mirror of
https://github.com/Comfy-Org/ComfyUI-Manager.git
synced 2026-11-04 23:37:31 +00:00
[GH-ISSUE #2724] RX 6800 / Windows 11 / Desktop 0.8.24 crashes on startup with exit code 3221225477 during torch.cuda.is_available() #1023
Labels
No milestone
No project
No assignees
1 participant
Notifications
Due date
No due date set.
Dependencies
No dependencies set.
Reference
Comfy-Org/ComfyUI-Manager#1023
Loading…
Add table
Add a link
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?
Originally created by @JrouleEnGolf5 on GitHub (Mar 21, 2026).
Original GitHub issue: https://github.com/Comfy-Org/ComfyUI-Manager/issues/2724
Expected Behavior
ComfyUI Desktop should start normally and open the UI.
Actual Behavior
ComfyUI Desktop crashes at startup with:
Python process exited with code 3221225477 and signal null
This appears to happen during GPU probing. In my logs, the crash occurs when Python reaches torch.cuda.is_available().
Environment
GPU: AMD Radeon RX 6800 (16 GB VRAM)
Architecture: RDNA2 / gfx1030
OS: Windows 11
ComfyUI Desktop: v0.8.24
PyTorch: 2.9.1+rocmsdk20260116
HIP: 7.2.26024-f6f897bd3d
AMD Adrenalin driver: 26.1.1 (and 26.3.1)
Reproduction
Fresh install of ComfyUI Desktop on Windows.
On startup, Desktop crashes immediately and shows:
Python process exited with code 3221225477 and signal null
I also reproduced the failure directly outside ComfyUI with the bundled Python environment:
..venv\Scripts\python.exe -c "import torch; print(torch.version); print(torch.version.hip); print(torch.cuda.is_available())"
Output stops after:
2.9.1+rocmsdk20260116
7.2.26024-f6f897bd3d
The line for torch.cuda.is_available() is never printed because Python crashes at that point.
Additional Notes
This is a fresh install. ComfyUI has never successfully started on this machine.
Disabling extensions/plugins did not resolve the issue.
Logs suggest the crash is triggered during GPU detection, not during workflow execution.
The Windows exit code 3221225477 suggests a native access violation / hard crash rather than a Python exception.
What I already tested
Fresh install
Updated GPU driver
Disabling optional components/plugins
Direct PyTorch test outside the UI
Logs / Evidence
Relevant behavior:
Desktop crashes immediately at startup
The same crash happens when calling torch.cuda.is_available() manually
Exit code is consistently 3221225477
I can provide the full logs if needed.
Question
Is this a known issue with:
AMD RX 6800 / RDNA2 on Windows Desktop
the current ROCm/PyTorch bundle
or driver compatibility with ComfyUI Desktop 0.8.24?
Any recommended workaround would be appreciated.