to a RX 6900XT. So for the resolution I'm using (5120x1440) I have approximately 3x the rendering power available.
Unfortunately, ETS2/ATS are heavily limited by the CPU, so the gains were really more in the range of maybe 20% at the same settings.
It did allow me to ramp up the scaling and that helps a bit with some problems (like aliasing), but that's mostly it.
Of course, there would be something to gain using a stronger CPU, but there isn't that much to gain, maybe 20%, even if you get one of the new Zen 4 CPUs.

In the end, it's something I can live with, reaching only around 50-70 fps on very high/ultra settings, but it's not optimal for a driving simulator.
The goal should be something in the range of 100-150 fps.
Vulkan really could help to get there, due to the reduced CPU overhead, even if the engine remains single threaded for the time being, but ultimately, this game really needs multithreading.

[flight50]

Why would we need 100-150fps though? That processing power can go elsewhere. 50-70 fps isn't to bad at all if its all smooth. My card and monitor was getting me 110-144. It wasn't consistent though. I have since forced fps to 60 and the game runs pretty good at 60fps and 200% scaling.

[xXCARL1992Xx]

because of input delay, higher FPS more precise control, this is already visible between 30 VS 60 fps, and i have the bad feeling that the physics are tied to the fps but im not sure about that and i hope they are not tied to it

also VR, the more frames you get the better it is, might not be as grave as for games where you actively move instead of drive but still, 90+fps is better then 60 which isnt even an option on most headsets to set the refreshrate to (80 to 140/144 on the Valve Index)

[mmorselli]

Why would we need 100-150fps though? That processing power can go elsewhere. 50-70 fps isn't to bad at all if its all smooth. My card and monitor was getting me 110-144. It wasn't consistent though. I have since forced fps to 60 and the game runs pretty good at 60fps and 200% scaling.

VR

if I had to play in flat monitor only I wouldn't ask much more from this game, the framerate is decent and the aliasing, while bad, is tolerable

In VR, however, an RTX 3090 is not enough to get steady 90 fps, and because the head produces continuous micromovements the aliasing is intolerable

But in VR this game is not even the same game, what in flat monitor is just a nice entertainment in VR becomes a wonderful experience

[Berniyh]

Why would we need 100-150fps though? That processing power can go elsewhere. 50-70 fps isn't to bad at all if its all smooth. My card and monitor was getting me 110-144. It wasn't consistent though. I have since forced fps to 60 and the game runs pretty good at 60fps and 200% scaling.

50-70 fps is ok to drive, but only "ok".
If you've ever experienced driving a simulator with double that, then you'll know the difference.
In addition to what Carl already mentioned higher framerate will also improve the readability of signs (although that also depends on the amount of ghosting the display does) and generally it feels much smoother. At least that's my experience with other games.
So it's not necessary to have more fps than around 60 on a regular monitor, but if you can achieve that, the experience will be significantly improved.
It's a different story for VR, though. There you really want to have much higher framerate, because it can reduce motion sickness. It's a highly subjective thing (and personally I'm not affected fortunately), but I've heard multiple people tell me that they are affected more by motion sickness in case of lower fps.

[Some newbie driver]

I can fix the performance problem by buying more and more powerful computers, but I can't do anything for the awful Truck Simulator antialiasing

That's not true, because the load on the CPU increases faster than the improvements of the hardware. CPUs in past years are being powerful in absolute raw power indeed; but single core improvements aren't increasing so fast and there's the actual bottleneck for SCS games. You can find lot of people in this forums complaining about bad performance on computers with way higher specs than needed and not so far than other more humble computers. There's a lot of diminishing returns in overspending for SCS games at their current state. If we don't have proper AA is because it will still tank more performance, no matter the hardware you throw at it. SCS is more than aware they have to change the architecture of the game; that will give them the performance not just in FPS numbers, but the room to put in there proper AA and lots of other things.

@Berniyh the approach I guess SCS will do is to move multi-thread the game as a whole (now is partial and not affecting almost main actions) but graphics on single thread. So, what it would be optimal on DX11; leaving a whole tread only for graphics and the rest of the loads on other threads; being able to process a lot more things for the game apart of the graphics (specially, AI vehicles control; increasing traffic density and improving its behavior). If they later move to DX12/Vulkan they can keep the game like that while keep improving and eventually multi-thread also the graphic part itself to fully leverage those newer APIs. If they pretend to do that first; it means they had to do three things at the same time: multi-thread the whole game engine, multi-thread the graphic part of the engine and switch to new API. Seems a too risky gable IMHO.

Regards

[JoeAlex23]

I'm usually not the "one" to tell people how to play a game, but in my experience, it's pointless to push these games over 60FPS on a screen, for VR i totally get it, the more FPS the better, but for casual playing? 60 is enough, the reason a lot of people have problems with graphics in ETS2/ATS is because they push the game engine too hard, the more you push it, the worse it performs no matter how powerful your PC is.

[Some newbie driver]

For slow paced games like this ones, 60 should be more than enough IF (and that's the big IF) they are stable. Not only stable in the big number (a 60 steady FPS value) but also if the delay between every frame is stable. Even if some hardware of some players can achieve more than decent FPS values, i'm pretty sure the 99% and 95% percentiles figures are far away from the average FPS an the delay between frames would show too high dispersion to be perceived as smooth as we would like.

Then, of course, there's the VR scenario, with its own problems.

Regards

[Berniyh]

@Berniyh the approach I guess SCS will do is to move multi-thread the game as a whole (now is partial and not affecting almost main actions) but graphics on single thread. So, what it would be optimal on DX11; leaving a whole tread only for graphics and the rest of the loads on other threads; being able to process a lot more things for the game apart of the graphics (specially, AI vehicles control; increasing traffic density and improving its behavior). If they later move to DX12/Vulkan they can keep the game like that while keep improving and eventually multi-thread also the graphic part itself to fully leverage those newer APIs. If they pretend to do that first; it means they had to do three things at the same time: multi-thread the whole game engine, multi-thread the graphic part of the engine and switch to new API. Seems a too risky gable IMHO.

That would be the DCS approach.
But yeah, I do agree that it might be the better way, since the limitation of using a single core might currently be the biggest one.

Just note, that it can be done the other way round, as XP11/XP12 demonstrates. It's not great in the transition time, leaving a lot of potential unused, but it can still perform better than DX11 singlethreaded.
(Actually, XP12 is not singlethreaded as such, but it's strongly limited by the main CPU thread.)

I'm usually not the "one" to tell people how to play a game, but in my experience, it's pointless to push these games over 60FPS on a screen, for VR i totally get it, the more FPS the better, but for casual playing? 60 is enough, the reason a lot of people have problems with graphics in ETS2/ATS is because they push the game engine too hard, the more you push it, the worse it performs no matter how powerful your PC is.

Well it's something we could debate over for some time, personally I'd prefer a slightly higher framerate, but it's surely personal preference to some degree.
But think about it this way: if you lighten the burden on the CPU, you could use the freed ressources in another way as well. For improved optics like e.g. higher rendering distance or generally more of everything (e.g. more vehicles, more people, more animations, etc.), better anti-aliasing, filtering, post-processing and whatnot.
Right now, if you have detailed scenery, you might bring even powerful systems to a limit.