CUDA

Installation

Linux

Reference

• Install the latest nvidia drivers from the standard repo, e.g. nvidia-drivers-440
  
• Install Cuda Toolkit separately without the drivers.
  
  • Use one of the deb install options.
• You may also want to install the following:
  • cuDnn
    
  • TensorRT

Adapted from tensorflow

# Add NVIDIA package repositories
wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1804/x86_64/cuda-repo-ubuntu1804_10.1.243-1_amd64.deb
sudo apt-key adv --fetch-keys https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1804/x86_64/7fa2af80.pub
sudo dpkg -i cuda-repo-ubuntu1804_10.1.243-1_amd64.deb
sudo apt update
wget http://developer.download.nvidia.com/compute/machine-learning/repos/ubuntu1804/x86_64/nvidia-machine-learning-repo-ubuntu1804_1.0.0-1_amd64.deb
sudo apt install ./nvidia-machine-learning-repo-ubuntu1804_1.0.0-1_amd64.deb
sudo apt update

# Install NVIDIA driver
sudo apt install -y nvidia-driver-450
# Reboot. Check that GPUs are visible using the command: nvidia-smi

# Install development and runtime libraries (~4GB)
sudo apt install -y --no-install-recommends cuda-10-1 \
    libcudnn7=7.6.4.38-1+cuda10.1  \
    libcudnn7-dev=7.6.4.38-1+cuda10.1


# Install TensorRT. Requires that libcudnn7 is installed above.
sudo apt install -y --no-install-recommends libnvinfer6=6.0.1-1+cuda10.1 \
    libnvinfer-dev=6.0.1-1+cuda10.1 \
    libnvinfer-plugin6=6.0.1-1+cuda10.1

For tensorflow and pytorch, you may need to add LD_LIBRARY_PATH=/usr/local/cuda/lib64 to your environment variables.
You can also do this in PyCharm.

GCC Versions

nvcc sometimes only supports older gcc/g++ versions.
To make it use those by default, create the following symlinks:

• sudo ln -s /usr/bin/gcc-6 /usr/local/cuda/bin/gcc
• sudo ln -s /usr/bin/g++-6 /usr/local/cuda/bin/g++

References

• An Even Easier Introduction To Cuda