ikrambenabdelouahab / manual_tf_gpu

Manual_tf_GPU

This is a step by step guide to install latest version of TensorFlow on GPU

07/01/2022

Tested successfully on Windows 10 & NVIDIA Quadro T1000 with Max-Q Design

Why manual installation ?

After many tries to get things done using "conda install tensorflow-gpu", It failed and gives a ton of errors. So I propose doing things manually to save time. And to make sure everything is well installed. This is a manual installation guide of TensorFlow to avoid any kind of errors.

Step 1: Hardware check

• Using DxDialog check GPU series and version

Step 2: Install driver (I already have it)

• https://www.nvidia.com/download/index.aspx?lang=en-us

Step 3: Install Visual Studio IDE

• https://visualstudio.microsoft.com/fr/vs/

Step 4: Install CUDA Toolkit

• Version 11.5
• https://developer.nvidia.com/cuda-downloads

Step 5: Install cuDNN

• Create account or log in
• Make sure to choose the version suitable with CUDA 11.5
• V 8.3.1
• https://developer.nvidia.com/cudnn
• Download the zip folder. Decompress it in C:\tools\

Step 6: Add environment variables

• https://www.tensorflow.org/install/gpu

1.	C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.5\bin
2.	C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.5\libnvvp
3.	C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.5\include
4.	C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.0\extras\CUPTI\lib64
5.	C:\tools\cuda-windows-x86_64-8.3.1.22_cuda11.5-archive\bin
6.	C:\tools\cuda-windows-x86_64-8.3.1.22_cuda11.5-archive\include

(*) check paths depending on your OS and used versions !

Step 7: Install Miniconda3

• https://docs.conda.io/en/latest/miniconda.html
• Latest Miniconda Installer Links => Miniconda3 Windows 64-bit
• Conda version 4.11.0
• Python 3.9.5
• Add Miniconda3 to my path environment variable

Step 8: Install Jupyter

• Go to Anaconda Prompt (base)

# conda install -y jupyter

• Test by launching a notebook

# jupyter notebook

Step 9: Create a new environment

• Create a new environment called: tensorflow

# conda create -y --name tensorflow python=3.9

• Activate the new environment

# conda activate tensorflow

Step 10: Jupyter Kernel

• Use Anaconda Prompt
• Navigate to ‘tensorflow’ environment

# conda install ipykernel
# python -m ipykernel install --user --name tensorflow --display-name "Python 3.9 (tensorflow)"

• Check the environment in Jupyter notebook

Step 11: Install TensorFlow

# pip install tensorflow

• Installed latest version at this day: 2.7
• Check installation

> import tensorflow as tf
> print(tf.__version__)
> print(len(tf.config.list_physical_devices('GPU'))>0) # should return True

Step 12: Install additional packages for our project

(tensorflow) # pip install pandas
(tensorflow) # pip install pillow
(tensorflow) # pip install tensorflow-addons
(tensorflow) # pip install pydot
(tensorflow) # pip install matplotlib

References

• https://github.com/jeffheaton/t81_558_deep_learning/blob/master/install/manual_setup2.ipynb
• https://github.com/jeffheaton/t81_558_deep_learning/blob/master/install/manual_setup.ipynb
• https://www.youtube.com/watch?v=OEFKlRSd8Ic