Torch not compiled with CUDA enabled

[foysalremon]

I think it's because you don't add support for other gpu like intel arc which doesn't have cuda. If so, requesting to add support

Error Details

• Node Type: CatVTONWrapper
• Exception Type: AssertionError
• Exception Message: Torch not compiled with CUDA enabled

Stack Trace

  File "G:\ComfyUI\execution.py", line 323, in execute
    output_data, output_ui, has_subgraph = get_output_data(obj, input_data_all, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb)

  File "G:\ComfyUI\execution.py", line 198, in get_output_data
    return_values = _map_node_over_list(obj, input_data_all, obj.FUNCTION, allow_interrupt=True, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb)

  File "G:\ComfyUI\execution.py", line 169, in _map_node_over_list
    process_inputs(input_dict, i)

  File "G:\ComfyUI\execution.py", line 158, in process_inputs
    results.append(getattr(obj, func)(**inputs))

  File "G:\ComfyUI\custom_nodes\ComfyUI_CatVTON_Wrapper\py\cat_vton.py", line 42, in catvton
    pipeline = CatVTONPipeline(

  File "G:\ComfyUI\custom_nodes\ComfyUI_CatVTON_Wrapper\py\catvton\pipeline.py", line 42, in __init__
    self.vae = AutoencoderKL.from_pretrained(os.path.join(folder_paths.models_dir, "CatVTON", "sd-vae-ft-mse")).to(device, dtype=weight_dtype)

  File "G:\ComfyUI\comfyui_env\lib\site-packages\torch\nn\modules\module.py", line 1160, in to
    return self._apply(convert)

  File "G:\ComfyUI\comfyui_env\lib\site-packages\torch\nn\modules\module.py", line 810, in _apply
    module._apply(fn)

  File "G:\ComfyUI\comfyui_env\lib\site-packages\torch\nn\modules\module.py", line 810, in _apply
    module._apply(fn)

  File "G:\ComfyUI\comfyui_env\lib\site-packages\torch\nn\modules\module.py", line 833, in _apply
    param_applied = fn(param)

  File "G:\ComfyUI\comfyui_env\lib\site-packages\torch\nn\modules\module.py", line 1158, in convert
    return t.to(device, dtype if t.is_floating_point() or t.is_complex() else None, non_blocking)

  File "G:\ComfyUI\comfyui_env\lib\site-packages\torch\cuda\__init__.py", line 289, in _lazy_init
    raise AssertionError("Torch not compiled with CUDA enabled")

System Information

• ComfyUI Version: v0.2.2-58-g38c6908
• Arguments: main.py --bf16-unet
• OS: nt
• Python Version: 3.10.11 (tags/v3.10.11:7d4cc5a, Apr 5 2023, 00:38:17) [MSC v.1929 64 bit (AMD64)]
• Embedded Python: false
• PyTorch Version: 2.1.0a0+cxx11.abi

Devices

• Name: xpu
  • Type: xpu
  • VRAM Total: 8319483904
  • VRAM Free: 4993683456
  • Torch VRAM Total: 3487563776
  • Torch VRAM Free: 161763328

