Closed michael-pptf closed 9 months ago
I think there can be two probable reasons:
python_embeded\python.exe pip install torch torchvision --index-url https://download.pytorch.org/whl/cu118
Hello get this error:
[ReActor] 01:39:13 - STATUS - Working: source face index [0], target face index [0] [ReActor] 01:39:14 - STATUS - Analyzing Source Image... 2023-12-04 01:39:14.4227271 [E:onnxruntime:Default, provider_bridge_ort.cc:1480 onnxruntime::TryGetProviderInfo_CUDA] D:\a_work\1\s\onnxruntime\core\session\provider_bridge_ort.cc:1193 onnxruntime::ProviderLibrary::Get [ONNXRuntimeError] : 1 : FAIL : LoadLibrary failed with error 126 "" when trying to load "C:\Users\Roberto\stable\ComfyUI_windows_portable\python_embeded\Lib\site-packages\onnxruntime\capi\onnxruntime_providers_cuda.dll"
EP Error D:\a_work\1\s\onnxruntime\python\onnxruntime_pybind_state.cc:743 onnxruntime::python::CreateExecutionProviderInstance CUDA_PATH is set but CUDA wasn't able to be loaded. Please install the correct version of CUDA and cuDNN as mentioned in the GPU requirements page (https://onnxruntime.ai/docs/execution-providers/CUDA-ExecutionProvider.html#requirements), make sure they're in the PATH, and that your GPU is supported. when using ['CUDAExecutionProvider'] Falling back to ['CUDAExecutionProvider', 'CPUExecutionProvider'] and retrying. 2023-12-04 01:39:14.5799241 [E:onnxruntime:Default, provider_bridge_ort.cc:1480 onnxruntime::TryGetProviderInfo_CUDA] D:\a_work\1\s\onnxruntime\core\session\provider_bridge_ort.cc:1193 onnxruntime::ProviderLibrary::Get [ONNXRuntimeError] : 1 : FAIL : LoadLibrary failed with error 126 "" when trying to load "C:\Users\Roberto\stable\ComfyUI_windows_portable\python_embeded\Lib\site-packages\onnxruntime\capi\onnxruntime_providers_cuda.dll"
ERROR:root:!!! Exception during processing !!! ERROR:root:Traceback (most recent call last): File "C:\Users\Roberto\stable\ComfyUI_windows_portable\python_embeded\Lib\site-packages\onnxruntime\capi\onnxruntime_inference_collection.py", line 419, in init self._create_inference_session(providers, provider_options, disabled_optimizers) File "C:\Users\Roberto\stable\ComfyUI_windows_portable\python_embeded\Lib\site-packages\onnxruntime\capi\onnxruntime_inference_collection.py", line 463, in _create_inference_session sess.initialize_session(providers, provider_options, disabled_optimizers) RuntimeError: D:\a_work\1\s\onnxruntime\python\onnxruntime_pybind_state.cc:743 onnxruntime::python::CreateExecutionProviderInstance CUDA_PATH is set but CUDA wasn't able to be loaded. Please install the correct version of CUDA and cuDNN as mentioned in the GPU requirements page (https://onnxruntime.ai/docs/execution-providers/CUDA-ExecutionProvider.html#requirements), make sure they're in the PATH, and that your GPU is supported.
