daveboat / ubuntu_cuda_instructions

Some directions (mainly for my own use) for how to set up CUDA, CUDnn for deep learning on Ubuntu 18.04

Installing CUDA 10.0 and CUDnn 7.6 on Ubuntu 18.04, for pytorch (~1.3) and tensorflow (~2.0, 1.14-1.15) to both be able to use the GPU Also git, anaconda, pytorch, tensorflow, etc

1. Install nvidia drivers if they aren't installed already

• Add nvidia PPA

sudo add-apt-repository ppa:graphics-drivers
sudo apt-get update

• Install latest nvidia driver for the graphics card

sudo apt-get install nvidia-driver-430

• Reboot

sudo reboot

• Check nvidia driver is installed

nvidia-smi

2. Install CUDA

• Install dependencies

sudo apt-get install freeglut3 freeglut3-dev libxi-dev libxmu-dev

• Choose the correct downloads from https://developer.nvidia.com/cuda-downloads

wget https://developer.nvidia.com/compute/cuda/10.0/Prod/local_installers/cuda_10.0.130_410.48_linux
wget http://developer.download.nvidia.com/compute/cuda/10.0/Prod/patches/1/cuda_10.0.130.1_linux.run

• Install CUDA 10.0 and patch (Don't install driver, since we already installed the driver)

sudo sh cuda_10.0.130_410.48_linux
sudo sh cuda_10.0.130.1_linux.run

• Add CUDA directories to PATH variable

echo 'export PATH=/usr/local/cuda-10.0/bin:$PATH' >> ~/.bashrc
echo 'export LD_LIBRARY_PATH=/usr/local/cuda-10.0/lib64:$LD_LIBRARY_PATH' >> ~/.bashrc
source ~/.bashrc
sudo ldconfig

• Check CUDA is installed

nvcc --version

3. Install CUDnn

• Download the correct version of CUDnn (I got CUDnn version 7.6 for Cuda 10.0) from https://developer.nvidia.com/cudnn (requires an nVidia developer account)

• Install

sudo dpkg -i libcudnn7_7.6.5.32-1+cuda10.0_amd64.deb

4. Install Anaconda

• Get install script from https://www.anaconda.com/distribution/#linux

wget https://repo.anaconda.com/archive/Anaconda3-2019.10-Linux-x86_64.sh

• Run install script (do not sudo!)

sh Anaconda3-2019.10-Linux-x86_64.sh

5. (Optional) Install Git

sudo apt install git

6. (Optional) Make a dummy conda environment, install pytorch and tensorflow, check that they can access the GPU

conda create -n dummy python=3.7
conda activate dummy
pip install tensorflow-gpu
pip install torch torchvision

python
>>> import torch
>>> torch.cuda.is_available()
>>> import tensorflow as tf
>>> tf.test.is_gpu_available()