protonscr

Possible regression concerning F1 2020 shaders in v3.0

vkd3dclosed
HansKristian-Work/vkd3d-proton#2849 · opened 2026-02-26 by digimezzo · updated 2026-02-27 · 2 comments · github
Ddigimezzo 2026-02-26 github

Using vkd3d-proton-3.0 with F1 2020, (DX11 as well as DX12) I'm experiencing the artifacts that are described in issue #611. I'm running F1 2020 via Bottles and have created a Bottle that uses vkd3d-proton-2.14.1 as a mattter of test, and with vkd3d-proton-2.14.1 I don't get these artifacts. Is there a regression concerning the shader workaround in vkd3d-proton-3.0?

Software information

F1 2020

System information

  • OS: Arch Linux x86_64
  • GPU: NVIDIA GeForce RTX 4070
  • Wine version: ge-proton10-32
  • VKD3D-Proton version: 3.0
Ddoitsujin maintainer 2026-02-26 github

Did it actually break for DX11? The workaround for the game should still apply there, but I'll admit that besides looking at shader disassembly I haven't really tested it when we moved to the new compiler.

For DX12, the workaround kinda got lost when we moved to the new shader compiler, there's currently no way to enable it because there's no interface to pass through app quirks to dxbc-spirv from dxil-spirv, and I don't know how to really make that work given that dxbc-spirv is invoked very early.

Ddigimezzo 2026-02-26 github

Thank you for the quick response. I was convinced that it broke for DX11 as well, as I tested it with both the DX11 as the DX12 executable over the course of 2 days. But I must have made a mistake. After your reply, I decided to try with DX11 once more, and it doesn't break with DX11, it only breaks with DX12. My apologies for the confusion.

Proton versions