TensorFlow 2 with GPU on Windows: Step-by-step instructions

How to properly install CUDA and cuDNN on Windows to use TensorFlow with GPU support

Software requirements

• Python 3.9–3.11
• pip version 19.0 or higher for Linux (requires manylinux2014 support) and Windows. pip version 20.3 or higher for macOS.
• Windows Native Requires Microsoft Visual C++ Redistributable for Visual Studio 2015, 2017 and 2019

The following NVIDIA® software required for GPU support

• NVIDIA® GPU drivers version 450.80.02 or higher.
• CUDA® Toolkit 11.8.
• cuDNN SDK 8.6.0.

Prerequisites

• Download GeForce Experience to update drivers

• Download Microsoft Visual C++ drivers

• Download Visual Studio and Visual Studio Code

• Download CUDA Toolkit version - 11.8 or Lattest version (not recommended)

• Download cuDNN v8.9.7 for CUDA v11.8.0:

• Download NVIDIA Nsight Developer Tools:

  • Nsight Integration for Visual Studio
  • Nsight Compute
  • Nsight Graphics
  • Nsight Systems

Installation

• Cuda Toolkit

  Open the downloaded file cuda_11.8.0_522.06_windows.exe and follow the installation instructions.

  

  In the options select express installation

  

• CUDNN

  Unzip the archive cudnn-windows-x86_64-8.9.7.29_cuda11-archive.zip and move with replace all files in the lib, include and bin folders on C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.8 with files from the corresponding archive folders:

  

  Add the following lines to your system and user $Path variable:

      C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.8
      C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.8\bin
      C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.8\include
      C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.8\lib
      C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.8\lib\x64

  

• NVIDIA Nsight Developer Tools

  Install all the NVIDIA Nsight developer tools that we downloaded earlier

  

Verifying installation success

• Run the following commands in a terminal

  nvcc --version # Shows CUDA version

  

  nvidia-smi # Shows the NVIDIA system management interface

  

  Caution: TensorFlow 2.10 was the last TensorFlow release that supported GPU on native-Windows. Starting with TensorFlow 2.11, you will need to install TensorFlow in WSL2, or install tensorflow or tensorflow-cpu and, optionally, try the TensorFlow-DirectML-Plugin`

  # Upgrade pip to latest version
  python -m pip install --upgrade pip
  # Anything above 2.10 is not supported on the GPU on Windows Native
  python -m pip install "tensorflow<2.11"
  # Verify the installation:
  python -c "import tensorflow as tf; print(tf.config.list_physical_devices('GPU'))"
  # If a list of GPU devices is returned, you've installed TensorFlow successfully.

  

  # If a tensor is returned, you've installed TensorFlow successfully.
  python -c "import tensorflow as tf; print(tf.reduce_sum(tf.random.normal([1000, 1000])))"

  

  🎉 Thank you for your attention! 🎉

  🔝 Back to top 🔝