Skip to content

Instantly share code, notes, and snippets.

@patel-zeel

patel-zeel/installing_nvidia_driver_cuda_cudnn_linux.md

Forked from kmhofmann/installing_nvidia_driver_cuda_cudnn_linux.md
Last active December 17, 2021 18:34
Show Gist options
  • Save patel-zeel/2e515e35407cc987691df2cdcb8bd9e1 to your computer and use it in GitHub Desktop.
Save patel-zeel/2e515e35407cc987691df2cdcb8bd9e1 to your computer and use it in GitHub Desktop.
Download ZIP
Installing the NVIDIA driver, CUDA and cuDNN on Linux
Raw
installing_nvidia_driver_cuda_cudnn_linux.md

https://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html

For tensorflow to work

  • Remove any preinstalled drivers, or nvidia stuff with this
  • Use GCP provided script to install the latest cuda and drivers.
  • Reboot the system and verify nvidia-smi and nvcc
  • Download tar file for cudnn from here
  • Extract and copy library files from /lib folder to $LD_LIBRARY_PATH folder (maybe /usr/local/cuda-11/lib64/)
  • pip install tensorflow should work with GPU support now.

Install seperate versions of cuda, cudnn manually: https://medium.com/analytics-vidhya/install-cuda-11-2-cudnn-8-1-0-and-python-3-9-on-rtx3090-for-deep-learning-fcf96c95f7a1

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment