Member-only story
GPU Support for Tensorflow
Step-by-step installation of CUDA toolkit in Windows
Introduction
In this article I’m going to show you how to install Tensorflow with GPU support. I will walk you through the procedure of installing Tensorflow in windows environment and get GPU support from Nvidia GeForce RTX graphics card. The steps are below.
1. Install Visual Studio (some libraries from CUDA will not work without Visual Studio)
2. Install CUDA toolkit
3. Install CuDNN
4. Install Tensorflow
Let’s Dive In.
Initial Checks
To get the things started we first need to visit the website https://www.tensorflow.org/install/gpu. Here in the hardware requirements section you will see the devices which are currently supported for deep learning. It states that Nvidia GPU card with CUDA architecture from 3.5 to 8.0 and higher then 8.0 are supported. In the software requirements section, it will show you the required software that we need to download and install. The first one is NVIDIA GPU driver; the second one is the CUDA toolkit. CUDA stands for Compute Unified Device Architecture. This is a parallel computing platform developed by Nvidia. The next one is CUPTI which comes with CUDA. CuDNN stands for…
