With vkd3d-proton master, you can try VKD3D_CONFIG=defer_resource_destruction. It works around some game bugs in Valhalla and I wouldn't be surprised if Shadows inherits the same bugs.
With vkd3d-proton master, you can try VKD3D_CONFIG=defer_resource_destruction. It works around some game bugs in Valhalla and I wouldn't be surprised if Shadows inherits the same bugs.
I'll try it out next time I play the game. I've finished it for now though so I don't know when that will be. Thanks for the tip though. :)
@Dwarsen have you found a working solution for this? I'm running an RTX 2080 Ti (575.64.05) Proton experimental-10.0-20250724b on Nobara Linux 42 and have non-recovering freezes when playing. I've tried many different Proton versions, Pyroveil, GPU drivers, environment variables, DLL overrides, in-game settings, etc. and nothing's worked.
@Dwarsen have you found a working solution for this? I'm running an RTX 2080 Ti (575.64.05) Proton experimental-10.0-20250724b on Nobara Linux 42 and have non-recovering freezes when playing. I've tried many different Proton versions, Pyroveil, GPU drivers, environment variables, DLL overrides, in-game settings, etc. and nothing's worked.
I haven't played the game that much the last few weeks because I've kind of finished it. For what it's worth, I have been running with the workaround Hans-Kristian suggested and I haven't had any hangs since, but I can't prove that it's helped. You could try it out? Proton-GE 10-11 should have a vkd3d version new enough to contain it.
Also, I did experiment a bit with settings and noticed that activating frame-gen will cause a hang almost immediately (as in, instead of being able to play for hours we are now talking about seconds), so if you use that you should turn it off.
I am experiencing GPU hangs in AC Shadows as well, but the signature on RDNA 4 (RX 9070 XT, gfx1201) with Ray Tracing and HDR enabled appears to diverge from the existing reports. Instead of a single CPC-based fault at a near-null address, I am seeing four rapid page faults through the TCP (Texture Cache Per-pipe) unit at high memory addresses. These faults are logged with WALKER_ERROR: 0x5 and PERMISSION_FAULTS: 0x5, eventually escalating into a comp_1.0.1 compute ring timeout. While the driver initially recovers through a reset, a second timeout occurs about 11 seconds later, leading to a total system hang.
amdgpu 0000:03:00.0: amdgpu: [gfxhub] page fault (src_id:0 ring:173 vmid:6 pasid:32819) amdgpu 0000:03:00.0: amdgpu: Process ACShadows.exe pid 8255 thread vkd3d_queue pid 8475 amdgpu 0000:03:00.0: amdgpu: in page starting at address 0x00008003c405a000 from client 10 amdgpu 0000:03:00.0: amdgpu: GCVM_L2_PROTECTION_FAULT_STATUS:0x0064115A amdgpu 0000:03:00.0: amdgpu: Faulty UTCL2 client ID: TCP (0x8) amdgpu 0000:03:00.0: amdgpu: WALKER_ERROR: 0x5 amdgpu 0000:03:00.0: amdgpu: PERMISSION_FAULTS: 0x5 amdgpu 0000:03:00.0: amdgpu: ring comp_1.0.1 timeout, signaled seq=765961, emitted seq=765963 amdgpu 0000:03:00.0: amdgpu: Ring comp_1.0.1 reset succeeded amdgpu 0000:03:00.0: [drm] device wedged, but recovered through reset
Interestingly, testing with VKD3D_CONFIG=defer_resource_destruction shifts the failure state entirely. The page faults move to a null address (0x0000000000000000) and the timeout moves to the graphics ring (gfx_0.0.0). In this scenario, the MES firmware becomes completely unresponsive to reset commands, failing with MES ring buffer is full and requiring a hard reboot.
amdgpu 0000:03:00.0: amdgpu: [gfxhub] page fault (src_id:0 ring:24 vmid:5 pasid:32817) amdgpu 0000:03:00.0: amdgpu: in page starting at address 0x0000000000000000 from client 10 amdgpu 0000:03:00.0: amdgpu: ring gfx_0.0.0 timeout, signaled seq=4250597, emitted seq=4250599 amdgpu 0000:03:00.0: amdgpu: MES(1) failed to respond to msg=RESET amdgpu 0000:03:00.0: amdgpu: Ring gfx_0.0.0 reset failed amdgpu 0000:03:00.0: amdgpu: GPU reset begin! amdgpu 0000:03:00.0: amdgpu: MES(1) failed to respond to msg=INVALIDATE_TLBS amdgpu 0000:03:00.0: amdgpu: MES ring buffer is full
Edit; For added context, my launch flags are: PROTON_ENABLE_HDR=1 PROTON_USE_WAYLAND=1 FSR4=1 VKD3D_CONFIG=defer_resource_destruction %command%, i'm running on GE-Proton10-34
proton experimentalx2 2025-08PROTON_ENABLE_HDR=1x1 2026-05PROTON_USE_WAYLAND=1x1 2026-05VKD3D_CONFIG=defer_resource_destructionx1 2026-05VKD3D_CONFIG=defer_resource_destruction.x2 2025-08
Have been getting random GPU hangs in AC Shadows; sometimes the GPU manages to reset, sometimes not, and when it doesnt the entire system is brought down and a hard reset is required.
Happened sporadically at first, now happens a lot but seemingly still randomly. Can be seconds after starting the game, can take hours.
Tried different mesa and proton versions, none seem to work perfectly. Changing proton version to force a shader recompilation does seem to help for a little while though, but that is just my guess.
Attaching a log from journal showing some of the amdgpu errors. I'm not 100% sure it is a vkd3d issue but as far as I've understood the errors in question shouldn't be related to the kernel driver?
Software information
Assassins Creed Shadows, mostly max settings.
System information
Log files
journal-ac-gpuhang.txt