The above exception was the direct cause of the following exception:
Traceback (most recent call last): File "C:\Users\Roberto\stable\ComfyUI_windows_portable\ComfyUI\execution.py", line 153, in recursive_execute output_data, output_ui = get_output_data(obj, input_data_all) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\Users\Roberto\stable\ComfyUI_windows_portable\ComfyUI\execution.py", line 83, in get_output_data return_values = map_node_over_list(obj, input_data_all, obj.FUNCTION, allow_interrupt=True) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\Users\Roberto\stable\ComfyUI_windows_portable\ComfyUI\execution.py", line 76, in map_node_over_list results.append(getattr(obj, func)(slice_dict(input_data_all, i))) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\Users\Roberto\stable\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui-reactor-node\nodes.py", line 117, in execute script.process( File "C:\Users\Roberto\stable\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui-reactor-node\scripts\reactor_faceswap.py", line 96, in process result = swap_face( ^^^^^^^^^^ File "C:\Users\Roberto\stable\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui-reactor-node\scripts\reactor_swapper.py", line 195, in swap_face source_faces = analyze_faces(source_img) ^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\Users\Roberto\stable\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui-reactor-node\scripts\reactor_swapper.py", line 114, in analyze_faces face_analyser = copy.deepcopy(getAnalysisModel()) ^^^^^^^^^^^^^^^^^^ File "C:\Users\Roberto\stable\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui-reactor-node\scripts\reactor_swapper.py", line 69, in getAnalysisModel ANALYSIS_MODEL = insightface.app.FaceAnalysis( ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\Users\Roberto\stable\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui-reactor-node\reactor_log_patch.py", line 48, in patched_faceanalysis_init model = model_zoo.get_model(onnx_file, kwargs) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\Users\Roberto\stable\ComfyUI_windows_portable\python_embeded\Lib\site-packages\insightface\model_zoo\model_zoo.py", line 96, in get_model model = router.get_model(providers=providers, provider_options=provider_options) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\Users\Roberto\stable\ComfyUI_windows_portable\ComfyUI\custom_nodes\comfyui-reactor-node\reactor_log_patch.py", line 21, in patched_get_model session = PickableInferenceSession(self.onnx_file, kwargs) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\Users\Roberto\stable\ComfyUI_windows_portable\python_embeded\Lib\site-packages\insightface\model_zoo\model_zoo.py", line 25, in init super().init(model_path, kwargs) File "C:\Users\Roberto\stable\ComfyUI_windows_portable\python_embeded\Lib\site-packages\onnxruntime\capi\onnxruntime_inference_collection.py", line 430, in init raise fallback_error from e File "C:\Users\Roberto\stable\ComfyUI_windows_portable\python_embeded\Lib\site-packages\onnxruntime\capi\onnxruntime_inference_collection.py", line 425, in init self._create_inference_session(self._fallback_providers, None) File "C:\Users\Roberto\stable\ComfyUI_windows_portable\python_embeded\Lib\site-packages\onnxruntime\capi\onnxruntime_inference_collection.py", line 463, in _create_inference_session sess.initialize_session(providers, provider_options, disabled_optimizers) RuntimeError: D:\a_work\1\s\onnxruntime\python\onnxruntime_pybind_state.cc:743 onnxruntime::python::CreateExecutionProviderInstance CUDA_PATH is set but CUDA wasn't able to be loaded. Please install the correct version of CUDA and cuDNN as mentioned in the GPU requirements page (https://onnxruntime.ai/docs/execution-providers/CUDA-ExecutionProvider.html#requirements), make sure they're in the PATH, and that your GPU is supported.
Tried python_embeded\python.exe pip install torch torchvision --index-url https://download.pytorch.org/whl/cu118
get this: PS C:\Users\Roberto\stable\ComfyUI_windows_portable> python_embeded\python.exe pip install torch torchvision --index-url https://download.pytorch.org/whl/cu118 C:\Users\Roberto\stable\ComfyUI_windows_portable\python_embeded\python.exe: can't open file 'C:\Users\Roberto\stable\ComfyUI_windows_portable\pip': [Errno 2] No such file or directory PS C:\Users\Roberto\stable\ComfyUI_windows_portable>
Hello! Sorry, my fault... Missed "-m" flag before pip 🤦♂️
python_embeded\python.exe -m pip install torch torchvision --index-url https://download.pytorch.org/whl/cu118
Im actually have the exact same issue right now:
Error occurred when executing ReActorFaceSwap:
D:\a\_work\1\s\onnxruntime\python\onnxruntime_pybind_state.cc:743 onnxruntime::python::CreateExecutionProviderInstance CUDA_PATH is set but CUDA wasn't able to be loaded. Please install the correct version of CUDA and cuDNN as mentioned in the GPU requirements page (https://onnxruntime.ai/docs/execution-providers/CUDA-ExecutionProvider.html#requirements), make sure they're in the PATH, and that your GPU is supported.