Logs

2024-09-22 16:26:39,791 - root - INFO - Total VRAM 7934 MB, total RAM 32556 MB
2024-09-22 16:26:39,791 - root - INFO - pytorch version: 2.1.0a0+cxx11.abi
2024-09-22 16:26:39,792 - root - INFO - Set vram state to: NORMAL_VRAM
2024-09-22 16:26:39,792 - root - INFO - Device: xpu
2024-09-22 16:26:39,809 - root - INFO - Using pytorch cross attention
2024-09-22 16:26:40,395 - root - INFO - [Prompt Server] web root: G:\ComfyUI\web
2024-09-22 16:26:43,101 - root - INFO - 
Import times for custom nodes:
2024-09-22 16:26:43,101 - root - INFO -    0.0 seconds: G:\ComfyUI\custom_nodes\websocket_image_save.py
2024-09-22 16:26:43,101 - root - INFO -    0.0 seconds: G:\ComfyUI\custom_nodes\ComfyUI_pose_inter
2024-09-22 16:26:43,101 - root - INFO -    0.0 seconds: G:\ComfyUI\custom_nodes\cg-use-everywhere
2024-09-22 16:26:43,101 - root - INFO -    0.0 seconds: G:\ComfyUI\custom_nodes\ComfyUI-Custom-Scripts
2024-09-22 16:26:43,101 - root - INFO -    0.0 seconds: G:\ComfyUI\custom_nodes\ComfyUI_CatVTON_Wrapper
2024-09-22 16:26:43,101 - root - INFO -    0.1 seconds: G:\ComfyUI\custom_nodes\comfyui_controlnet_aux
2024-09-22 16:26:43,101 - root - INFO -    0.2 seconds: G:\ComfyUI\custom_nodes\ComfyUI_Comfyroll_CustomNodes
2024-09-22 16:26:43,101 - root - INFO -    0.2 seconds: G:\ComfyUI\custom_nodes\ComfyUI-Manager
2024-09-22 16:26:43,101 - root - INFO -    0.3 seconds: G:\ComfyUI\custom_nodes\ComfyUI-Dwpose-Tensorrt
2024-09-22 16:26:43,102 - root - INFO -    0.4 seconds: G:\ComfyUI\custom_nodes\ComfyUI-Easy-Use
2024-09-22 16:26:43,102 - root - INFO -    0.8 seconds: G:\ComfyUI\custom_nodes\ComfyUI_LayerStyle
2024-09-22 16:26:43,102 - root - INFO - 
2024-09-22 16:26:43,109 - root - INFO - Starting server

2024-09-22 16:26:43,109 - root - INFO - To see the GUI go to: http://127.0.0.1:8188
2024-09-22 16:26:46,973 - root - INFO - got prompt
2024-09-22 16:26:47,032 - comfyui_segment_anything - WARNING - using extra model: G:\ComfyUI\models\sams\sam_vit_h_4b8939.pth
2024-09-22 16:27:02,235 - root - ERROR - !!! Exception during processing !!! Torch not compiled with CUDA enabled
2024-09-22 16:27:02,236 - root - ERROR - Traceback (most recent call last):
  File "G:\ComfyUI\execution.py", line 323, in execute
    output_data, output_ui, has_subgraph = get_output_data(obj, input_data_all, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb)
  File "G:\ComfyUI\execution.py", line 198, in get_output_data
    return_values = _map_node_over_list(obj, input_data_all, obj.FUNCTION, allow_interrupt=True, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb)
  File "G:\ComfyUI\execution.py", line 169, in _map_node_over_list
    process_inputs(input_dict, i)
  File "G:\ComfyUI\execution.py", line 158, in process_inputs
    results.append(getattr(obj, func)(**inputs))
  File "G:\ComfyUI\custom_nodes\ComfyUI_CatVTON_Wrapper\py\cat_vton.py", line 42, in catvton
    pipeline = CatVTONPipeline(
  File "G:\ComfyUI\custom_nodes\ComfyUI_CatVTON_Wrapper\py\catvton\pipeline.py", line 42, in __init__
    self.vae = AutoencoderKL.from_pretrained(os.path.join(folder_paths.models_dir, "CatVTON", "sd-vae-ft-mse")).to(device, dtype=weight_dtype)
  File "G:\ComfyUI\comfyui_env\lib\site-packages\torch\nn\modules\module.py", line 1160, in to
    return self._apply(convert)
  File "G:\ComfyUI\comfyui_env\lib\site-packages\torch\nn\modules\module.py", line 810, in _apply
    module._apply(fn)
  File "G:\ComfyUI\comfyui_env\lib\site-packages\torch\nn\modules\module.py", line 810, in _apply
    module._apply(fn)
  File "G:\ComfyUI\comfyui_env\lib\site-packages\torch\nn\modules\module.py", line 833, in _apply
    param_applied = fn(param)
  File "G:\ComfyUI\comfyui_env\lib\site-packages\torch\nn\modules\module.py", line 1158, in convert
    return t.to(device, dtype if t.is_floating_point() or t.is_complex() else None, non_blocking)
  File "G:\ComfyUI\comfyui_env\lib\site-packages\torch\cuda\__init__.py", line 289, in _lazy_init
    raise AssertionError("Torch not compiled with CUDA enabled")
AssertionError: Torch not compiled with CUDA enabled

