Tutorial : Installing CUDA 8 on Ubuntu 16

Introduction

CUDA is a parallel computing platform and programming model that makes using a GPU for general purpose computing simple and elegant. The developer still programs in the familiar C, C++, Fortran, or an ever expanding list of supported languages, and incorporates extensions of these languages in the form of a few basic keywords.

These keywords let the developer express massive amounts of parallelism and direct the compiler to the portion of the application that maps to the GPU.

Below is the steps to perform installation of cuda8 & cuDNN 7.1.4 on ubuntu16 by removing latest nvidia packages on new instance

Installation Steps

Perform system update and upgrade

apt update
apt upgrade
reboot

Remove the existing installed Cuda Packages

apt remove cuda-*
apt autoremove

Install Nvidia driver which supports Tesla v100

wget http://us.download.nvidia.com/tesla/440.33.01/NVIDIA-Linux-x86_64-440.33.01.run
chmod +x NVIDIA-Linux-x86_64-440.33.01.run
./NVIDIA-Linux-x86_64-440.33.01.run

To check nvidia driver output

nvidia-smi

ignore cuda version which is seen in this output

Install Cuda8 debian packages

wget https://developer.nvidia.com/compute/cuda/8.0/Prod2/local_installers/cuda-repo-ubuntu1604-8-0-local-ga2_8.0.61-1_amd64-deb
dpkg -i cuda-repo-ubuntu1604-8-0-local-ga2_8.0.61-1_amd64-deb
apt-get update
apt-get install cuda-8.0

Install patch update for Cuda8

wget https://developer.nvidia.com/compute/cuda/8.0/Prod2/patches/2/cuda-repo-ubuntu1604-8-0-local-cublas-performance-update_8.0.61-1_amd64-deb
dpkg -i cuda-r  epo-ubuntu1604-8-0-local-cublas-performance-update_8.0.61-1_amd64-deb

Reboot the system

reboot

To check Cuda version

nvcc --version

To Install cuDNN version 7.1.4 for Cuda

Download three packages for ubuntu 16 from archive cudnn directory

Installation and verification steps available here