CUDA driver initialization failed, you might not have a CUDA gpu.

[lsainzvillalba]

I have followed the installation and setup instructions. I am running now the demo training pipeline that you provide with your zebrafish dataset. And I get this error: RuntimeError: CUDA driver initialization failed, you might not have a CUDA gpu. It appears after:
Total stack number -----> 1
Noise image name -----> fish_localbrain.tif
Noise image shape -----> (6955, 492, 492)

If I run the check that you specify in the colab notebook:
if torch.cuda.is_available():
print('\033[1;31mGPU accessiable. Use GPU for computation.\033[0m')
gpu_id = torch.cuda.current_device()
total_memory = torch.cuda.get_device_properties(gpu_id).total_memory/1024/1024
alloc_memory = torch.cuda.memory_allocated(0)/1024/1024
print('GPU ID: ', gpu_id, '|', torch.cuda.get_device_name(),
'| Memory: {:.0f} MB'.format(total_memory))
! nvcc --version
print('PyTorch version: ', torch.version)
else:
print('\033[1;31mNo GPU support. Please enable GPUs for the notebook:\033[0m')
print(' 1. Navigate to Edit → Notebook Settings')
print(' 2. Select GPU from the Hardware Accelerator drop-down')

I get that torch.cuda is not available and hence it tells me to do the 1 and 2 commands. But I do not have notebook settings in Edit or nothing related to GPU settings in the drop down menus.
I am running an Ubuntu 18.04 Spyder Anaconda Python 3.6 .
So how can I solve this issue?

[cabooster]

To solve this issue, the following methods may be helpful:

1. If you have a GPU device, you can directly go to step 2. If not, you can run DeepCAD-RT directly in your browser using a cloud GPU without configuring the environment by directly clicking this link, where you can see Edit in the upper left corner.
   

2. Please check whether you install CUDA. Run the following command in a terminal window:

   nvcc -V
   

   If this results in something like the following, it means that you have installed CUDA. Please remember the CUDA version (e.g. 11.1).

   nvcc: NVIDIA (R) Cuda compiler driver
   Copyright (c) 2005-2020 NVIDIA Corporation
   Built on Mon_Oct_12_20:09:46_PDT_2020
   Cuda compilation tools, release 11.1, V11.1.105
   Build cuda_11.1.TC455_06.29190527_0
   

​ If it reports an error, you can install CUDA with this tutoiral.

3. Make sure that your Pytorch version matches your CUDA version. You can select the correct Pytorch version from https://pytorch.org/get-started/previous-versions/.

This issue is similar to the solved problem in Stackoverflow. You can also refer to the answers in this link.

[lsainzvillalba]

Thank you very much!

…

El 29 mar 2022, a las 9:14, cabooster ***@***.***> escribió: To solve this issue, the following methods may be helpful: If you have a GPU device, you can directly go to step 2. If not, you can run DeepCAD-RT directly in your browser using a cloud GPU without configuring the environment by directly clicking this link <https://colab.research.google.com/github/cabooster/DeepCAD-RT/blob/main/DeepCAD_RT_pytorch/notebooks/DeepCAD_RT_demo_colab.ipynb>, where you can see Edit in the upper left corner. <https://user-images.githubusercontent.com/71710781/160554184-2da60f4f-f341-4df0-90d2-e5e5816beb6c.png> Please check whether you install CUDA. Run the following command in a terminal window: nvcc -V If this results in something like the following, it means that you have installed CUDA. Please remember the CUDA version (e.g. 11.1). nvcc: NVIDIA (R) Cuda compiler driver Copyright (c) 2005-2020 NVIDIA Corporation Built on Mon_Oct_12_20:09:46_PDT_2020 Cuda compilation tools, release 11.1, V11.1.105 Build cuda_11.1.TC455_06.29190527_0 ​ If it reports an error, you can install CUDA with this tutoiral <https://docs.nvidia.com/cuda/pdf/CUDA_Installation_Guide_Linux.pdf>. Make sure that your Pytorch version matches your CUDA version. You can select the correct Pytorch version from https://pytorch.org/get-started/previous-versions/ <https://pytorch.org/get-started/previous-versions/>. This issue is similar to the solved problem in Stackoverflow <https://stackoverflow.com/questions/60987997/why-torch-cuda-is-available-returns-false-even-after-installing-pytorch-with>. You can also refer to the answers in this link. — Reply to this email directly, view it on GitHub <#8 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ASYWKR6H5C2Z5JGAZZ5BOILVCKUUVANCNFSM5R2XHGLQ>. You are receiving this because you authored the thread.

[lsainzvillalba]

I found also this video really useful to solve the issue dependencies for cuda installation in ubuntu. https://www.youtube.com/watch?v=GljujCLixzE