Squad (UE5) - GPU crash on respawn
Having similar issue for few weeks. Mine is random crashes (with freeze sometimes). Proton-GE 10-34 ,10-33 and 10-32 are all having this issue.
Having similar issue for fews weeks. Mine is random crashes (with freeze sometimes). Both Proton-GE 10-34 and 10-33 having this issue.
Just had the spawn crash 10s ago lol.
It still happens. It could be the game’s fault or another component, such as the kernel, Mesa, firmware, or vkd3d-proton. You can try updating vkd3d-proton to the latest version — just download Proton Experimental (from .steam/steam/steamapps/common/Proton - Experimental/files/lib/wine/vkd3d-proton/ ) and copy the newer files into Proton-GE 10-34 ( .steam/steam/compatibilitytools.d/GE-Proton10-34/files/lib/wine/vkd3d-proton/ ). I get fewer crashes after doing that, but it still happens occasionally.
Some people say it’s caused by FSR4, but playing that game without it is rough — FSR3 looks terrible, and TSR causes a big FPS drop. I feel older FSR 4 have less crashes, but latest vkd3d-proton with 4.1 its much better. Version from proton-ge 10-34 its really bad with FSR 4.1 for me.
Of course, you can see many fixes in Mesa, the kernel, firmware for AMDGPU, or memory management in vkd3d-proton. Which one actually gives me fewer crashes, I have no idea.
Another setting you can try is increasing vm.max_map_count, like on SteamOS: sudo sysctl -w vm.max_map_count=2147483642
This is the same value used in SteamOS and may help with stability in some games on UE5 like Squad. But the feeling for me is: great, it helped — only one crash. Then suddenly, three crashes in a row. I must say i dont often give up and have fewer respawns :) but it still happens.
Actually iam waiting for 10.4 patch and UE5.7 upgrade.
It still happens. It could be the game’s fault or another component, such as the kernel, Mesa, firmware, or vkd3d-proton. You can try updating vkd3d-proton to the latest version — just download Proton Experimental (from .steam/steam/steamapps/common/Proton - Experimental/files/lib/wine/vkd3d-proton/ ) and copy the newer files into Proton-GE 10-34 ( .steam/steam/compatibilitytools.d/GE-Proton10-34/files/lib/wine/vkd3d-proton/ ). I get fewer crashes after doing that, but it still happens occasionally.
Some people say it’s caused by FSR4, but playing that game without it is rough — FSR3 looks terrible, and TSR causes a big FPS drop. I feel older FSR 4 have less crashes, but latest vkd3d-proton with 4.1 its much better. Version from proton-ge 10-34 its really bad with FSR 4.1 for me.
Of course, you can see many fixes in Mesa, the kernel, firmware for AMDGPU, or memory management in vkd3d-proton. Which one actually gives me fewer crashes, I have no idea.
Another setting you can try is increasing vm.max_map_count, like on SteamOS: sudo sysctl -w vm.max_map_count=2147483642
This is the same value used in SteamOS and may help with stability in some games on UE5 like Squad. But the feeling for me is: great, it helped — only one crash. Then suddenly, three crashes in a row. I must say i dont often give up and have fewer respawns :) but it still happens.
Actually iam waiting for 10.4 patch and UE5.7 upgrade.
I'm having the issue without using FSR4 tho. Tried this launch option but still crashes. I'm frustrated.
I've been facing a similar, if not the same issue, though I always had FSR off.
Unless it was somehow activated while ignoring the settings menu, it is not specific to FSR. At least not for me.
After testing various config options and reading some logs and crashdumps, I've come to believe that Squad is freeing live resources and then trying to use them (use-after-free), causing the crashes.
My guess is that on Windows this does not show up (as frequently?) because the drivers/D3D12 are more lenient about resource destruction/more aggressively defer it.
So lease try this launch option:
VKD3D_CONFIG=defer_resource_destruction %command%
This is what is used for AC Valhalla in application_override[] at /libs/vk3d/device.c:609.
