protonscr

Starfield GPU Hang

vkd3dopen
HansKristian-Work/vkd3d-proton#3130 · opened 2026-06-27 by Selene0623 · updated 2026-08-03 · 4 comments · github
SSelene0623 2026-06-27 github

Software information

Starfield v1.16.244, heavily modded (~200 ESM mods, 14 SFSE plugins). SFSE v0.2.21. Custom shaders via SFShaderInjector. Launch via sfse_loader.exe through gamescope (Wayland, 2560x1080@200Hz, HDR enabled).

System information

  • GPU: AMD Radeon RX 9060 XT (RDNA4, Navi 44, PCI 0000:0a:00.0)
  • Driver: Mesa 26.1.2-arch3.1 (RADV/ACO)
  • Wine version: Proton CachyOS native 11.0-20260602-native
  • vkd3d-proton version: 3.1.0 (build ee737e324376289)

Issue

VKD3D-proton detects Broken NULL PRT with SMEM and applies per-game Starfield workarounds (2 configs + shader quirks), but the workarounds are incomplete. During terrain/material loading, a GPUVM page fault occurs from the SQC data client (SMEM scalar load) at address 0x0000000000001000, causing a full GPU reset with VRAM loss.

dmesg sequence:

Process Main pid 16080 thread vkd3d_queue pid 16925
Ring gfx_0.0.0 reset failed
GPU reset begin!. Source: 1
MES(1) failed to respond to msg=REMOVE_QUEUE (x5)
failed to halt cp gfx
gfxhub page fault (src_id:0 ring:24 vmid:6 pasid:283)
  in page starting at address 0x0000000000001000 from client 10
GCVM_L2_PROTECTION_FAULT_STATUS:0x00601431
  Faulty UTCL2 client ID: SQC (data) (0xa)
  MORE_FAULTS: 0x1, PERMISSION_FAULTS: 0x3, RW: 0x0
MODE1 reset
VRAM is lost due to GPU reset!

Reproduction: On save load, always during terrain/material loading. Always the same SQC data client fault at 0x1000. Same fault signature reported on RX 9070 XT (https://gitlab.freedesktop.org/mesa/mesa/-/work_items/15689), (https://gitlab.freedesktop.org/mesa/mesa/-/work_items/15708) and RX 7900 XTX (https://gitlab.freedesktop.org/mesa/mesa/-/work_items/15737) GitLab for MESA

Already tested with no effect: disabled raytracing, disabled frame gen/upscaling, multiple Proton versions (Experimental, CachyOS native), LTS kernel, deleted vkd3d-proton cache, removed VKD3D_CONFIG=force_host_cached.

VKD3D-proton log (relevant lines from steam-1716740.log):

3503.886:info:vkd3d-proton:vkd3d_bindless_state_add_binding: Device supports VK_EXT_descriptor_buffer! (x3)
3503.786:info:vkd3d-proton:vkd3d_instance_apply_application_workarounds: Detected game Starfield.exe, adding 2 configs, removing 0 configs.
3503.786:info:vkd3d-proton:vkd3d_instance_apply_application_workarounds: Detected game Starfield.exe, adding shader quirks for specific shaders.

The game runs until terrain/material loading, then VKD3D-proton never receives a device_lost callback — the GPU hangs so hard the kernel resets it before the error propagates. The last game output before the hang: "TERRAIN: Loaded 72 material slots".

Log files

  • Proton log: steam-1716740.log (attached)
HHansKristian-Work maintainer 2026-06-30 github

Starfield v1.16.244, heavily modded (~200 ESM mods, 14 SFSE plugins). SFSE v0.2.21. Custom shaders via SFShaderInjector.

Maybe not use a heavily modded game so that others at least have a chance at reproducing?

Also, that Mesa link is broken.

SSelene0623 2026-07-29 github

Starfield v1.16.244, heavily modded (~200 ESM mods, 14 SFSE plugins). SFSE v0.2.21. Custom shaders via SFShaderInjector.

Maybe not use a heavily modded game so that others at least have a chance at reproducing?

Also, that Mesa link is broken.

I tried on a vanilla Starfield install and got the exact same error. Same log entry, same everything. No need to post again. I also fixed the links. The only difference from before when I had it working was I had a 16:9 1080p 60Hz SDR Samsung SyncMaster TN Panel (a dinosaur) over DVI->HDMI and recently switched to an Acer Nitro 21:9 1080P 200Hz HDR UltraWide VA Panel with FreeSync over DP. after that switch the game drops the display link and freezes, taking KWin with it.

AAttemptToCallNil 2026-08-03 github

I experience the same or a similar issue: a GPU hang on save load. (If this needs its own issue report, I'll see what I can do.) I have an RX 6700 XT. The reported values are mostly identical (only the address is different).

The issue I experience occurs on save load only and appears to depend on the area in the save. In my case, I've approximately narrowed it down to "areas where the exterior of a space station is modeled". Actually completing a mission with an affected area is possible in a continuous session as long as no saves made in affected areas are loaded.

Loading saves made in these places caused hangs:

  • On a ship next to the Nova Galactic staryard
  • In the first zone of the Nova Galactic staryard (the small area with an exterior window)
  • On the station where the Shattered Space DLC campaign starts

Loading saves made in the these places did not cause hangs:

  • In the second zone of the Nova Galactic staryard (where all the enemies are; I don't think there are any exterior windows there)
  • In orbit around Kreet after resolving the initial storyline mission there
  • In the Lodge, on planets, or in most other places

If the issue I describe is the same issue OP experiences, this information may help with reproducing it.

I do not believe it to be a save corruption issue (the Nova Galactic mission is one of the first things I did in the game), and I doubt save corruption should cause GPU hangs anyway.

I speculate the issue may be a bug or regression introduced by the Terran Armada DLC / Free Lanes update, because I don't recall this issue happening before. I admit, however, that the conditions for triggering the issue are narrow, and I can't be sure I simply failed to trigger it before the update by chance.

Rrunar-work 2026-08-03 github

It would be helpful if you could provide some save files that trigger this hang. They should be in steamapps/compatdata/1716740/pfx/drive_c/users/steamuser/My Documents/My Games/Starfield/Saves/

Launch options