Comfy-Org / Comfy-Org/ComfyUI

rocm unhandled OOM exception

Open
#11,622 3 comments 0 reactions 0 assignees View on GitHub
Potential Bug
Dominant language
Python
Stars
133k
Forks
15.7k
Avg merge
1d 10h
Merged PRs (30d)
153

Description

### Custom Node Testing

- [ ] I have tried disabling custom nodes and the issue persists (see [how to disable custom nodes](https://docs.comfy.org/troubleshooting/custom-node-issues#step-1%3A-test-with-all-custom-nodes-disabled) if you need help)

### Expected Behavior

comfyui catches the out-of-memory error and logs it in the comfyui.log file

### Actual Behavior

the error is not logged in comfyui.log. It only gets displayed in the console.
A typical error message looks like this:
:0:C:\develop\TheRock\rocm-systems\projects\clr\rocclr\device\device.cpp:379 : 110090412676 us: Memobj map does not have ptr: 0x8ae80000

### Steps to Reproduce

fresh install of ComfyUI 7.0
git clone manager
run the Z-Image Turbo (new) text-to-image workflow, using the bf16 model
run with only the default command line arguments :
.\python_embeded\python.exe -s .\ComfyUI\main.py --windows-standalone-build

I have this issue mainly when using the zit-bf16 model, and the wan22-fp8 models.
Below is the debug log for the zit-bf16 run

