[AMD RX 6750 XT] High VRAM usage (hits 10.4/10.4 GB): freeze in Hyrule Field, slow Castle Town load, owl model missing (v2.0.14) #2058
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#2058
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?
GPU: AMD Radeon RX 6750 XT (12 GB), driver 32.0.21045.16010, Windows 11
CPU/RAM: Ryzen 5 3600, 16 GB
Hookshot: v2.0.14, 1920x1080 output, FSR 3.1 (frame generation 2x)
The main problem is VRAM usage, which keeps climbing until it hits the limit and then the game stalls or freezes.
What happens
VK_ERROR_DEVICE_LOSTand no checkpoint dump; the process stays alive but stops presenting frames.hookshot_log_vram_freeze.zip
What the log shows right before the freeze (
hookshot_log.txt)VRAM: 10693 MB used of 10693 MB allowed; texture budget dropped from 6144 to 4077 MB195 new ASTC textures left for a later framestatus: 600 frames, the slowest 2008.0 ms; VRAM 10.4 of 10.4 GB296.9 draws left out (not ready), 841 jobs queued, replay submit ~54 msthe slowest 28434.2 ms, VRAM 9.9 of 10.4 GB) that it recovered fromCould VRAM usage be reduced (texture streaming / budget, freeing textures between areas)? Possibly related: #1950, #1857.
Happy to test builds and send more logs on this GPU. Log attached.
I checked the attached log. It shows the 2.008 s frame at the 10.4 GB limit and the 841 queued translation jobs, then resumes presenting. Near the end it reports 105,120 presents and a normal window close. That makes the attachment useful for the long-stall case. A log ending during a run that never recovered would help isolate the hard freeze.
Can you reproduce either case with the current release candidate, using the same GPU, output and frame generation settings? If you do, please attach that log and note the last area or transition on screen.
Tested v2.1.0 with RX_VK_COMPACT=1 (same GPU, 1080p output, cap 90).
VRAM: fixed. Compaction gave 2.8 GB back to the driver and VRAM stayed at 9.2–9.8 of 10.8 GB, never at the limit.
Still left:
New log attached.
hookshot_log_amd_fps.zip
Thanks for the report. Reports of the same problem are now collected in #993, which lists this one with your details and links back here, so one fix closes them together and the open list stays readable. Follow #993 for the fix, and add new details there (or here). If this turns out to be a separate problem it will be reopened.