Installing Tensorflow GPU on Ubuntu 20.04, This Is the Way

This guide is the best way to install Tensorflow GPU when other methods are not successful on my Ubuntu 20.04 machine.

1. Removing the old package.

sudo apt-get clean
sudo apt-get update
sudo apt-get purge cuda
sudo apt-get purge nvidia-*
sudo apt-get autoremove

2. Install Nvidia package repositories

wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2004/x86_64/cuda-ubuntu2004.pin
sudo mv cuda-ubuntu2004.pin /etc/apt/preferences.d/cuda-repository-pin-600
sudo apt-key adv --fetch-keys https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2004/x86_64/7fa2af80.pub
sudo add-apt-repository "deb https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2004/x86_64/ /"
sudo apt-get update

3. Install Nvidia cudnn package repositories

wget http://developer.download.nvidia.com/compute/machine-learning/repos/ubuntu2004/x86_64/nvidia-machine-learning-repo-ubuntu2004_1.0.0-1_amd64.deb

sudo apt install ./nvidia-machine-learning-repo-ubuntu2004_1.0.0-1_amd64.deb
sudo apt-get update

4. Install Nvidia cuda and cudnn

sudo apt-get install cuda-11-2 nvidia-cuda-toolkit libcudnn8 libcudnn8-dev
sudo reboot # Reboot your machine

5. Install the Tensorflow on the system Python3.8 from Ubuntu

sudo apt install python3 python3-pip
sudo pip3 install tensorflow

6. Test in Python

python3 -c 'import tensorflow as tf
print("Num GPUs Available: ", len(tf.config.experimental.list_physical_devices("GPU")))'

The result should be like this:

If found error “Failed to call ThenRnnBackward with model config” use this code to allow memory growth.

gpu_devices = tf.config.experimental.list_physical_devices('GPU')
for device in gpu_devices:
    tf.config.experimental.set_memory_growth(device, True)