It seems to be working for me so far and I've had no crashes since I applied it, but do be aware that I might be totally wrong about this or that it may not work for everybody. Please share whether it helped or not.
Proton versions
ge-proton10-34x2 2026-04proton experimentalx2 2026-04
Launch options
VKD3D_CONFIG=defer_resource_destructionx1 2026-08
Software information
I am experiencing an issue in Squad (Unreal Engine 5) where the game crashes during respawn on the hub, when the player is moved to another part of the map (heavy streaming / reload event).
During normal gameplay the game generally runs well without major issues, but sometimes there are short freezes lasting around 1–2 seconds. These do not crash the game and it continues running normally afterward.
In the past I did have similar crashes occasionally, but they were rare. Disabling Steam pre-shaders or clearing mesa shader caches usually helped, especially after installing new Mesa drivers, kernel or firmware.
Recently, after updating to a newer vkd3d-proton version (via Proton-GE 10-34), the behavior changed. The first respawn usually works fine, but every next respawn results in a crash.
Rolling back to an earlier vkd3d-proton version (Proton-GE 10-33) resolves the issue — during last full gameplay session I did not experience crashes, only occasional short freezes without GPU reset.
System information
Log files
amdgpu 0000:03:00.0: amdgpu: [gfxhub] page fault (src_id:0 ring:24 vmid:7 pasid:32804)
kernel: amdgpu 0000:03:00.0: amdgpu: Process GameThread pid 324969 thread vkd3d_queue pid 325393
kernel: amdgpu 0000:03:00.0: amdgpu: in page starting at address 0x000080000b800000 from client 10
kernel: amdgpu 0000:03:00.0: amdgpu: GCVM_L2_PROTECTION_FAULT_STATUS:0x00701431
kernel: amdgpu 0000:03:00.0: amdgpu: Faulty UTCL2 client ID: SQC (data) (0xa)
kernel: amdgpu 0000:03:00.0: amdgpu: MORE_FAULTS: 0x1
kernel: amdgpu 0000:03:00.0: amdgpu: WALKER_ERROR: 0x0
kernel: amdgpu 0000:03:00.0: amdgpu: PERMISSION_FAULTS: 0x3
kernel: amdgpu 0000:03:00.0: amdgpu: MAPPING_ERROR: 0x0
kernel: amdgpu 0000:03:00.0: amdgpu: RW: 0x0
kernel: amdgpu 0000:03:00.0: amdgpu: [gfxhub] page fault (src_id:0 ring:24 vmid:7 pasid:32804)
kernel: amdgpu 0000:03:00.0: amdgpu: Process GameThread pid 324969 thread vkd3d_queue pid 325393
kernel: amdgpu 0000:03:00.0: amdgpu: in page starting at address 0x000080000b800000 from client 10
kernel: amdgpu 0000:03:00.0: amdgpu: [gfxhub] page fault (src_id:0 ring:24 vmid:7 pasid:32804)
kernel: amdgpu 0000:03:00.0: amdgpu: Process GameThread pid 324969 thread vkd3d_queue pid 325393
kernel: amdgpu 0000:03:00.0: amdgpu: in page starting at address 0x000080000b800000 from client 10
kernel: amdgpu 0000:03:00.0: amdgpu: [gfxhub] page fault (src_id:0 ring:24 vmid:7 pasid:32804)
kernel: amdgpu 0000:03:00.0: amdgpu: Process GameThread pid 324969 thread vkd3d_queue pid 325393
kernel: amdgpu 0000:03:00.0: amdgpu: in page starting at address 0x000080000b800000 from client 10
kernel: amdgpu 0000:03:00.0: amdgpu: PERMISSION_FAULTS: 0x3
kernel: amdgpu 0000:03:00.0: amdgpu: MAPPING_ERROR: 0x0
kernel: amdgpu 0000:03:00.0: amdgpu: RW: 0x0