2024-09-22 16:27:02,237 - root - INFO - Prompt executed in 15.26 seconds
2024-09-22 16:27:12,577 - root - INFO - got prompt
2024-09-22 16:27:12,626 - root - ERROR - !!! Exception during processing !!! Torch not compiled with CUDA enabled
2024-09-22 16:27:12,626 - root - ERROR - Traceback (most recent call last):
  File "G:\ComfyUI\execution.py", line 323, in execute
    output_data, output_ui, has_subgraph = get_output_data(obj, input_data_all, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb)
  File "G:\ComfyUI\execution.py", line 198, in get_output_data
    return_values = _map_node_over_list(obj, input_data_all, obj.FUNCTION, allow_interrupt=True, execution_block_cb=execution_block_cb, pre_execute_cb=pre_execute_cb)
  File "G:\ComfyUI\execution.py", line 169, in _map_node_over_list
    process_inputs(input_dict, i)
  File "G:\ComfyUI\execution.py", line 158, in process_inputs
    results.append(getattr(obj, func)(**inputs))
  File "G:\ComfyUI\custom_nodes\ComfyUI_CatVTON_Wrapper\py\cat_vton.py", line 42, in catvton
    pipeline = CatVTONPipeline(
  File "G:\ComfyUI\custom_nodes\ComfyUI_CatVTON_Wrapper\py\catvton\pipeline.py", line 42, in __init__
    self.vae = AutoencoderKL.from_pretrained(os.path.join(folder_paths.models_dir, "CatVTON", "sd-vae-ft-mse")).to(device, dtype=weight_dtype)
  File "G:\ComfyUI\comfyui_env\lib\site-packages\torch\nn\modules\module.py", line 1160, in to
    return self._apply(convert)
  File "G:\ComfyUI\comfyui_env\lib\site-packages\torch\nn\modules\module.py", line 810, in _apply
    module._apply(fn)
  File "G:\ComfyUI\comfyui_env\lib\site-packages\torch\nn\modules\module.py", line 810, in _apply
    module._apply(fn)
  File "G:\ComfyUI\comfyui_env\lib\site-packages\torch\nn\modules\module.py", line 833, in _apply
    param_applied = fn(param)
  File "G:\ComfyUI\comfyui_env\lib\site-packages\torch\nn\modules\module.py", line 1158, in convert
    return t.to(device, dtype if t.is_floating_point() or t.is_complex() else None, non_blocking)
  File "G:\ComfyUI\comfyui_env\lib\site-packages\torch\cuda\__init__.py", line 289, in _lazy_init
    raise AssertionError("Torch not compiled with CUDA enabled")
AssertionError: Torch not compiled with CUDA enabled

2024-09-22 16:27:12,627 - root - INFO - Prompt executed in 0.05 seconds

Attached Workflow

Please make sure that workflow does not contain any sensitive information such as API keys or passwords.

