VR - experimental support

[xXCARL1992Xx]

if there is no AI there is no other vehicle on the road except you and random events when you have enabled it

[KingOfTheRoad77]

Ah okay.
Well, that finally would really explain why FPS drops are so severe in settlements and cities with a lot of crossings, traffic lights and possibilities for AI Routing.

[D4NTRUCK]

I am using various bus mods in VR and when I barely turn my head to either side it does a jerky rotation.

For example: when I turn my head a little to the left, the movement to move my head out of the cabin begins.
Is there a way to correct that? To remove it?

[dhrto]

Yep, see first post in this thread where it is explained.

The interior camera is using a limited variant of angle-dependent movement of the head from the normal game. It only keeps the window lookout and moves its start to about 85 degress. The movement can be disabled entirely by setting g_hmd_no_artifical_movement profile cvar to 1 before starting the game.

[D4NTRUCK]

Thank you very much!!

[KingOfTheRoad77]

Happy to do some formal benchmarking this weekend. Maybe between us in this thread we could establish some VR benchmarking discipline where we drive the same route (same time leaving etc) so like Munich to Rome in ETS at 10 AM or Denver to Billings or something.

After having the strange feeling that things became even worse than Montana since 1.46 / Texas, I finally made a benchmark savegame for a fix delivery route from Dallas to Lufkin:

[ external image ]

General conditions:
- Pimax 8K-X
- Ryzen 5600X, 32GB DDR4-3600, RTX 3080
- OpenXR @90% resolution = 3.416 x 2.844
- Start at daytime, 9am
- (Perhaps) important mods: Weather 2.3, Grime's Early Autumun, Daniel's Random Events
- I directly included both weight stations (waypoints 2 & 4) into routing since it's 60-80% sure I'll catch at least one - so better including both directly for better comparison of the results.
- Always driving according to traffic rules (preferably constant influence of total gaming / driving time for avg. FPS)
- Measurement with RTSS plug-in for HWInfo64


First, due to last conversations here, I especially wanted to find out whether AI traffic had some (significant) impact or not on performance.

So, here is the result:

g_traffic="0.0"
=> Avg. FPS 58
(First half of the trip sunny, then overcast, a little drizzle from Centerville till Lufkin)

g_traffic="1.0"
=> Avg. FPS 52
(First sunny, then overcast, then partly sunny again, no drizzle or rain)

(PS: I did another trial yesterday with g_traffic="0.3" on two different routes between Fort Worth and Dallas before having defined this one above => Avg. FPS 59)

=> So, there is an impact of AI - but not as critical I'd get stable 75 FPS without any traffic - far way away from it.

=> I am still not really sure where's the major bottleneck for me - I'm still thinking, GPU is more important for my Pimax 8K-X in a first step than the CPU...


I'd really appreciate if anyone would participate in such benchmarking and perhaps defines a delivery from Dallas to Lufkin, too, posting the results for g_traffic="0.0" and "1.0" in a similar way here afterwards

[Shiva]

What happens if you do the route with a start time around 5 in the morning?

[KingOfTheRoad77]

What happens if you do the route with a start time around 5 in the morning?

I suppose the GPU will implode!!
... or was it the truck in the game that exploded...?!?
Can't remember
___

I've got a 4080 today for test purposes.

Good news:
It behaves absolutely as expected under 3DMark: 20,2k total score, 27,5k graphics score. A little lower than the ones in my screenshot on last page - but they all have a little higher GPU clocks. So I am completely fine there

But - 2 bad news:
1. Absolutely no significant gain in ATS @ VR!!
=> On my benchmarking route from Dallas to Lufkin => 55 FPS - instead of 52!
And the 4080 is almost sleeping - 220-250 Watts consumption only, max load <75%, temperature 55-60°...

=> So, although my tests with the 5800X3D earlier this year weren't convincing (@ 3080), and although 3DMark now scales as predicted with the 4080 - ATS @ VR simply is "ignoring" everything so far...


2. Furthermore, I've got a very strange bug with the 4080 in ATS-VR:

When moving my head and looking around, the image in the HMD shifts a little bit in the same direction for the first milliseconds. Not much, perhaps only a couple of millimeters - but directly noticeable and causing slight nausea after a couple of minutes driving.
Happens with 3D image while driving as well as in the menus. When I move my head like a lying "8", the whole image slightly "swings" with a little delay...
(I don't know what's the technical expression for this.)

I checked the lighthouse; loosened and fixed the screws once again => no change.
I installed latest Pimax drivers, hoping there could have been some bug fix for the new RTX 40x0 => no change.
I enabled low latency mode in NVidia GeForce Panel => no change.
I cleaned my Pimax, thought the sensors might be a bit dusty/dirty => no change.

I put back the 3080 => and the problem is gone!

=> So it is definitely the 4080 apparently having some issues with VR, OpenXR or at least with Pimax 8K-X, lighthouse tracking, or whatever.


So, once again not very successful...

And I don't think the 7900 XTX would behave much different. (At least in terms of FPS; concerning the "swinging image" bug, I hope, it's just a temporary driver bug of the 4080.)


So, I think it's time again to order a new 5800X3D once again, now, that it is much cheaper than half a year ago, and to redo some driving tests...
At the time of my first trials, ATS was on 1.43 or 1.44, and I was using fholger's VR mod under SteamVR - whereas some of you already were able to use OpenXR with your Reverbs. So, perhaps, the current combination of ATS 1.46 (including Montana and Texas now) and OpenXR will do the trick.


I like and love working on my PC and its hardware - but it's a bit silly that I just don't have enough money to just order everything I need once and for all, do my tests, keep what's working best, and then I send back or sell via ebay everything else only once and without wasting to much money...

Dammit...

[Shiva]

Cheers for testing.

[Phantom_Mark]

I saw virtually no difference going from the 3080ti to a 4090, upgrading from a Z590 Motherboard with 64gb of DDR4 3600 and a 10900k to a Z790 and 32gb DDR5 7200 and the 13900k, at that point it is almost the fastest possible spec you can buy pretty much ? fingers crossed........I won't hold my breath tho, my experience at the moment in VR is at least stable and very playable to some degree, so anything better is better.