01_installation.md

Installation

https://www.tensorflow.org/install

0. Optional: GPU Support

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

1. Install Visual Studion 2017 Free Version and C++ Redistributable:

   • https://www.techspot.com/downloads/6278-visual-studio.html
   • https://support.microsoft.com/en-us/help/2977003/the-latest-supported-visual-c-downloads

2. Install CUDA Toolkit 10.1 https://developer.nvidia.com/cuda-toolkit-archive

3. Install NVIDIA cuDNN version 7 https://developer.nvidia.com/cudnn

Download (you need an account) and copy the dll, include, and lib files in the corresponding directories of the CUDA Toolkit installation directory

4. On Windows: Modify environment variables (paths must match your installation directory):

• Add variable

CUDA_PATH = C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.1\bin

Add 2 entries to PATH variable:

C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.1\bin
C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.1\libnvvp

1. Installation

1. Create a virtual environment and activate it (e.g. with conda or virtualenv)

conda create -n tf python=3.8
conda activate tf

2. Install with

pip install tensorflow

2. Verification:

import tensorflow as tf
print(tf.__version__)

# test gpu
physical_devices = tf.config.list_physical_devices("GPU")
print(len(physical_devices))