TERA Online (x86 33.08 client) crash with DXVK 2.5.X

[Kiciuk]

Old TERA Online client(UE3/UDK) crashes on launch after it spawn window and initialize cursor.
It appears to work with DXVK 2.4

Game is being launched with launcher and runs some kind of custom anti-cheat(afterburner doesn't show anything) so i have no idea how to attach RenderDoc or Apitrace so i cannot provide dumps.
In log files i also attached log from working 2.4 version it differs a bit it looks like.

System information

• OS: Windows 10
• GPU: Radeon RX 570/RX Vega 11(different systems both affected)
• Driver: Adrenaline 24.9.1(Same driver for both gpus)
• DXVK version: 2.5.1-2.5.3

Log files

TERA_d3d9.log
TERA_d3d9 _dxvk_2_4.log

[Blisto91]

Hi there.
Have you tried making a apitrace? Or is it just an assumption that it won't work? I would be surprised if it allowed the dxvk files but not the apitrace ones

[Kiciuk]

Hi there. Have you tried making a apitrace? Or is it just an assumption that it won't work? I would be surprised if it allowed the dxvk files but not the apitrace ones

Not 100% sure but game is being launched on windows and it works with default and if i do as it says to do for apitrace(so place its dlls in Binaries instead of DXVK) then i would be dumping state of D3D9 native renderer which works, and isn't apitrace of dxvk what im supposed to dump?

[Blisto91]

No in most cases that is fine as we just replay it with dxvk locally where it should hopefully reproduce then

[Kiciuk]

No in most cases that is fine as we just replay it with dxvk locally where it should hopefully reproduce then

Okay managed to do a dump, is 7z compression okay?
If yes will upload it to gdrive soon

[Blisto91]

That is fine yes

[Kiciuk]

That is fine yes

https://drive.google.com/file/d/1VFY7PsFh9TQHydJhphI6IFu-hBxIk7AV/view?usp=drive_link

[K0bin]

The trace replays fine. No crashes with it.

[Kiciuk]

Could be driver issue?
It dies after spawning cursor for me..

[doitsujin]

Could be anything, if the log is anything to go by we literally die during device creation, which is pretty much outside of our control at that point.

It's also extremely annoying to work with games that have literally been shut down already, so I'd honestly just suggest sticking to 2.4.1 if that works for you, I don't really see a good way for us to figure this out remotely.

[Blisto91]

I haven't been able to reproduce this in game either. I've started the same game version on Windows with both my Intel b580 and AMD rdna2 iGPU using dxvk 2.5.3. Also tried same driver version on AMD

[Kiciuk]

I haven't been able to reproduce this in game either. I've started the same game version on Windows with both my Intel b580 and AMD rdna2 iGPU using dxvk 2.5.3. Also tried same driver version on AMD

Didn't amd branch-out polaris from rdna driver?
I also tried older driver 24.3.1 and it also fails.
From replaying using qapitrace i get something like this:

Probably last resort of rulling amd driver bug would be trying to run on linux/mesa but it ain't that easy so far.

I was messing around renderdoc and some d3dretrace cmd had some " a Vulkan feature bufferDeviceAddressCaptureReplay is not available"