File "Y:\ComfyUI\execution.py", line 153, in recursive_execute
output_data, output_ui = get_output_data(obj, input_data_all)
File "Y:\ComfyUI\execution.py", line 83, in get_output_data
return_values = map_node_over_list(obj, input_data_all, obj.FUNCTION, allow_interrupt=True)
File "Y:\ComfyUI\execution.py", line 76, in map_node_over_list
results.append(getattr(obj, func)(**slice_dict(input_data_all, i)))
File "Y:\ComfyUI\custom_nodes\comfyui-reactor-node\nodes.py", line 117, in execute
script.process(
File "Y:\ComfyUI\custom_nodes\comfyui-reactor-node\scripts\reactor_faceswap.py", line 96, in process
result = swap_face(
File "Y:\ComfyUI\custom_nodes\comfyui-reactor-node\scripts\reactor_swapper.py", line 195, in swap_face
source_faces = analyze_faces(source_img)
File "Y:\ComfyUI\custom_nodes\comfyui-reactor-node\scripts\reactor_swapper.py", line 114, in analyze_faces
face_analyser = copy.deepcopy(getAnalysisModel())
File "Y:\ComfyUI\custom_nodes\comfyui-reactor-node\scripts\reactor_swapper.py", line 69, in getAnalysisModel
ANALYSIS_MODEL = insightface.app.FaceAnalysis(
File "Y:\ComfyUI\venv\lib\site-packages\insightface\app\face_analysis.py", line 31, in __init__
model = model_zoo.get_model(onnx_file, **kwargs)
File "Y:\ComfyUI\venv\lib\site-packages\insightface\model_zoo\model_zoo.py", line 96, in get_model
model = router.get_model(providers=providers, provider_options=provider_options)
File "Y:\ComfyUI\venv\lib\site-packages\insightface\model_zoo\model_zoo.py", line 40, in get_model
session = PickableInferenceSession(self.onnx_file, **kwargs)
File "Y:\ComfyUI\venv\lib\site-packages\insightface\model_zoo\model_zoo.py", line 25, in __init__
super().__init__(model_path, **kwargs)
File "Y:\ComfyUI\venv\lib\site-packages\onnxruntime\capi\onnxruntime_inference_collection.py", line 430, in __init__
raise fallback_error from e
File "Y:\ComfyUI\venv\lib\site-packages\onnxruntime\capi\onnxruntime_inference_collection.py", line 425, in __init__
self._create_inference_session(self._fallback_providers, None)
File "Y:\ComfyUI\venv\lib\site-packages\onnxruntime\capi\onnxruntime_inference_collection.py", line 463, in _create_inference_session
sess.initialize_session(providers, provider_options, disabled_optimizers)
Running in a venv of course. Here is what the venv sees:
(venv) PS Y:\ComfyUI> nvidia-smi
Mon Dec 4 14:11:25 2023
+---------------------------------------------------------------------------------------+
| NVIDIA-SMI 546.01 Driver Version: 546.01 CUDA Version: 12.3 |
|-----------------------------------------+----------------------+----------------------+
| GPU Name TCC/WDDM | Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|=========================================+======================+======================|
| 0 NVIDIA GeForce RTX 3090 WDDM | 00000000:01:00.0 On | N/A |
| 59% 67C P2 303W / 420W | 9687MiB / 24576MiB | 52% Default |
| | | N/A |
+-----------------------------------------+----------------------+----------------------+
Not sure why its not loading cuda
Hello! Sorry, my fault... Missed "-m" flag before pip 🤦♂️
python_embeded\python.exe -m pip install torch torchvision --index-url https://download.pytorch.org/whl/cu118
Hello. I tried but this is what it tells me: PS C:\Users\Roberto\stable\ComfyUI_01> python_embeded\python.exe -m pip install torch torchvision --index-url https://download.pytorch.org/whl/cu118 Looking in indexes: https://download.pytorch.org/whl/cu118 Requirement already satisfied: torch in c:\users\roberto\stable\comfyui_01\python_embeded\lib\site-packages (2.1.1+cu121) Requirement already satisfied: torchvision in c:\users\roberto\stable\comfyui_01\python_embeded\lib\site-packages (0.16.1+cu121) Requirement already satisfied: filelock in c:\users\roberto\stable\comfyui_01\python_embeded\lib\site-packages (from torch) (3.13.1) Requirement already satisfied: typing-extensions in c:\users\roberto\stable\comfyui_01\python_embeded\lib\site-packages (from torch) (4.8.0) Requirement already satisfied: sympy in c:\users\roberto\stable\comfyui_01\python_embeded\lib\site-packages (from torch) (1.12) Requirement already satisfied: networkx in c:\users\roberto\stable\comfyui_01\python_embeded\lib\site-packages (from torch) (3.2.1) Requirement already satisfied: jinja2 in c:\users\roberto\stable\comfyui_01\python_embeded\lib\site-packages (from torch) (3.1.2) Requirement already satisfied: fsspec in c:\users\roberto\stable\comfyui_01\python_embeded\lib\site-packages (from torch) (2023.10.0) Requirement already satisfied: numpy in c:\users\roberto\stable\comfyui_01\python_embeded\lib\site-packages (from torchvision) (1.26.2) Requirement already satisfied: requests in c:\users\roberto\stable\comfyui_01\python_embeded\lib\site-packages (from torchvision) (2.31.0) Requirement already satisfied: pillow!