[GH-ISSUE #26] How to disable CUDA and to run on PyTorch? #9

New issue

Open

opened 2026-10-26 22:23:28 +00:00 by bsakaguchi · 2 comments

bsakaguchi commented

2026-10-26 22:23:28 +00:00

Owner

Originally created by @yurivict on GitHub (Mar 19, 2026).
Original GitHub issue: https://github.com/Comfy-Org/comfy-aimdo/issues/26

comfy-kitchen allows to disable CUDA and it passes all tests.

How to disable CUDA in comfy-aimdo ?

We would like to run ComfyUI on PyTorch w/out CUDA, but comfy-aimdo fails:

comfy-aimdo failed to load: Cannot open "/usr/local/lib/python3.11/site-packages/comfy_aimdo/aimdo.so"
NOTE: comfy-aimdo is currently only support for Nvidia GPUs

We are on FreeBSD where CUDA is not currently available.
I am trying to create FreeBSD ports for all ConfyUI components, and I've already did this, but there's the above failure from comfy-aimdo that I am trying to fix.

Originally created by @yurivict on GitHub (Mar 19, 2026). Original GitHub issue: https://github.com/Comfy-Org/comfy-aimdo/issues/26 comfy-kitchen allows to disable CUDA and it passes all tests. How to disable CUDA in comfy-aimdo ? We would like to run ComfyUI on PyTorch w/out CUDA, but comfy-aimdo fails: ``` comfy-aimdo failed to load: Cannot open "/usr/local/lib/python3.11/site-packages/comfy_aimdo/aimdo.so" NOTE: comfy-aimdo is currently only support for Nvidia GPUs ``` We are on FreeBSD where CUDA is not currently available. I am trying to create FreeBSD ports for all ConfyUI components, and I've already did this, but there's the above failure from comfy-aimdo that I am trying to fix.

bsakaguchi commented

2026-10-26 22:23:29 +00:00

Author

Owner

@rattus128 commented on GitHub (Mar 20, 2026):

comfy-kitchen allows to disable CUDA and it passes all tests.

How to disable CUDA in comfy-aimdo ?

We would like to run ComfyUI on PyTorch w/out CUDA, but comfy-aimdo fails:
comfy-aimdo failed to load: Cannot open "/usr/local/lib/python3.11/site-packages/comfy_aimdo/aimdo.so"
NOTE: comfy-aimdo is currently only support for Nvidia GPUs
We are on FreeBSD where CUDA is not currently available. I am trying to create FreeBSD ports for all ConfyUI components, and I've already did this, but there's the above failure from comfy-aimdo that I am trying to fix.

Does this cause you problems in comfyUI after it fails?

The expectation is ComfyUI falls back to non-aimdo execution in full for unsupported platforms and this message is expected.

@rattus128 commented on GitHub (Mar 20, 2026): > comfy-kitchen allows to disable CUDA and it passes all tests. > > How to disable CUDA in comfy-aimdo ? > > We would like to run ComfyUI on PyTorch w/out CUDA, but comfy-aimdo fails: > > ``` > comfy-aimdo failed to load: Cannot open "/usr/local/lib/python3.11/site-packages/comfy_aimdo/aimdo.so" > NOTE: comfy-aimdo is currently only support for Nvidia GPUs > ``` > > We are on FreeBSD where CUDA is not currently available. I am trying to create FreeBSD ports for all ConfyUI components, and I've already did this, but there's the above failure from comfy-aimdo that I am trying to fix. Does this cause you problems in comfyUI after it fails? The expectation is ComfyUI falls back to non-aimdo execution in full for unsupported platforms and this message is expected.

bsakaguchi commented

2026-10-26 22:23:29 +00:00

Author

Owner

@yurivict commented on GitHub (Mar 21, 2026):

comfyui fails w/out the arguments --disable-dynamic-vram --cpu.
The errors begin with aimdo, but maybe aimdo isn't a core reason.

