Check GPU with Python

My codes to check if GPUs are available after system reinstallation and creating new conda envs.

conda environments and packages backup

Install Python Packages

conda install -c conda-forge opencv

conda install -c conda-forge matplotlib
conda install -c conda-forge tqdm

conda install -c conda-forge notebook nb_conda_kernels jupyterlab tqdm
conda install -c conda-forge ipywidgets

One line:

conda install -c conda-forge matplotlib tqdm notebook nb_conda_kernels ipywidgets ipykernel

PyTorch and CUDA related

https://anaconda.org/nvidia/cuda-toolkit
conda install -c nvidia cuda-toolkit

Install PyTorch for CUDA 11.7

PyTorch GET STARTED

conda install pytorch torchvision torchaudio pytorch-cuda=12.1 -c pytorch -c nvidia

Check the cuda version of pytorch installation (or if the pytorch supports cuda):

python3 -c "import torch; print(\"CUDA:\", torch.version.cuda)"

python3 -c "import torch; print('Torch:', torch.__version__); print('CUDA:', torch.version.cuda); import torchvision; print('Torchvision:', torchvision.__version__)"

python3 -c "import torch; print('Torch:', torch.__version__); print('CUDA:', torch.version.cuda); import torchvision; print('Torchvision:', torchvision.__version__); print([i/2**30 for i in torch.cuda.mem_get_info()])"

Exports for CUDA (needed for tensorflow)

export CUDA_HOME=/usr/local/cuda
export PATH=${CUDA_HOME}/bin:${PATH}
export LD_LIBRARY_PATH=${CUDA_HOME}/lib64:$LD_LIBRARY_PATH

Problems with GCC for conda

>> `GLIBCXX_3.4.30' not found >> reference

export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$HOME/xxx/lib

Other packages

pip install gdown