[CUDA] CUDA_HOME environment variable is not set. Please set it to your CUDA install root

OSError: CUDA_HOME environment variable is not set. Please set it to your CUDA install root.

에러의 원인은 pytorch가 gpu가 아닌 cpu로 (재)설치 되었을 수 있습니다.

torch.cuda.is_available()이 False이므로 pytorch가 cpu 버전으로 깔렸음을 알 수 있습니다.

python
Python 3.8.6 | packaged by conda-forge | (default, Jan 25 2021, 23:21:18) 
[GCC 9.3.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import torch
>>> torch.cuda.is_available()
False
>>> print(torch.__version__)
2.4.0

conda list로 확인한 결과 torch 관련 모듈이 전부 cpu인 것을 확인할 수 있습니다.

conda list

# Name                    Version                   Build    Channel
...
pytorch                   2.4.0               py3.8_cpu_0    pytorch
...
torchaudio                2.4.0                  py38_cpu    pytorch
torchvision               0.19.0                 py38_cpu    pytorch

conda uninstall pytorch이후 nvcc -V로 확인한 cuda 버전과 맞는 torch를 설치해줍니다.

nvcc -V
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2022 NVIDIA Corporation
Built on Wed_Sep_21_10:33:58_PDT_2022
Cuda compilation tools, release 11.8, V11.8.89
Build cuda_11.8.r11.8/compiler.31833905_0

conda install pytorch torchvision torchaudio pytorch-cuda=11.8 -c pytorch -c nvidia

다시 pip install -e .를 하여 제대로 .cu 관련 submodule을 설치 완료하였습니다.

pip install -e .
Obtaining file:///mai_nas/KHS/SoSC/submodules/diff-gaussian-rasterization-depth
  Preparing metadata (setup.py) ... done
Installing collected packages: diff_gaussian_rasterization_depth
  Running setup.py develop for diff_gaussian_rasterization_depth
Successfully installed diff_gaussian_rasterization_depth-0.0.0