{"last_node_id":6,"last_link_id":6,"nodes":[{"id":3,"type":"LayerMask: SegmentAnythingUltra V2","pos":{"0":414,"1":424},"size":{"0":428.4000244140625,"1":342},"flags":{},"order":2,"mode":0,"inputs":[{"name":"image","type":"IMAGE","link":1}],"outputs":[{"name":"image","type":"IMAGE","links":null,"shape":3},{"name":"mask","type":"MASK","links":[2,3],"slot_index":1,"shape":3}],"properties":{"Node name for S&R":"LayerMask: SegmentAnythingUltra V2"},"widgets_values":["sam_vit_h (2.56GB)","GroundingDINO_SwinT_OGC (694MB)",0.3,"VITMatte",6,6,0.01,0.99,false,"shirt, pants,","cuda",2],"color":"rgba(27, 80, 119, 0.7)"},{"id":4,"type":"LayerMask: MaskPreview","pos":{"0":903,"1":444},"size":{"0":277.20001220703125,"1":246},"flags":{},"order":3,"mode":0,"inputs":[{"name":"mask","type":"MASK","link":2}],"outputs":[],"properties":{"Node name for S&R":"LayerMask: MaskPreview"},"color":"rgba(27, 80, 119, 0.7)"},{"id":5,"type":"CatVTONWrapper","pos":{"0":901,"1":32},"size":{"0":315,"1":218},"flags":{},"order":4,"mode":0,"inputs":[{"name":"image","type":"IMAGE","link":5},{"name":"mask","type":"MASK","link":3},{"name":"refer_image","type":"IMAGE","link":4}],"outputs":[{"name":"image","type":"IMAGE","links":[6],"slot_index":0,"shape":3}],"properties":{"Node name for S&R":"CatVTONWrapper"},"widgets_values":[25,"fp16",549557970686997,"randomize",40,2.5]},{"id":6,"type":"PreviewImage","pos":{"0":1273,"1":33},"size":{"0":323.8887023925781,"1":455.15203857421875},"flags":{},"order":5,"mode":0,"inputs":[{"name":"images","type":"IMAGE","link":6}],"outputs":[],"properties":{"Node name for S&R":"PreviewImage"}},{"id":1,"type":"LoadImage","pos":{"0":45,"1":28},"size":{"0":309.2167053222656,"1":333.2966613769531},"flags":{},"order":0,"mode":0,"inputs":[],"outputs":[{"name":"IMAGE","type":"IMAGE","links":[1,5],"slot_index":0,"shape":3},{"name":"MASK","type":"MASK","links":null,"shape":3}],"properties":{"Node name for S&R":"LoadImage"},"widgets_values":["13735824_1108784102510186_1530724342781074013_o.jpg","image"]},{"id":2,"type":"LoadImage","pos":{"0":45,"1":423},"size":{"0":308.3948059082031,"1":348.0320739746094},"flags":{},"order":1,"mode":0,"inputs":[],"outputs":[{"name":"IMAGE","type":"IMAGE","links":[4],"slot_index":0,"shape":3},{"name":"MASK","type":"MASK","links":null,"shape":3}],"properties":{"Node name for S&R":"LoadImage"},"widgets_values":["71AihluQr1L._AC_SX569__upscayl_4x_ultrasharp.jpg","image"]}],"links":[[1,1,0,3,0,"IMAGE"],[2,3,1,4,0,"MASK"],[3,3,1,5,1,"MASK"],[4,2,0,5,2,"IMAGE"],[5,1,0,5,0,"IMAGE"],[6,5,0,6,0,"IMAGE"]],"groups":[],"config":{},"extra":{"ds":{"scale":0.9090909090909095,"offset":[291.0420536792561,101.14600019424024]}},"version":0.4}

Additional Context

(Please add any additional context or steps to reproduce the error here)

[chflame163]

I did not have corresponding device testing, so in order to prevent problems, I removed the code for these devices.

[foysalremon]

@chflame163 Could this be added? I have an Intel Arc 750 available for testing. Alternatively, could it be provided in a separate branch for those who need it? Non-CUDA GPUs are becoming more popular these days, so it would be great to support them.

[zixuan96]

I did not have corresponding device testing, so in order to prevent problems, I removed the code for these devices.

I'm using Apple M2 and M3 chip. I will be very grateful if you could add them back, so that other guys and I can use this extension and then provide feedback on Non-CUDA GPUs.

[chflame163]

I tried to run it on intel CPU，but was unsuccessful, so I don't plan to continue. Sorry guys.

[malinowskij]

Just edit: cat_vton.py from 'cuda' to 'mps:

But now I have other error :D :

Ok, no longer valid problem.
I modified code that now use windowed attention and it works.

But... not as expected :D
From this:

To:

Time for buying RTX with CUDA support, not apple silicon 😛

[zer0factor]

like @malinowskij . Solved with these step.

ComfyUI_CatVTON_Wrapper
cat_vton.py
device = "mps"

segment_anything
build_sam_hq.py
state_dict = torch.load(f, map_location=torch.device("mps"))

ComfyUI_CatVTON_Wrapper
attn_processor.py
Did some memory managment here.

[eljabbaryhicham]

i tried your solution but the node won't load!!

[zer0factor]

i tried your solution but the node won't load!!

You need to install whatever is missing. Which we wouldn’t know what from just one sentence.