### Debug Logs

```powershell
## ComfyUI-Manager: installing dependencies done.
[2026-01-04 14:02:34.516] ** ComfyUI startup time: 2026-01-04 14:02:34.516
[2026-01-04 14:02:34.516] ** Platform: Windows
[2026-01-04 14:02:34.516] ** Python version: 3.12.10 (tags/v3.12.10:0cc8128, Apr 8 2025, 12:21:36) [MSC v.1943 64 bit (AMD64)]
[2026-01-04 14:02:34.516] ** Python executable: C:\Comfy\python_embeded\python.exe
[2026-01-04 14:02:34.516] ** ComfyUI Path: C:\Comfy\ComfyUI
[2026-01-04 14:02:34.516] ** ComfyUI Base Folder Path: C:\Comfy\ComfyUI
[2026-01-04 14:02:34.516] ** User directory: C:\Comfy\ComfyUI\user
[2026-01-04 14:02:34.516] ** ComfyUI-Manager config path: C:\Comfy\ComfyUI\user\__manager\config.ini
[2026-01-04 14:02:34.522] ** Log path: C:\Comfy\ComfyUI\user\comfyui.log

Prestartup times for custom nodes:
[2026-01-04 14:02:35.108] 1.6 seconds: C:\Comfy\ComfyUI\custom_nodes\comfyui-manager
[2026-01-04 14:02:35.108]
[2026-01-04 14:02:36.175] Checkpoint files will always be loaded safely.
[2026-01-04 14:02:36.419] Total VRAM 24560 MB, total RAM 64669 MB
[2026-01-04 14:02:36.419] pytorch version: 2.9.0+rocmsdk20251116
[2026-01-04 14:02:36.419] Set: torch.backends.cudnn.enabled = False for better AMD performance.
[2026-01-04 14:02:36.419] AMD arch: gfx1100
[2026-01-04 14:02:36.419] ROCm version: (7, 1)
[2026-01-04 14:02:36.419] Set vram state to: NORMAL_VRAM
[2026-01-04 14:02:36.419] Device: cuda:0 AMD Radeon RX 7900 XTX : native
[2026-01-04 14:02:36.428] Using async weight offloading with 2 streams
[2026-01-04 14:02:36.428] Enabled pinned memory 29101.0
[2026-01-04 14:02:37.140] Using sub quadratic optimization for attention, if you have memory or speed issues try using: --use-split-cross-attention
[2026-01-04 14:02:38.455] Python version: 3.12.10 (tags/v3.12.10:0cc8128, Apr 8 2025, 12:21:36) [MSC v.1943 64 bit (AMD64)]
[2026-01-04 14:02:38.455] ComfyUI version: 0.7.0
[2026-01-04 14:02:38.492] ComfyUI frontend version: 1.35.9
[2026-01-04 14:02:38.492] [Prompt Server] web root: C:\Comfy\python_embeded\Lib\site-packages\comfyui_frontend_package\static
[2026-01-04 14:02:38.896] Total VRAM 24560 MB, total RAM 64669 MB
[2026-01-04 14:02:38.896] pytorch version: 2.9.0+rocmsdk20251116
[2026-01-04 14:02:38.896] Set: torch.backends.cudnn.enabled = False for better AMD performance.
[2026-01-04 14:02:38.896] AMD arch: gfx1100
[2026-01-04 14:02:38.896] ROCm version: (7, 1)
[2026-01-04 14:02:38.896] Set vram state to: NORMAL_VRAM
[2026-01-04 14:02:38.896] Device: cuda:0 AMD Radeon RX 7900 XTX : native
[2026-01-04 14:02:38.912] Using async weight offloading with 2 streams
[2026-01-04 14:02:38.912] Enabled pinned memory 29101.0
[2026-01-04 14:02:39.211] ComfyUI-GGUF: Allowing full torch compile
[2026-01-04 14:02:39.227] ### Loading: ComfyUI-Manager (V3.39)
[2026-01-04 14:02:39.227] [ComfyUI-Manager] network_mode: offline
[2026-01-04 14:02:39.227] [ComfyUI-Manager] ComfyUI per-queue preview override detected (PR #11261). Manager's preview method feature is disabled. Use ComfyUI's --preview-method CLI option or 'Settings > Execution > Live preview method'.
[2026-01-04 14:02:39.295] ### ComfyUI Revision: 150 [f59f71cf] *DETACHED | Released on '2025-12-30'
[2026-01-04 14:02:39.306] [ComfyUI-Manager] All startup tasks have been completed.
[2026-01-04 14:02:39.306]
Import times for custom nodes:
[2026-01-04 14:02:39.306] 0.0 seconds: C:\Comfy\ComfyUI\custom_nodes\websocket_image_save.py
[2026-01-04 14:02:39.306] 0.0 seconds: C:\Comfy\ComfyUI\custom_nodes\ComfyUI-GGUF
[2026-01-04 14:02:39.306] 0.1 seconds: C:\Comfy\ComfyUI\custom_nodes\comfyui-manager
[2026-01-04 14:02:39.306]
[2026-01-04 14:02:39.622] Context impl SQLiteImpl.
[2026-01-04 14:02:39.622] Will assume non-transactional DDL.
[2026-01-04 14:02:39.624] No target revision found.
[2026-01-04 14:02:39.638] Starting server

[2026-01-04 14:02:39.638] To see the GUI go to: http://127.0.0.1:8188
[2026-01-04 14:02:41.909] [DEPRECATION WARNING] Detected import of deprecated legacy API: /scripts/ui.js. This is likely caused by a custom node extension using outdated APIs. Please update your extensions or contact the extension author for an updated version.
[2026-01-04 14:02:41.913] [DEPRECATION WARNING] Detected import of deprecated legacy API: /extensions/core/groupNode.js. This is likely caused by a custom node extension using outdated APIs. Please update your extensions or contact the extension author for an updated version.
[2026-01-04 14:02:42.147] [DEPRECATION WARNING] Detected import of deprecated legacy API: /scripts/ui/components/buttonGroup.js. This is likely caused by a custom node extension using outdated APIs. Please update your extensions or contact the extension author for an updated version.
[2026-01-04 14:02:42.150] [DEPRECATION WARNING] Detected import of deprecated legacy API: /scripts/ui/components/button.js. This is likely caused by a custom node extension using outdated APIs. Please update your extensions or contact the extension author for an updated version.
[2026-01-04 14:02:45.267] got prompt
[2026-01-04 14:02:45.289] Using split attention in VAE
[2026-01-04 14:02:45.289] Using split attention in VAE
[2026-01-04 14:02:45.460] VAE load device: cuda:0, offload device: cpu, dtype: torch.bfloat16
[2026-01-04 14:02:47.833] CLIP/text encoder model load device: cuda:0, offload device: cpu, current: cpu, dtype: torch.float16
[2026-01-04 14:02:48.405] Requested to load ZImageTEModel_
[2026-01-04 14:02:49.654] loaded completely; 22891.42 MB usable, 7672.25 MB loaded, full load: True
[2026-01-04 14:02:51.396] model weight dtype torch.bfloat16, manual cast: None
[2026-01-04 14:02:51.397] model_type FLOW
[2026-01-04 14:02:56.501] Requested to load Lumina2
[2026-01-04 14:02:59.661] Unloaded partially: 7672.25 MB freed, 0.00 MB remains loaded, 2320.62 MB buffer reserved, lowvram patches: 0
[2026-01-04 14:03:02.659] loaded completely; 18675.99 MB usable, 11739.54 MB loaded, full load: True
[2026-01-04 14:03:07.389]
100%|████████████████████████████████████████████████████████████████████████████████████| 4/4 [00:04<00:00, 1.07s/it]
100%|████████████████████████████████████████████████████████████████████████████████████| 4/4 [00:04<00:00, 1.14s/it]
[2026-01-04 14:03:07.391] Requested to load AutoencodingEngine

:0:C:\develop\TheRock\rocm-systems\projects\clr\rocclr\device\device.cpp:379 : 110090412676 us: Memobj map does not have ptr: 0x8ae80000
```

### Other

_No response_

Contributor guide

Open the contributing guide

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.