GPU hang → device lost (VK_ERROR_DEVICE_LOST, Windows TDR) at Zora's River on AMD RX 6800 XT (v2.0.11) #1465
Labels
No labels
accessibility
autorelease: pending
autorelease: tagged
bug
dependencies
documentation
duplicate
enhancement
github_actions
good first issue
help wanted
invalid
question
testing wanted
wontfix
No milestone
No assignees
1 participant
Due date
No due date set.
Dependencies
No dependencies set
Reference
MagicBeanDev/Hookshot#1465
Loading…
Reference in a new issue
No description provided.
Delete branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Hookshot version
2.0.11
What kind of problem
Crash, freeze or black screen
What happened
Summary
At Zora's River the game freezes for about 4–5 seconds, then the GPU is lost (
submit … failed (-4),device lost). Windows resets the display driver (LiveKernelEvent 141), and the game has to be closed. It keeps happening there on v2.0.11. Frame generation and the shader caches don't seem to be the cause: it still crashes with frame generation off and a freshly rebuilt pipeline cache.Environment
d6369f1e), built with HookshotBuilder v2.0.110x800161immediateOptions that may matter:
quality_preset=0,frame_generation=0(crash also happens with2),frame_pacing=1,upscaling=1,resolution=1,output_res=1,robust_buffers=0.Steps to reproduce
What the log shows
Everything is normal (about 60 fps, ~12 ms GPU time a frame, ~2 GB of 16 GB VRAM in use). Then a write-back stops completing, the submit thread stalls on a slot for ~3.7 s, and the device is lost:
In an earlier run it looked the same, with
slow submit: 3681.1 ms … slot wait 3670.8and thensubmit 29072 failed (-13)andgraphics pipeline … not made (-4).The completed counter stops 4 submits behind and never moves again. The GPU isn't simply overloaded (frames were light just before); it looks like a genuine hang in one of those submissions.
Frequency
Windows logged LiveKernelEvent 141 (display driver timeout/reset) 16 times in about 15 hours on 2026-11-09/10. The ones I checked line up with Hookshot crash logs, and each of those shows the same pattern (
wait for write-back … timed out→-4).What I tried
frame_generation2 → 0 (log confirms interpolation is off)mkw_vk_pipelines.bin,hookshot_pipelines.bin,hookshot_pipeline_states.binand let them rebuild (took ~30 min)graphics pipeline … not made (-13)errors at startup went away, so those came from a stale cache and aren't the cause.robust_buffers=1robust_buffers=0the log printsvk: robust buffer access on; with=1that line disappears. Is that inverted, or expected?Other things in the log that might be related
view of texture 0x1133ccb000 (vk 139) as vk 95 … layers 0..15 not made. If I'm reading the VkFormat numbers right, that's a BC4 texture viewed as R16G16B16A16_UINT (same 8-byte block size).0 programs (10 failed)shows up shortly after boot on every run.subgroups … pinned to 32 on stages 0x20,task and mesh programs built without the driver's optimizer (AMD; RX_VK_MESH_OPT=1: with it),matrix multiplies in software.Where in the game
Zora's River
How often
Every time
Graphics card and driver version
RX 6800 XT
CPU, RAM and operating system
Ryzen 5 5600X, 32GB, Windows 11
Changes from the defaults
No response
Other software hooked into the game
No, nothing hooked in
hookshot_log.txt
hookshot_log.txt
hookshot_log_checkpoints.txt
Screenshots or video
No response
Before you send it
Thanks for the report. This is the same problem as #996 (AMD freezes and driver resets (device lost)), so it's tracked there from now on: please add any new logs or details to #996. Closing as a duplicate.