Installing TensorFlow-GPU 2.4.1 on WSL2
Hi Everyone! Thank you for watching this video on installing TensorFlow 2.4.1 on WSL 2 on Windows 10 Home.
In this tutorial, you’ll learn how to install TensorFlow in a Linux distribution (in our case Ubuntu 18.04) and allow it to connect to your local GPU.
I won’t be covering deployment of TensorFlow on Docker.
▬▬▬▬▬▬ T I M E S T A M P S ⏰ ▬▬▬▬▬▬
00;00;00 Introduction to Installing TensorFlow 2.4.1 on WSL2
00;02;24 TensorFlow Installation Prerequisites
00;03;28 Steps to Install TensorFlow 2.4.1 on WSL2
00;05;35 Install CUDA Toolkit 11.0 for Windows
00;07;14 Configure WSL 2
00;08;28 Install Ubuntu 18.04 on WSL 2
00;09;20 Install CUDA Toolkit 11.0 for Linux on Ubuntu 18.04 in WSL 2
00;10;17 Patch CUDA 11.0 Libraries for Linux
00;10;40 Installing Anaconda on WSL 2
00;11;54 Installing TensorFlow 2.4.1 on WSL2
00;12;14 Configuring Jupyter Lab on WSL 2
00;12;59 Test TensorFlow-GPU 2.4.1
00;13;25 TensorFlow 2.4.1 Installation of WSL2 Recap
▬▬▬▬▬▬ Library List and Commands 🔗 ▬▬▬▬▬▬
List of CUDA Libraries Needing Installed / Updated:
libnvinfer7_7.1.3-1+cuda11.0_amd64
libcudnn8_8.0.4.30-1+cuda11.0_amd64
libcudnn8-dev_8.0.4.30-1+cuda11.0_amd64
libnvinfer-dev_7.1.3-1+cuda11.0_amd64
libnvinfer-plugin7_7.1.3-1+cuda11.0_amd64
Command Reference:
Update Ubuntu 18.04:
sudo apt update && sudo apt upgrade
Install NVIDIA CUDA Toolkit 11.0:
sudo apt-key adv --fetch-keys http://developer.download.nvidia.com/...
Ugh... Because YouTube does not allow "Angled Brackets" aka greater than signs, in their descriptions, place a redirect (greater than sign) after 86_64 /" and before /etc
sudo sh -c 'echo "deb http://developer.download.nvidia.com/... /" (redirect) /etc/apt/sources.list.d/cuda.list'
sudo apt-get update
sudo apt-get install -y cuda-toolkit-11-0
Test NVIDIA CUDA Toolkit:
cd /usr/local/cuda/samples/4_Finance/BlackScholes
sudo make BlackScholes
../../bin/x86_64/linux/release/BlackScholes
Library Updates / Install:
wget https://developer.download.nvidia.com...
sudo apt install ./libnvinfer7_7.1.3-1+cuda11.0_amd64.deb
wget https://developer.download.nvidia.com...
sudo apt install ./libcudnn8_8.0.4.30-1+cuda11.0_amd64.deb
wget https://developer.download.nvidia.com...
sudo apt install ./libcudnn8-dev_8.0.4.30-1+cuda11.0_amd64.deb
wget https://developer.download.nvidia.com...
sudo apt install ./libnvinfer-dev_7.1.3-1+cuda11.0_amd64.deb
wget https://developer.download.nvidia.com...
sudo apt install ./libnvinfer-plugin7_7.1.3-1+cuda11.0_amd64.deb
Install Development Environment:
sudo apt install python3-pip
pip3 –version
wget https://repo.continuum.io/archive/Ana...
bash Anaconda[YOUR VERSION].sh
*Open a new terminal window*
conda update conda
conda update anaconda
conda update python
conda update --all
pip3 install tensorflow-gpu
▬▬▬▬▬▬ Useful Links 🔗 ▬▬▬▬▬▬
Windows Insider Program:
https://insider.windows.com/en-us/
NVIDIA Developer Program: https://developer.nvidia.com/develope...
CUDA on Windows Subsystem for Linux (WSL) - Public Preview: https://developer.nvidia.com/cuda/wsl
Enable NVIDIA CUDA in WSL 2: https://docs.microsoft.com/en-us/wind...
NVIDIA CUDA Library Location: http://developer.download.nvidia.com/...
▬▬▬▬▬▬ Connect with me 👋 ▬▬▬▬▬▬
[email protected]