Linux Port – Feedback, Known Issues, and FAQ

works fine, ~75-80 fps in Ultra Preset, windowed 1440p
driver: nvidia-open-beta-dkms 530.41.03-1
(https://github.com/NVIDIA/open-gpu-kernel-modules
https://aur.archlinux.org/packages/nvidia-open-beta-dkms)


also doesn't seem to crash when changing graphics settings (at least it didn't crash when going from default to Ultra)


DPAD on my Logitech G29 still isn't detected


I just noticed, something is wrong with the renderer, the graphics (even though I have Ultra preset selected) look bad, there is no anti-aliasing (the toggle in the menu does nothing) (on my PC, RX 6700 XT)



(possibly because of previous issue): Seems to run a lot more stable, hasn't crashed since*! nah, still crashes all the time

*when I click on Quit Game, it crashes (also happened in 0.27)


so it just crashed out of nowhere, but I can't really interpret the logs (something with the shader/ the materials?) https://pastebin.com/efNiSWpM

 

Trying to launch the game on the Steam Deck after the latest update gives this error message:
Error: 0x3000001
Error description: Failed to get documents directory path

Update: running on Proton GE 7-53 fixes this issue

 

Both this native version and the Windows Vulkan version are broken for me. Trying to load any map ends up with a black screen and I need to reboot the PC.

AMD RX 6600 GPU, AMD Ryzen 5 5600 CPU and 16GB or RAM.

 

doesn't fix the problem, but with alt + print + e you can reset your system (it tells the kernel directly to terminate all processes except init, search SysRQ if you want to know more)

 

Hey I've noticed that vram usage went down by about 200mb with the last update in utah and I'm getting a bit more Fps than before, though that fps increase might be because of the new nvidia 530 driver.

However the new automation test center map performs horrible, while I get 90+ Fps on utah with 5 traffic vehicles, I only get 30-40 fps in the Automation test map with only one vehicle.

This is an older image of utah where you can see the vram usage before the update, it was 2.85GiB: https://i.imgur.com/GE7zQmu.png

Now with the new update, 2.64GiB: https://i.imgur.com/W5BcAXp.png

And this is automation with just one vehicle, vram usage is now higher than on either version of Utah, this map used to take about 2.5 GiB now it takes 2.88 GiB: https://i.imgur.com/YLUu2vd.png

In all the images the settings are low settings without AA and with partial shadows.

 

This just happened when I tried to load the K-Series on Utah with normal graphics settings and x16 antialiasing:

[ 5391.066774] [drm:amdgpu_job_timedout [amdgpu]] *ERROR* ring gfx_0.0.0 timeout, signaled seq=1
117042, emitted seq=1117045
[ 5391.067331] [drm:amdgpu_job_timedout [amdgpu]] *ERROR* Process information: process BeamNG.drive.x6 pid 20016 thread BeamNG.dri:cs0 pid 20017

Kernel: 6.2.6
Distribution: Artix Linux
GPU: RX 5700XT
Mesa version: 23.0.1

 

Wow! 0.28 fixed the crashing I was seeing on Linux. Additionally, I have noticed much improved performance, especially with AI vehicles. I can now easily run 50 parked cars + 12 active AI cars + pooling + ultra graphics preset@1440p at ~120 FPS on West Coast USA. (RX 7900 XTX + 64GB DDR4 + NVME SSD + Ryzen 9 3900X).

 

I'd also like to report that 0.28 is working wonderfully for me with the native Linux binary. I haven't played a lot yet with it, but the performance increases and stability improvements over previous versions are awesome!

I've got a 6900XT, 64GB DDR4, NVME and a Ryzen 9 3900X @ 1440p, running kernel 6.1 and mesa 22.2.5. I'm seeing really similar performance numbers as above. With Windows version in Proton, I get about 15-20FPS with AI traffic on. With the Linux version of 0.28, now it's around 120FPS. There's some hitching still, but I suspect that's shader compilation related since it got better the more that I played. Using the Windows version with the Vulkan renderer, I was able to get my GPU to timeout, but I haven't seen that with the Linux version yet. All in all, I'm really excited with the improvements I'm seeing, it's like a whole new game, thanks devs!

