Confyuai AMD GPU crash - AMD Radeon RX 6650 XT: failed to run amdgpu-arch binary not found.

[Delta-Dedalus]

Expected behavior:
generate any content.

Actual behavior:
Confyuai crashes.

How they responded:
Install the current AMD experimental portable ComfyUi build, run the bat, A [WARNING] failed to run amdgpu-arch: binary not found—error can be observed in the CMD console. Try to run a prompt, check console, model hangs on loading and freezes the PC.

We are talking about generations of images or modifications thereof. Even when using only the models provided, the installation does not work.

I enclose the CMD register.

Nuovo documento di testo (2).txt

[Delta-Dedalus]

For those who, for security reasons, do not want to download the text file, I am pasting the registry below:

XComfyUI_windows_portable.python_embededpython.exe -s ComfyUImain.py --windows-standalone-build
[START] Security scan
[DONE] Security scan

ComfyUI-Manager installing dependencies done.

ComfyUI startup time 2025-12-26 223715.026
Platform Windows
Python version 3.12.10 (tagsv3.12.100cc8128, Apr 8 2025, 122136) [MSC v.1943 64 bit (AMD64)]
Python executable XComfyUI_windows_portablepython_embededpython.exe
ComfyUI Path XComfyUI_windows_portableComfyUI
ComfyUI Base Folder Path XComfyUI_windows_portableComfyUI
User directory XComfyUI_windows_portableComfyUIuser
ComfyUI-Manager config path XComfyUI_windows_portableComfyUIuser__managerconfig.ini
Log path XComfyUI_windows_portableComfyUIusercomfyui.log

Prestartup times for custom nodes
2.9 seconds XComfyUI_windows_portableComfyUIcustom_nodesComfyUI-Manager

[WARNING] failed to run amdgpu-arch binary not found.
Checkpoint files will always be loaded safely.
Total VRAM 8176 MB, total RAM 16220 MB
pytorch version 2.9.0+rocmsdk20251116
Set torch.backends.cudnn.enabled = False for better AMD performance.
AMD arch gfx1032
ROCm version (7, 1)
Set vram state to NORMAL_VRAM
Device cuda0 AMD Radeon RX 6650 XT native
Enabled pinned memory 7299.0
Using sub quadratic optimization for attention, if you have memory or speed issues try using --use-split-cross-attention
Python version 3.12.10 (tagsv3.12.100cc8128, Apr 8 2025, 122136) [MSC v.1943 64 bit (AMD64)]
ComfyUI version 0.6.0
ComfyUI frontend version 1.34.9
[Prompt Server] web root XComfyUI_windows_portablepython_embededLibsite-packagescomfyui_frontend_packagestatic
Total VRAM 8176 MB, total RAM 16220 MB
pytorch version 2.9.0+rocmsdk20251116
Set torch.backends.cudnn.enabled = False for better AMD performance.
AMD arch gfx1032
ROCm version (7, 1)
Set vram state to NORMAL_VRAM
Device cuda0 AMD Radeon RX 6650 XT native
Enabled pinned memory 7299.0
ComfyUI-GGUF Allowing full torch compile

Loading ComfyUI-Manager (V3.39)

[ComfyUI-Manager] network_mode public
[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'.

ComfyUI Revision 284 [e4c61d7] DETACHED Released on '2025-12-23'

Import times for custom nodes
0.0 seconds XComfyUI_windows_portableComfyUIcustom_nodeswebsocket_image_save.py
0.1 seconds XComfyUI_windows_portableComfyUIcustom_nodesComfyUI-GGUF
0.3 seconds XComfyUI_windows_portableComfyUIcustom_nodesComfyUI-Manager

[ComfyUI-Manager] default cache updated httpsraw.githubusercontent.comltdrdataComfyUI-Managermainmodel-list.json[ComfyUI-Manager] default cache updated httpsraw.githubusercontent.comltdrdataComfyUI-Managermainalter-list.json[ComfyUI-Manager] default cache updated httpsraw.githubusercontent.comltdrdataComfyUI-Managermaingithub-stats.json
[ComfyUI-Manager] default cache updated httpsraw.githubusercontent.comltdrdataComfyUI-Managermainextension-node-map.json
[ComfyUI-Manager] default cache updated httpsraw.githubusercontent.comltdrdataComfyUI-Managermaincustom-node-list.json
Context impl SQLiteImpl.
Will assume non-transactional DDL.
No target revision found.
Starting server

To see the GUI go to http127.0.0.18188
[DEPRECATION WARNING] Detected import of deprecated legacy API scriptsui.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.
[DEPRECATION WARNING] Detected import of deprecated legacy API extensionscoregroupNode.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.
[DEPRECATION WARNING] Detected import of deprecated legacy API scriptsuicomponentsbuttonGroup.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.
[DEPRECATION WARNING] Detected import of deprecated legacy API scriptsuicomponentsbutton.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.
FETCH ComfyRegistry Data 5115
FETCH ComfyRegistry Data 10115
FETCH ComfyRegistry Data 15115
FETCH ComfyRegistry Data 20115
FETCH ComfyRegistry Data 25115
FETCH ComfyRegistry Data 30115
FETCH ComfyRegistry Data 35115
FETCH ComfyRegistry Data 40115
FETCH ComfyRegistry Data 45115
FETCH ComfyRegistry Data 50115
FETCH ComfyRegistry Data 55115
FETCH ComfyRegistry Data 60115
FETCH ComfyRegistry Data 65115
FETCH ComfyRegistry Data 70115
FETCH ComfyRegistry Data 75115
FETCH ComfyRegistry Data 80115
FETCH ComfyRegistry Data 85115
FETCH ComfyRegistry Data 90115
FETCH ComfyRegistry Data 95115
FETCH ComfyRegistry Data 100115
FETCH ComfyRegistry Data 105115
FETCH ComfyRegistry Data 110115
FETCH ComfyRegistry Data 115115
FETCH ComfyRegistry Data [DONE]
[ComfyUI-Manager] default cache updated httpsapi.comfy.orgnodes
FETCH DATA from httpsraw.githubusercontent.comltdrdataComfyUI-Managermaincustom-node-list.json [DONE]
[ComfyUI-Manager] All startup tasks have been completed.
got prompt
Using split attention in VAE
Using split attention in VAE
VAE load device cuda0, offload device cpu, dtype torch.bfloat16
Requested to load AutoencoderKL
0 models unloaded.
loaded completely; 3214.29 MB usable, 160.31 MB loaded, full load True

XComfyUI_windows_portablepause
Premere un tasto per continuare . . .

[lcepi]

У меня также, rdna2 не поддерживается я так понимаю

[hamzeht8]

https://github.com/guinmoon/rocm7_builds
i use this one
my gpu is rx6600xt
its works fine

[hamzeht8]

And ignore the 'arch binary not found' error because this is for Linux, not Windows