=8.3.*,>=5.3.0 in c:\users\roberto\stable\comfyui_01\python_embeded\lib\site-packages (from torchvision) (9.5.0) Requirement already satisfied: MarkupSafe>=2.0 in c:\users\roberto\stable\comfyui_01\python_embeded\lib\site-packages (from jinja2->torch) (2.1.3) Requirement already satisfied: charset-normalizer<4,>=2 in c:\users\roberto\stable\comfyui_01\python_embeded\lib\site-packages (from requests->torchvision) (3.3.2) Requirement already satisfied: idna<4,>=2.5 in c:\users\roberto\stable\comfyui_01\python_embeded\lib\site-packages (from requests->torchvision) (3.4) Requirement already satisfied: urllib3<3,>=1.21.1 in c:\users\roberto\stable\comfyui_01\python_embeded\lib\site-packages (from requests->torchvision) (1.26.18) Requirement already satisfied: certifi>=2017.4.17 in c:\users\roberto\stable\comfyui_01\python_embeded\lib\site-packages (from requests->torchvision) (2023.11.17) Requirement already satisfied: mpmath>=0.19 in c:\users\roberto\stable\comfyui_01\python_embeded\lib\site-packages (from sympy->torch) (1.3.0) PS C:\Users\Roberto\stable\ComfyUI_01>
Not sure why its not loading cuda
The environment doesnt seem to know about it:
(venv) PS Y:\ComfyUI> python
Python 3.10.6 (tags/v3.10.6:9c7b4bd, Aug 1 2022, 21:53:49) [MSC v.1932 64 bit (AMD64)] on win32
Type "help", "copyright", "credits" or "license" for more information.
>>> import tensorflow as tf
2023-12-04 15:01:59.300300: I tensorflow/core/util/port.cc:113] oneDNN custom operations are on. You may see slightly different numerical results due to floating-point round-off errors from different computation orders. To turn them off, set the environment variable `TF_ENABLE_ONEDNN_OPTS=0`.
WARNING:tensorflow:From Y:\ComfyUI\venv\lib\site-packages\keras\src\losses.py:2976: The name tf.losses.sparse_softmax_cross_entropy is deprecated. Please use tf.compat.v1.losses.sparse_softmax_cross_entropy instead.
>>> tf.config.list_physical_devices('GPU')
[]
CUDA Execution Provider requires CUDA 11.8 https://onnxruntime.ai/docs/execution-providers/CUDA-ExecutionProvider.html You guys just need to download the appropriate build of Comfy This one for example (latest with CU11.8): https://github.com/comfyanonymous/ComfyUI/releases/download/latest/ComfyUI_windows_portable_nvidia_cu118_or_cpu_23_09_2023.7z Or any other you want but with cu118 https://github.com/comfyanonymous/ComfyUI/releases
I will add it to the documentation
Ah, I gotcha, the current iteration of reactor-node requests cuda11? I should have looked for that requirement, thanks. Any chance it'll be updated to support 12x, since thats the direction Comfy has gone?
Edit: Nevermind, I see that its onxx that is requiring it, so I assume you need to wait for them to add support.
Edit: Nevermind, I see that its onxx that is requiring it, so I assume you need to wait for them to add support.
Yeah, that's right
@Gourieff and @ilcristofer I was able to get this all working with CUDA 12. Just needed to make wheel build of onxx using CUDA/NN 12.
git clone --recursive https://github.com/Microsoft/onnxruntime
cd onnxruntime
.\build.bat --update --build --use_cuda --cudnn_home "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDNN\cudnn" --cuda_home "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.2" --skip_onnx_tests --build_wheel --parallel
Obvs change the paths to wherever you have cuda/cudnn
Its now working just fine in the latest version of ComfyUI I have a wheel file if you just want to pip install it @ilcristofer - feel free to inspect it first if safety is a concern. https://file.io/4RAvXi1dHVCK
@4DeepThought2 brilliant! 👍
here is my case: after start server:
Total VRAM 24576 MB, total RAM 32679 MB Set vram state to: NORMAL_VRAM Device: cuda:0 NVIDIA GeForce RTX 3090 : cudaMallocAsync VAE dtype: torch.bfloat16 Torch version: 2.1.1+cu121 Total VRAM 24576 MB, total RAM 32679 MB Set vram state to: NORMAL_VRAM Device: cuda:0 NVIDIA GeForce RTX 3090 : cudaMallocAsync VAE dtype: torch.bfloat16 Torch version: 2.1.1+cu121
I installed: ComfyUI_windows_portable_nvidia_cu121_or_cpu.7z
still met the same problem, any suggestion?
