How to Install TensorFlow with Cuda and cuDNN support in Windows – 2024

Hey guys, In this blog we will see how we can install Tensorflow with CUDA and cuDNN. This process can be really tricky and frustrating if you are doing it for the very first time. So without any further due, let’s do it…

Step 1 – Decide versions for CUDA,cuDNN, and Visual Studio

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

  • We will download the following versions.
  • CUDA 11.2
  • cuDNN 8.1
  • Microsoft Visual Studio 2019

Step 2 – Download the CUDA Toolkit

Download Link – https://developer.nvidia.com/cuda-11.2.2-download-archive

NOTE – There is no Windows 11 version for CUDA 11.2 that’s why I used Windows 10 version and it worked like charm 🙂

Windows > x86_64 > 10 > exe(local) > Download

install tensorflow with cuda and cudnn

Step 3 – Download cuDNN

Download Link – https://developer.nvidia.com/rdp/cudnn-archive

  • If you have an account on Nvidia simply Login otherwise click on Join Now.
  • We will select the latest version of cuDNN (8.1.1) for CUDA 11.2 and we will download the zip for Windows.
install tensorflow with cuda and cudnn

Step 4 – Download Visual Studio 2019 Community.

Download Link – https://my.visualstudio.com/Downloads?PId=8228

  • Login to Microsoft and then search Visual Studio 2019 and download the Community version.
  • It will download a setup.
  • Install that setup.
  • It will ask to download workloads, just skip it and just install Visual Studio Core Editor.

Step 5 – Extracting and merging files

  • Once you have successfully downloaded CUDA and cuDNN, install the CUDA toolkit by double-clicking on it.
  • Agree and Continue > Express (Recommended). It can restart while the installation process.
  • Then extract the cuDNN zip. It will produce 4 files.
  • Copy all the files from the cuDNN folder and paste them into C:/Program Files/NVIDIA GPU Computing Toolkit/CUDA/v11.2 and replace the files in the destination.
  • Now open the bin Folder and copy the path from the address bar.
install tensorflow with cuda and cudnn
  • Now open the start menu and type env and you will see an option “Edit the System Environment Variables“. A window like the one below will pop up.
install tensorflow with cuda and cudnn
  • Click on Environment Variables here.
  • In the user variables (in the top section) double click on Path and click New.
  • Now paste the bin path you copied in a new field and hit OK.
  • Now do the same steps for libnvvp folder also.
  • Copy the path to libnvvp folder and add a new record in path.
install tensorflow with cuda and cudnn

RESTART YOUR COMPUTER.

Step 6 – Check the successful installation of CUDA

  • Run the nvidia-smi command in your terminal.
  • You can see in the top right corner, CUDA Version: 11.2

Step 7 – Create a conda environment and install TensorFlow

  • Now open your terminal and create a new conda environment.
  • Use the following command and hit “y“.
  • Here gpu is the name that I gave to my conda environment.
conda create -n gpu python=3.9
  • Activate the conda environment and install tensorflow-gpu.
conda activate gpu
pip install tensorflow-gpu
  • Now simply copy the code below and paste it into a file named test.py.
import tensorflow as tf

tf.compat.v1.disable_eager_execution()

with tf.device('/gpu:0'):
    a = tf.constant([1.0, 2.0, 3.0, 4.0, 5.0, 6.0], shape=[2, 3], name='a')
    b = tf.constant([1.0, 2.0, 3.0, 4.0, 5.0, 6.0], shape=[3, 2], name='b')
    c = tf.matmul(a, b)

with tf.compat.v1.Session() as sess:
    print (sess.run(c))
  • Now run the python file in the conda environment.
  • If you see these results, bingo!! we have successfully installed TensorFlow with CUDA and cuDNN support 🙂
install tensorflow with cuda and cudnn

So this is how you will install TensorFlow with Cuda and cuDNN in Windows, thanks for reading it and I hope you are taking something with you after reading this and till the next time ?…

Check out my other machine learning projectsdeep learning projectscomputer vision projectsNLP projectsFlask projects at machinelearningprojects.net

Abhishek Sharma
Abhishek Sharma

Started my Data Science journey in my 2nd year of college and since then continuously into it because of the magical powers of ML and continuously doing projects in almost every domain of AI like ML, DL, CV, NLP.

Articles: 517

5 Comments

  1. Hello,

    That was very helpfull ! 🙂

    I would like to know if you know how to install MirroredStrategy for multiple gpus configurations.

    I tried to follow Tensorflow’s directions but I always have problems with libraries or misplaced files. I end up with an error like :
    “`
    Registered devices: [CPU, GPU]
    Registered kernels:

    “`
    Thanks if you have an answer for me.

    Have a nice day!

  2. Hi,

    Thanks for this post. I think step 6 works differently: nvidia-smi shows the current driver and the CUDA version supported by this driver. I currently have Nvidia driver 527.56, which supports CUDA version 12.0, but I don’t have 12.0 installed.

  3. scratch that, i missed i have to double click on Path for that page to even show up. cheers, good guide 🙂

Leave a Reply

Your email address will not be published. Required fields are marked *