CUDA 11.2 Installation on Ubuntu 18.04

cuda-11.2_installation_on_Ubuntu-18.04

#!/bin/bash
## This gist contains instructions about cuda v11.2 and cudnn 8.1 installation in Ubuntu 18.04 for PyTorch

#############################################################################################
##### forked by : https://gist.github.com/Mahedi-61/2a2f1579d4271717d421065168ce6a73 ########
#############################################################################################


### steps ####
# verify the system has a cuda-capable gpu
# download and install the nvidia cuda toolkit and cudnn
# setup environmental variables
# verify the installation
###

### If you have previous installation remove it first. 
sudo apt-get purge nvidia*
sudo apt remove nvidia-*
sudo rm /etc/apt/sources.list.d/cuda*
sudo apt-get autoremove && sudo apt-get autoclean
sudo rm -rf /usr/local/cuda*


### to verify your gpu is cuda enable check
lspci | grep -i nvidia

### gcc compiler is required for development using the cuda toolkit. to verify the version of gcc install enter
gcc --version

# system update
sudo apt-get update
sudo apt-get upgrade


# install other import packages
sudo apt-get install g++ freeglut3-dev build-essential libx11-dev libxmu-dev libxi-dev libglu1-mesa libglu1-mesa-dev


# first get the PPA repository driver
sudo add-apt-repository ppa:graphics-drivers/ppa
sudo apt-key adv --fetch-keys http://developer.download.nvidia.com/compute/cuda/repos/ubuntu1804/x86_64/7fa2af80.pub
echo "deb https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1804/x86_64 /" | sudo tee /etc/apt/sources.list.d/cuda.list
sudo apt-get update

 # installing CUDA-11.2
sudo apt-get -o Dpkg::Options::="--force-overwrite" install cuda-11-2 cuda-drivers


# setup your paths
echo 'export PATH=/usr/local/cuda-11.2/bin:$PATH' >> ~/.bashrc
echo 'export LD_LIBRARY_PATH=/usr/local/cuda-11.2/lib64:$LD_LIBRARY_PATH' >> ~/.bashrc
source ~/.bashrc
sudo ldconfig

# install cuDNN v8.1

CUDNN_TAR_FILE="cudnn-11.2-linux-ppc64le-v8.1.1.33.tgz"
wget https://developer.nvidia.com/compute/machine-learning/cudnn/secure/8.1.1.33/11.2_20210301/cudnn-11.2-linux-ppc64le-v8.1.1.33.tgz
tar -xzvf ${CUDNN_TAR_FILE}


# copy the following files into the cuda toolkit directory.
sudo cp -P cuda/include/cudnn*.h /usr/local/cuda-11.2/include
sudo cp -P cuda/lib64/libcudnn* /usr/local/cuda-11.2/lib64/
sudo chmod a+r /usr/local/cuda-11.2/lib64/libcudnn*

# Finally, to verify the installation, check
nvidia-smi
nvcc -V

# install PyTorch (an open source machine learning framework)

pip install torch==1.7.1+cu110 torchvision==0.8.2+cu110 torchaudio==0.7.2 -f https://download.pytorch.org/whl/torch_stable.html

This is amazing. Just a small suggestion.
Replace sudo cp cuda/include/cudnn.h /usr/local/cuda/include with sudo cp cuda/include/cudnn*.h /usr/local/cuda/include so that it can copy cudnn_version.h as well.

• Updated

Download of Cudnn consumes about 80+ mins. Can you replace it officially url

This is really great, thank you! No suggestions, it worked perfectly for me. :)

Works perfectly, also with 11.3 CUDA version!

You are the best! Thank you so much! Works perfectly

Work like a charm. Thank you.

That cudnn link access is forbidden! Perhaps there was a change in access permissions on the server for this link.

@amehrish resolved !

This is wonderful. You are a life saver.
Just one more contribution; the wget https://developer.nvidia.com/compute/machine-learning/cudnn/secure/8.1.1.33/11.2_20210301/cudnn-11.2-linux-x64-v8.1.1.33.tgz will not work because you need to login to your NVIDIA account to download the tar file.
However, there is another open download link http://people.cs.uchicago.edu/~kauffman/nvidia/cudnn/cudnn-11.2-linux-x64-v8.2.0.53.tgz. Use that in the cudNN install command instead.
You can tar for more cuDNN versions here
http://people.cs.uchicago.edu/~kauffman/nvidia/cudnn/

Thank you...

I am going to add your suggestion link.

This link to download cudnn did not work for me.

This is wonderful. You are a life saver.
Just one more contribution; the wget https://developer.nvidia.com/compute/machine-learning/cudnn/secure/8.1.1.33/11.2_20210301/cudnn-11.2-linux-x64-v8.1.1.33.tgz will not work because you need to login to your NVIDIA account to download the tar file.
However, there is another open download link http://people.cs.uchicago.edu/~kauffman/nvidia/cudnn/cudnn-11.2-linux-x64-v8.2.0.53.tgz. Use that in the cudNN install command instead.
You can tar for more cuDNN versions here
http://people.cs.uchicago.edu/~kauffman/nvidia/cudnn/

Thank you...
I am going to add your suggestion link.

This link to download cudnn did not work for me.

Did not work for me as well, did you manage to find another link?

For the download link (which is forbidden), I had to create an NVIDIA account (free), manually download and then tar -xzvf cudnn-11.2-linux-ppc64le-v8.1.1.33.tgz. After copying cuDNN library files into the CUDA toolkit, I would recommend rebooting since there might be a "driver mismatch" error for nvidia-smi. Following reboot, everything looks great! Thanks!