@Gourieff and @ilcristofer I was able to get this all working with CUDA 12. Just needed to make wheel build of onxx using CUDA/NN 12.
git clone --recursive https://github.com/Microsoft/onnxruntime cd onnxruntime .\build.bat --update --build --use_cuda --cudnn_home "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDNN\cudnn" --cuda_home "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.2" --skip_onnx_tests --build_wheel --parallel
Obvs change the paths to wherever you have cuda/cudnn
Its now working just fine in the latest version of ComfyUI I have a wheel file if you just want to pip install it @ilcristofer - feel free to inspect it first if safety is a concern. https://file.io/4RAvXi1dHVCK
@4DeepThought2 thanks for the support. I tried to do as you say and after the command .\build.bat --update --build --use_cuda --cudnn_home "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\CUDDN\v8.9.7" --cuda_home "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.1" --skip_onnx_tests --build_wheel --parallel
I get this error 2023-12-07 23:45:57.006 build [ERROR] - Failed to resolve executable path for 'cmake'.
You need to install cmake for windows: https://cmake.org/download/
Yea, I gave up on this since downgrading to CUDA 11.8 would break so many other plugins.
Since it could work with CUDA 11.8, I am closing this issue.
@Gourieff and @ilcristofer I was able to get this all working with CUDA 12. Just needed to make wheel build of onxx using CUDA/NN 12.
git clone --recursive https://github.com/Microsoft/onnxruntime cd onnxruntime .\build.bat --update --build --use_cuda --cudnn_home "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDNN\cudnn" --cuda_home "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.2" --skip_onnx_tests --build_wheel --parallel
Obvs change the paths to wherever you have cuda/cudnn
Its now working just fine in the latest version of ComfyUI I have a wheel file if you just want to pip install it @ilcristofer - feel free to inspect it first if safety is a concern. https://file.io/4RAvXi1dHVCK
You need to install cmake for windows: https://cmake.org/download/
Hello. I tried as suggested and now it works, but not on the portable version. I did a manual installation of comfyui and it works (probably because it doesn't use the embedded Python version, but the standard one installed on the PC).
I believe, however, that starting from this, it's possible to modify the node to work with the latest portable version of comfyui.
Can the node be set to use the CPU instead?
I was about to open a new thread but it seems that I'm not the only one. So~~I'm pretty sure I had no more issues yesterday but here I am again. Now it's CUDA xD
I tried to follow the advices here but really I have no idea what I'm doing. I'm using ComfyUI portable (like, there is another version ?) and well, it didn't work.
https://onnxruntime.ai/docs/execution-providers/CUDA-ExecutionProvider.html#requirements
ComfyUI has a lot of builds on their Releases page ORT-GPU requires CUDA 11.8 (this is official requirements) So, you should download latest ComfyUI with Cu118 https://github.com/comfyanonymous/ComfyUI/releases On some Machines it works fine with Cu121, but on some - not (I don't know what it depends on...) But ORT-GPU 100% works with Cu118
Today Bing AI told me about Pinokio, from there I found out about FaceFusion which is okay, but not as good as your Reactor in ComfyUI (and it's censored) so~~ I have no other choice but to keep trying but to be honest, I'm tired of all that Python nightmare x) I already reinstalled ComfyUI and everything last time (like 2 days ago ?), I used the Download Link with stable pytorch 2.1 cu121 maybe that's how I end up with CUDA121 I don't know. I already read that onnxruntime doc page but it's very hardcore to put my head around this, anyway, should I install the unstable nightly build instead ? (isn't that unstable enough ? xDDD ) Or should I uninstall Cuda121 and put 118 instead (Cuda,Cdnn, maybe it's the same thing hell I don't know) I want to see the bottom of this
okay case closed (for now), I installed ComfyUI using Pinokio, reinstalled all the custom nodes for the 4th time this week x) AND NOW IT WORKS AGAIN I don't get it but I don't care anymore x) (for now lol)
Can the node be set to use the CPU instead?
If you want to force the node to use CPU you can go to ComfyUI\custom_nodes\comfyui-reactor-node\scripts\reactor_swapper.py and then in line 29 change providers to ["CPUExecutionProvider"]. It's a lazy workaround but it works for me.
