Update – DX11 workaround testing
I tested forcing DX11 with the launch option -dx11.
Results
DX11 (-dx11):
Successfully forces LogRHI: Using Forced RHI: D3D11 + DXVK
Avoids the original WaveSize crash
Game runs longer without an immediate hard crash
However, the DX11 path is also unstable:
Log contains ~232 000 access violations (Exception 0xc0000005) all hitting the same location:textlibnvidia-glcore.so.610.43.03 + 0xe44c64
These are continuously generated inside the NVIDIA driver (loaded via pressure-vessel).
The old periodic EXCEPTION_ILLEGAL_INSTRUCTION (0xc000001d) in gp.dll is still present.
So while -dx11 works around the original D3D12/WaveSize issue, it currently triggers a heavy stream of faults inside the NVIDIA 610.43.03 driver instead.
System: RTX 4080 + NVIDIA 610.43.03 + CachyOS (kernel 7.1.5) + Proton Experimental bleeding-edge (2026-08-01).
EDIT:
Dx11 crashes as well, just takes little longer:)
proton experimentalx2 2026-08PROTON_LOG=1x1 2026-08gp.dllx2 2026-080xc0000005x2 2026-080xc000001dx2 2026-080x80070057x1 2026-080x887a0005x1 2026-08
Compatibility Report
System Information
Full system information:
Computer Information:
Manufacturer: Gigabyte Technology Co., Ltd.
Model: X670 GAMING X AX
Form Factor: Desktop
No Touch Input Detected
Processor Information:
CPU Vendor: AuthenticAMD
CPU Brand: AMD Ryzen 9 7950X 16-Core Processor
CPU Family: 0x19
CPU Model: 0x61
CPU Stepping: 0x2
CPU Type: 0x0
Speed: 6074 MHz
32 logical processors
16 physical processors
Hyper-threading: Supported
FCMOV: Supported
SSE2: Supported
SSE3: Supported
SSSE3: Supported
SSE4a: Supported
SSE41: Supported
SSE42: Supported
AES: Supported
AVX: Supported
AVX2: Supported
AVX512F: Supported
AVX512PF: Unsupported
AVX512ER: Unsupported
AVX512CD: Supported
AVX512VNNI: Supported
SHA: Supported
CMPXCHG16B: Supported
LAHF/SAHF: Supported
PrefetchW: Unsupported
BMI1: Supported
BMI2: Supported
F16C: Supported
FMA: Supported
Operating System:
CachyOS (64 bit)
Kernel Name: Linux
Kernel Version: 7.1.5-1-cachyos
X Server Vendor: The X.Org Foundation
X Server Release: 12401013
X Window Manager: KWin
Steam Runtime Version: steam-runtime_1.0.20260714.251828
Client Information:
Version: 1785347151
Video Card:
Driver: NVIDIA Corporation NVIDIA GeForce RTX 4080/PCIe/SSE2
Driver Version: 4.6.0 NVIDIA 610.43.03
OpenGL Version: 4.6
Desktop Color Depth: 24 bits per pixel
Monitor Refresh Rate: 175 Hz
VendorID: 0x10de
DeviceID: 0x2704
Number of Monitors: 1
Number of Logical Video Cards: 1
Primary Display Resolution: 3440 x 1440
Desktop Resolution: 3440 x 1440
Primary Display Size: 31.89" x 13.78" (34.72" diag)
Primary VRAM: 16376 MB
Memory:
RAM: 31687 Mb
I confirm:
Symptoms
Game launches and runs for approximately 90–120 seconds, then crashes.
Primary failure sequence:
err:vkd3d-proton:d3d12_device_validate_shader_meta: Required WaveSize range [64, 64], but supported range is [32, 32].
warn:vkd3d-proton:d3d12_pipeline_state_init_compute: Failed to create Vulkan compute pipeline, hr 0x80070057
LogD3D12RHI: Error: Failed to create pipeline state with combined hash ..., error 80070057
text2. Shortly afterwards the device is lost:
err:vkd3d-proton:vkd3d_wait_for_gpu_timeline_semaphore: Failed to wait for Vulkan timeline semaphore, vr -4.
warn:vkd3d-proton:d3d12_device_mark_as_removed: Device ... is lost (reason 0x887a0005, "VK_ERROR_DEVICE_LOST")
text3. Cascade of access violations (
Exception 0xc0000005) → process exits.Additional recurring issue (caught by exception handler, not immediately fatal):
EXCEPTION_ILLEGAL_INSTRUCTION(0xc000001d) every ~20 seconds inMistfallHunter\Binaries\Win64\GSDK_I18N\Steam\gp.dllGame is Unreal Engine 5, forced to D3D12 (SM6), uses EOS.
This is the probably WaveSize 64 vs 32 mismatch that has been seen in several other UE5 titles on recent NVIDIA drivers + vkd3d-proton.
Reproduction
PROTON_LOG=1 %command%Proton log:
(https://files.catbox.moe/iu3c6b.xz)
Might be related UE5 titles on NVIDIA: