GPU ML Framework Installation: Ubuntu22.04 + Nvidia Driver 535 + CUDA 11.8 + cuDNN v8.7 + TensorFlow 2.13.0 + PyTorch 2.1.0

#TF_Torch_GPU_Installation

Install TensorFlow GPU and PyTorch GPU on a NVIDIA Graphics Card available computer.

1_install_driver_cuda_cudnn.sh

#!/bin/bash

### stage 1 ####
# verify the system has a cuda-capable gpu
# download and install the nvidia cuda toolkit and cudnn
# setup environmental variables
###

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

### remove previous installation
sudo apt purge '.*nvidia.*' '.*cuda.*' '.*cudnn.*'
sudo apt remove '.*nvidia.*' '.*cuda.*' '.*cudnn.*'
sudo rm /etc/apt/sources.list.d/cuda*
sudo apt-get autoremove && sudo apt-get autoclean
sudo rm -rf /usr/local/cuda*

### do system upgrade
sudo apt update && sudo apt upgrade -y
sudo apt install -y g++ freeglut3-dev build-essential libx11-dev libxmu-dev libxi-dev libglu1-mesa libglu1-mesa-dev

# install nvidia driver
sudo add-apt-repository ppa:graphics-drivers/ppa
sudo apt update
sudo apt install -y libnvidia-common-535 libnvidia-gl-535 nvidia-driver-535

# install cuda deb(network)
# Reference: https://developer.nvidia.com/cuda-11-8-0-download-archive?target_os=Linux&target_arch=x86_64&Distribution=Ubuntu&target_version=22.04&target_type=deb_network
wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2204/x86_64/cuda-keyring_1.0-1_all.deb
sudo dpkg -i cuda-keyring_1.0-1_all.deb
sudo apt-get update
sudo apt install -y cuda-11-8

# Note: you need to add below lines to ~/.bashrc or ~/.zshrc
echo 'export PATH=/usr/local/cuda/bin:$PATH' >> ~/."$(basename $SHELL)"rc
echo 'export LD_LIBRARY_PATH=/usr/local/cuda/lib64:$LD_LIBRARY_PATH' >> ~/."$(basename $SHELL)"rc

# install cuDNN v8.7
# Reference: https://developer.nvidia.com/cudnn
CUDNN_TAR_FILE="cudnn-linux-x86_64-8.7.0.84_cuda11-archive.tar.xz"
sudo wget https://developer.download.nvidia.com/compute/redist/cudnn/v8.7.0/local_installers/11.8/cudnn-linux-x86_64-8.7.0.84_cuda11-archive.tar.xz
sudo tar -xvf cudnn-linux-*.tar.xz
sudo mv cudnn-linux-x86_64-8.7.0.84_cuda11-archive cuda

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

# reboot to solve "Failed to initialize NVML: Driver/library version mismatch"
sudo reboot

2_install_tf_torch.sh

#!/bin/bash

### stage 2 ####
# verify the nvidia driver + cuda + cudnn installation
# install TensorFlow and PyTorch
###

# verify the installation
nvidia-smi
nvcc -V

# install TensorFlow GPU
python3 -m pip install nvidia-cudnn-cu11==8.6.0.163 tensorflow==2.13.0

# install PyTorch GPU
python3 -m pip install torch==2.1.0 torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118

3_gpu_check.sh

#!/bin/bash

### stage 3 ####
# check if framework is actually using GPU
###


python3 -c "import tensorflow as tf; print('TensorFlow: Version: ' + tf.__version__ + ', GPU Available: ', bool(len(tf.config.list_physical_devices('GPU'))))"

python3 -c "import torch; print('PyTorch: Version: ' + torch.__version__ + ', GPU Available: ', torch.cuda.is_available())"

4_additional_packages.md

Additional Packages

TF-Agents

Since tensorflow 2.13.0 is installed, if you need to install tf-agents, please use version 0.17.0.

See TF-Agents Releases

Installation:

pip install tensorflow==2.13.0 tensorflow_probability~=0.20.1 tf-agents==0.17.0

Notes.md

Tips

TensorFlow GPU: Set Memory Growth

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

Issue Fixes

TensorFlow NUMA Info

I ... successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355

Fix:

for a in /sys/bus/pci/devices/*; do echo 0 | sudo tee -a $a/numa_node; done