Dennis Núñez

Install Nvidia CUDA 9.0 and cuDNN 7.0 on Ubuntu 16.04 for Deep Learning

1. Install CUDA:

Go to: https://developer.nvidia.com/cuda-90-download-archive.

Select: LINUX --> x86_64 --> Ubuntu --> 16.04 --> deb (network).

Then:

sudo dpkg -i cuda-repo-ubuntu1604_9.0.176-1_amd64.deb sudo apt-key adv --fetch-keys http://developer.download.nvidia.com/compute/cuda/repos/ubuntu1604/x86_64/7fa2af80.pub sudo apt-get update sudo apt-get install cuda=9.0.176-1

Set up the development environment by modifying the PATH and LD_LIBRARY_PATH variables by adding them to the end of .bashrc file ( sudo gedit .bashrc ):

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

2. Install cuDNN:

Go to: https://developer.nvidia.com/cuda-90-download-archive.

Select: "Download cuDNN v7.0.5 (Dec 5, 2017), for CUDA 9.0" --> "cuDNN v7.0.5 Library for Linux".

Download the .tgz file.

Navigate to your directory containing the cuDNN Tar file (your cuDNN download path, usually ~/Downloads/):

cd ~/Downloads/

Unzip the cuDNN package:

tar -xzvf cudnn-10.2-linux-x64-v7.6.5.32.tgz

Copy the following files into the CUDA Toolkit directory, and change the file permissions:

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*

References:

- https://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#post-installation-actions.

- https://docs.nvidia.com/deeplearning/sdk/cudnn-install/index.html.