@Gourieff and @ilcristofer I was able to get this all working with CUDA 12. Just needed to make wheel build of onxx using CUDA/NN 12.
git clone --recursive https://github.com/Microsoft/onnxruntime cd onnxruntime .\build.bat --update --build --use_cuda --cudnn_home "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDNN\cudnn" --cuda_home "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.2" --skip_onnx_tests --build_wheel --parallel
Obvs change the paths to wherever you have cuda/cudnn
Its now working just fine in the latest version of ComfyUI I have a wheel file if you just want to pip install it @ilcristofer - feel free to inspect it first if safety is a concern. https://file.io/4RAvXi1dHVCK
where should we "git clone --recursive https://github.com/Microsoft/onnxruntime" ? like what dir ?
I found 'onnxruntime' & 'onnxruntime_gpu-1.17.0.dist-info' folders in : C:\Users(name)\AppData\Local\Programs\Python\Python310\Lib\site-packages\onnxruntime
is that it?
Can the node be set to use the CPU instead?
If you want to force the node to use CPU you can go to ComfyUI\custom_nodes\comfyui-reactor-node\scripts\reactor_swapper.py and then in line 29 change providers to ["CPUExecutionProvider"]. It's a lazy workaround but it works for me.
It would be nice to have some settings like in A1111 Reactor. I can set CPU or GPU in A1111 to run Reactor there.
Ran into this issue running cuda 12.1 on linux. What fixed it for me was
pip install ort-nightly-gpu==1.17.0.dev20240118002 --index-url https://aiinfra.pkgs.visualstudio.com/PublicPackages/_packaging/ort-cuda-12-nightly/pypi/simple/
which I found here
Can the node be set to use the CPU instead?
If you want to force the node to use CPU you can go to ComfyUI\custom_nodes\comfyui-reactor-node\scripts\reactor_swapper.py and then in line 29 change providers to ["CPUExecutionProvider"]. It's a lazy workaround but it works for me.
If on Linux and AMD(with RoCM) its also possible to just change the CUDAExecutionProvider to ROCMExecutionProvider and it works fine.
@4DeepThought2 It reported an error saying it couldn't be found, but I installed it.
@Gourieff and @ilcristofer I was able to get this all working with CUDA 12. Just needed to make wheel build of onxx using CUDA/NN 12.
git clone --recursive https://github.com/Microsoft/onnxruntime cd onnxruntime .\build.bat --update --build --use_cuda --cudnn_home "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDNN\cudnn" --cuda_home "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.2" --skip_onnx_tests --build_wheel --parallel
Obvs change the paths to wherever you have cuda/cudnn
Its now working just fine in the latest version of ComfyUI I have a wheel file if you just want to pip install it @ilcristofer - feel free to inspect it first if safety is a concern. https://file.io/4RAvXi1dHVCK
Please, be more detailed when posting. After compiling, what did you do? Where is the compiled file? What to do with it? Thanks.
To troubleshoot CUDA and ONNX Runtime issues:
Use Dependency Walker: Download and run Dependency Walker (http://www.dependencywalker.com/) to analyze onnxruntime_providers_cuda.dll. This will help identify missing dependencies. Check for missing files: The analysis revealed that the following files were missing:
cudnn64_8.dll cudnn_ops_infer64_8.dll cublasLt64_11.dll cublas64_11.dll
Locate the missing files: These files were found in the Python site-packages directory: AppData\Local\Programs\Python\Python39\Lib\site-packages Copy the files: The missing files were manually copied to the CUDA binary directory: C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.6\bin Verify the solution: After copying the files, the CUDA-related issues with ONNX Runtime were resolved.
This manual file relocation solved the immediate problem by ensuring all required CUDA and cuDNN libraries were in a location accessible to the system and ONNX Runtime. However, for a more robust long-term solution, it's recommended to properly reinstall CUDA and cuDNN according to NVIDIA's official documentation to ensure all components are correctly placed and registered in the system.
I have faced this error many times before but this video has solved the issue most of the time.
https://www.youtube.com/watch?v=Ywc5BUN5SSc&list=WL&index=1&t=10s
First, confirm
What happened?
Runtime error when ComfyUI reaches Reactor.
Steps to reproduce the problem
Your workflow
Sysinfo
Windows 10 LTSC, AMD Ryzen 7 7700X + RTX 3090 ImpactPack Controlnet Davemane42 Derfuu_Nodes facerestore_cf rgthree
Relevant console log
Additional information
No response