Errors:

$ comfyui 
comfy-aimdo failed to load: Cannot open "/usr/local/lib/python3.11/site-packages/comfy_aimdo/aimdo.so"
NOTE: comfy-aimdo is currently only support for Nvidia GPUs
Found comfy_kitchen backend triton: {'available': False, 'disabled': True, 'unavailable_reason': "ImportError: No module named 'triton'", 'capabilities': []}
Found comfy_kitchen backend cuda: {'available': False, 'disabled': True, 'unavailable_reason': 'Extension file not found: /usr/local/lib/python3.11/site-packages/comfy_kitchen/backends/cuda/_C.abi3.so', 'capabilities': []}
Found comfy_kitchen backend eager: {'available': True, 'disabled': False, 'unavailable_reason': None, 'capabilities': ['apply_rope', 'apply_rope1', 'dequantize_mxfp8', 'dequantize_nvfp4', 'dequantize_per_tensor_fp8', 'quantize_mxfp8', 'quantize_nvfp4', 'quantize_per_tensor_fp8', 'scaled_mm_mxfp8', 'scaled_mm_nvfp4']}
Checkpoint files will always be loaded safely.
Traceback (most recent call last):
  File "/usr/local/share/comfyui/main.py", line 197, in <module>
    import execution
  File "/usr/local/share/comfyui/execution.py", line 17, in <module>
    import comfy.model_management
  File "/usr/local/share/comfyui/comfy/model_management.py", line 256, in <module>
    total_vram = get_total_memory(get_torch_device()) / (1024 * 1024)
                                  ^^^^^^^^^^^^^^^^^^
  File "/usr/local/share/comfyui/comfy/model_management.py", line 206, in get_torch_device
    return torch.device(torch.cuda.current_device())
                        ^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.11/site-packages/torch/cuda/__init__.py", line 1094, in current_device
    _lazy_init()
  File "/usr/local/lib/python3.11/site-packages/torch/cuda/__init__.py", line 417, in _lazy_init
    raise AssertionError("Torch not compiled with CUDA enabled")
AssertionError: Torch not compiled with CUDA enabled

comfyui works fine with --disable-dynamic-vram --cpu.

@yurivict commented on GitHub (Mar 21, 2026): comfyui fails w/out the arguments ```--disable-dynamic-vram --cpu```. The errors begin with aimdo, but maybe aimdo isn't a core reason. Errors: ``` $ comfyui comfy-aimdo failed to load: Cannot open "/usr/local/lib/python3.11/site-packages/comfy_aimdo/aimdo.so" NOTE: comfy-aimdo is currently only support for Nvidia GPUs Found comfy_kitchen backend triton: {'available': False, 'disabled': True, 'unavailable_reason': "ImportError: No module named 'triton'", 'capabilities': []} Found comfy_kitchen backend cuda: {'available': False, 'disabled': True, 'unavailable_reason': 'Extension file not found: /usr/local/lib/python3.11/site-packages/comfy_kitchen/backends/cuda/_C.abi3.so', 'capabilities': []} Found comfy_kitchen backend eager: {'available': True, 'disabled': False, 'unavailable_reason': None, 'capabilities': ['apply_rope', 'apply_rope1', 'dequantize_mxfp8', 'dequantize_nvfp4', 'dequantize_per_tensor_fp8', 'quantize_mxfp8', 'quantize_nvfp4', 'quantize_per_tensor_fp8', 'scaled_mm_mxfp8', 'scaled_mm_nvfp4']} Checkpoint files will always be loaded safely. Traceback (most recent call last): File "/usr/local/share/comfyui/main.py", line 197, in <module> import execution File "/usr/local/share/comfyui/execution.py", line 17, in <module> import comfy.model_management File "/usr/local/share/comfyui/comfy/model_management.py", line 256, in <module> total_vram = get_total_memory(get_torch_device()) / (1024 * 1024) ^^^^^^^^^^^^^^^^^^ File "/usr/local/share/comfyui/comfy/model_management.py", line 206, in get_torch_device return torch.device(torch.cuda.current_device()) ^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/usr/local/lib/python3.11/site-packages/torch/cuda/__init__.py", line 1094, in current_device _lazy_init() File "/usr/local/lib/python3.11/site-packages/torch/cuda/__init__.py", line 417, in _lazy_init raise AssertionError("Torch not compiled with CUDA enabled") AssertionError: Torch not compiled with CUDA enabled ``` comfyui works fine with ```--disable-dynamic-vram --cpu```.

bsakaguchi referenced this issue

2026-10-26 22:23:57 +00:00

[GH-ISSUE #100] Blackwell (RTX 5060 Ti): intermittent Fault failed: 2 (cuMemSetAccess: device not ready) with DynamicVRAM/VBAR #36