Multiple-Version-of-CUDA-On-The-Same-Machine

Install CUDA

Since .deb file might override your current CUDA driver, I recommend installing it with runfile at NVIDIA Official Website After downloading CUDA runfile, run

sudo sh cuda_10.0.130_410.48_linux.run --silent --toolkit --override

where --silent means do everything without any interactive prompt and --override means set the cuda to default.

You may set any version of cuda to default by

sudo ln -nsf /usr/local/cuda-10.0 /usr/local/cuda

Install CuDNN

Download cuDNN .tgz file at CUDNN then

tar -xzvf cudnn-10.0-linux-x64-v7.6.4.38.tgz
sudo cp cuda/include/cudnn.h /usr/local/cuda/include
sudo cp cuda/lib64/libcudnn* /usr/local/cuda/lib64
sudo chmod a+r /usr/local/cuda/include/cudnn.h /usr/local/cuda/lib64/libcudnn*

Export LD_LIBRARY_PATH

Add all CUDA library paths to $LD_LIBRARY_PATH

export LD_LIBRARY_PATH=/usr/local/cuda-10.1/lib64:/usr/local/cuda-10.0/lib64

That's it!

Issues

Unsupported Compiler: X.X.X. Use --override to override this check

Install gcc-6 and g++-6, then link gcc-6 & g++-6 to gcc & g++

sudo apt install gcc-6 g++-6
cd /usr/bin/
sudo ln -sf gcc-6 gcc
sudo ln -sf g++-6 g++

then you are good to keep on installing cuda, remember to link the gcc & g++ back to what it was