kernel: amdgpu 0000:03:00.0: amdgpu: [gfxhub] page fault (src_id:0 ring:24 vmid:7 pasid:32804)
kernel: amdgpu 0000:03:00.0: amdgpu: Process GameThread pid 324969 thread vkd3d_queue pid 325393
kernel: amdgpu 0000:03:00.0: amdgpu: in page starting at address 0x000080000b800000 from client 10
kernel: amdgpu 0000:03:00.0: amdgpu: [gfxhub] page fault (src_id:0 ring:24 vmid:7 pasid:32804)
kernel: amdgpu 0000:03:00.0: amdgpu: Process GameThread pid 324969 thread vkd3d_queue pid 325393
kernel: amdgpu 0000:03:00.0: amdgpu: in page starting at address 0x000080000b800000 from client 10
kernel: amdgpu 0000:03:00.0: amdgpu: [gfxhub] page fault (src_id:0 ring:24 vmid:7 pasid:32804)
kernel: amdgpu 0000:03:00.0: amdgpu: Process GameThread pid 324969 thread vkd3d_queue pid 325393
kernel: amdgpu 0000:03:00.0: amdgpu: in page starting at address 0x000080000b800000 from client 10
kernel: amdgpu 0000:03:00.0: amdgpu: [gfxhub] page fault (src_id:0 ring:24 vmid:7 pasid:32804)
kernel: amdgpu 0000:03:00.0: amdgpu: Process GameThread pid 324969 thread vkd3d_queue pid 325393
kernel: amdgpu 0000:03:00.0: amdgpu: in page starting at address 0x000080000b800000 from client 10
kernel: amdgpu 0000:03:00.0: amdgpu: [gfxhub] page fault (src_id:0 ring:24 vmid:7 pasid:32804)
kernel: amdgpu 0000:03:00.0: amdgpu: Process GameThread pid 324969 thread vkd3d_queue pid 325393
kernel: amdgpu 0000:03:00.0: amdgpu: in page starting at address 0x000080000b800000 from client 10
kernel: amdgpu 0000:03:00.0: amdgpu: Dumping IP State
kernel: amdgpu 0000:03:00.0: amdgpu: Dumping IP State Completed
steam[323468]: radv/amdgpu: The CS has been cancelled because the context is lost. This context is innocent.
steam[323468]: radv: GPUVM fault detected at address 0x80000b800000.
steam[323468]: GCVM_L2_PROTECTION_FAULT_STATUS: 0x701431
steam[323468]: CLIENT_ID: (SQC (data)) 0xa
steam[323468]: MORE_FAULTS: 1
steam[323468]: WALKER_ERROR: 0
steam[323468]: PERMISSION_FAULTS: 3
steam[323468]: MAPPING_ERROR: 0
steam[323468]: RW: 0
kernel: amdgpu 0000:03:00.0: amdgpu: [drm] AMDGPU device coredump file has been created
kernel: amdgpu 0000:03:00.0: amdgpu: [drm] Check your /sys/class/drm/card1/device/devcoredump/data
kernel: amdgpu 0000:03:00.0: amdgpu: ring gfx_0.0.0 timeout, signaled seq=14444862, emitted seq=14444864
kernel: amdgpu 0000:03:00.0: amdgpu: Process GameThread pid 324969 thread vkd3d_queue pid 325393
kernel: amdgpu 0000:03:00.0: amdgpu: Starting gfx_0.0.0 ring reset
kernel: amdgpu 0000:03:00.0: amdgpu: Ring gfx_0.0.0 reset succeeded
kernel: amdgpu 0000:03:00.0: [drm] device wedged, but recovered through reset
EDIT:
It also seems to happen on Proton-GE 10-33, but mainly on a specific map (Yehorivka). After joining the game, the crash tends to occur around every second respawn rather than always on the immediate next one.
In general, during respawn there is a noticeable short pause while textures/assets are loading, and it looks like this streaming/reload moment is what triggers the issue. During normal gameplay I can also observe brief stalls, but without a crash.
I will try to test older Proton-GE versions more carefully.