The only issue I've run into with the Linux version so far is that it appears that my XBox One controller has an inverted Y axis which makes the radial menus a little awkward. I'm not sure if this is a configuration issue on my end or something I can toggle in the game.

 

Echoing the comments above, 0.28 has fixed the crashing issue for me as well. Running very smoothly again at more than 100fps.

A great update, thank you team!

 

you should be able to invert the axis for each binding individually in the Controls Settings ingame

 

Anyone else getting this?

 

try with RADV_DEBUG=zerovram
(RADV_DEBUG=zerovram ./BeamNG.drive.x64)

 

still an issue, let me try using mesa-git
Edit: nope, doesn't fix the issue, wonder what broke since I got another person to try it and it works on their machine

 

Try running with the environment variable RADV_PERFTEST=gpl if you are running Mesa 23.0. It will become the default in 23.1 but fixes this

Also, I have mine added to Steam as a non-Steam game which allows it to use the Steam controller APIs. Everything "just works"

 

0.28 is unstable for me. I'm able to get to the opening menu and select a free roam but the kernel GPU panics 90% of the time during the loading screen. I often hear sound of the truck engine before the screen goes black. AMD 6750XT GPU.

Dmesg output:

[ 48.839928] amdgpu 0000:03:00.0: amdgpu: [gfxhub] page fault (src_id:0 ring:24 vmid:6 pasid:32772, for process BeamNG.drive.x6 pid 1046 thread BeamNG.dri:cs0 pid 1047)
[ 48.839930] amdgpu 0000:03:00.0: amdgpu: in page starting at address 0x0000535b93af5000 from client 0x1b (UTCL2)
[ 48.839930] amdgpu 0000:03:00.0: amdgpu: GCVM_L2_PROTECTION_FAULT_STATUS:0x00000000
[ 48.839931] amdgpu 0000:03:00.0: amdgpu: Faulty UTCL2 client ID: CB/DB (0x0)
[ 48.839932] amdgpu 0000:03:00.0: amdgpu: MORE_FAULTS: 0x0
[ 48.839932] amdgpu 0000:03:00.0: amdgpu: WALKER_ERROR: 0x0
[ 48.839933] amdgpu 0000:03:00.0: amdgpu: PERMISSION_FAULTS: 0x0
[ 48.839933] amdgpu 0000:03:00.0: amdgpu: MAPPING_ERROR: 0x0
[ 48.839934] amdgpu 0000:03:00.0: amdgpu: RW: 0x0
[ 58.964198] [drm:amdgpu_job_timedout [amdgpu]] *ERROR* ring gfx_0.0.0 timeout, signaled seq=4588, emitted seq=4590
[ 58.964303] [drm:amdgpu_job_timedout [amdgpu]] *ERROR* Process information: process BeamNG.drive.x6 pid 1046 thread BeamNG.dri:cs0 pid 1047
[ 58.964377] amdgpu 0000:03:00.0: amdgpu: GPU reset begin!
[ 59.095716] amdgpu 0000:03:00.0: amdgpu: MODE1 reset
[ 59.095717] amdgpu 0000:03:00.0: amdgpu: GPU mode1 reset
[ 59.095766] amdgpu 0000:03:00.0: amdgpu: GPU smu mode1 reset
[ 59.602274] amdgpu 0000:03:00.0: amdgpu: GPU reset succeeded, trying to resume

I'm not running a DM so it returns to the terminal and I see this error along with the xserver stdout dump:

amdgpu: The CS has been rejected but the context isn't robust.

GDB does indicate that "Program terminated with signal SIGSEGV, Segmentation fault." but none of the thread backtraces show any errors. They're all hanging out in pthread_cond_timedwait at failure except for one thread in sigtimedwait.

I've tried various Mesas (22.3, 23) and kernels (5.15, 6.2) with same results. Beam and Xorg logs look fine. Have tried many times clearing out ~/.local/share/BeamNG.drive/0.28 with no luck. No other games do this and Steam version also runs fine.

Any suggestions??

 

The hotfix didn't fix my game. It still breaks my entire graphic session in Arch Linux.

The same problem with the Vulkan version with Proton.

 

The 0.28 linux port seems splendid so far, but I have issues with only certain map mods not getting detected.

Launching the game also throws an error in the top right: "Error in mounting mods"

Aside from that, everything works really good. Can't wait for further Vulkan/Linux improvements