Installing TensorFlow with GPU, CUDA and cuDNN support for windows 10

If you want to take advantage of your GPU for faster training and inference with TensorFlow, you’ll need to install the necessary libraries for CUDA and cuDNN support. Here’s how you can do it on a Windows 10 machine:

1. Download and install the latest version of CUDA Toolkit from NVIDIA’s website. Make sure to select the appropriate version for your GPU and operating system.
2. Download cuDNN from NVIDIA’s website. You’ll need to sign up for a free account in order to download cuDNN. Make sure to select the version that matches the CUDA version you installed in the previous step.
3. Extract the cuDNN archive and copy the contents into the CUDA Toolkit installation directory. This will typically be in the ‘C:Program FilesNVIDIA GPU Computing ToolkitCUDA’ folder.
4. Verify that CUDA and cuDNN are properly installed by checking the environment variables. You should see paths like ‘C:Program FilesNVIDIA GPU Computing ToolkitCUDAv10.0bin’ and ‘C:Program FilesNVIDIA GPU Computing ToolkitCUDAv10.0libx64’ in the PATH variable.
5. Install TensorFlow with GPU support by running the following pip install command in your command prompt:

pip install tensorflow-gpu

6. Test your TensorFlow installation by running a sample script that uses your GPU for computation. You should see a significant speedup compared to using the CPU for the same task.

By following these steps, you can make sure that TensorFlow is properly configured to take advantage of your GPU for faster training and inference. Happy coding!