Reference

https://docs.nvidia.com/datacenter/tesla/tesla-installation-notes/index.html#ubuntu-lts

https://levelup.gitconnected.com/install-the-cuda-driver-and-toolkit-on-linux-3494a4436d6

https://hiseon.me/linux/ubuntu/install_nvidia_driver/

https://askubuntu.com/questions/1292120/is-there-a-ppa-for-having-nvidia-cuda-toolkit-with-the-latest-cuda-11-1-for-ub

https://forums.developer.nvidia.com/t/installed-cuda-10-0-on-ubuntu-18-04-with-deb-file-nvidia-smi-couldnt-communicate-with-nvidia-driver/80788/2

---

Manual

1. Remove any NVIDIA traces that you may have on your system.

   sudo rm /etc/apt/sources.list.d/cuda*
   sudo apt remove --autoremove nvidia-cuda-toolkit
   sudo apt remove --autoremove nvidia-*
   sudo apt-get --purge remove "*cublus*" "*cuda*"
   sudo apt-get --purge remove "*nvidia*"
   

2. Install driver

   • 1.1 : auto install

     sudo ubuntu-drivers autoinstall
     

     1.2 : manual install

     sudo apt install sudo gnupg
     sudo apt update
     sudo add-apt-repository ppa:graphics-drivers
     sudo apt update
     sudo apt-get install nvidia-driver-460
     sudo apt-get install dkms nvidia-modprobe
     sudo reboot
     nvidia-smi
     

   @ Driver install 까지만 하고, nvidia-docker, conda 등 가상환경 방법들을 사용하는게 좋습니다

3. Get NVIDIA public key

   release="ubuntu"$(lsb_release -sr | sed -e "s/\.//g")
   echo $release
   sudo apt-key adv --fetch-keys http://developer.download.nvidia.com/compute/cuda/repos/"$release"/x86_64/7fa2af80.pub
   

4. Add NVIDIA to the Source List Directory

   release="ubuntu"$(lsb_release -sr | sed -e "s/\.//g")
   echo $release
   sudo sh -c 'echo "deb http://developer.download.nvidia.com/compute/cuda/repos/"$release"/x86_64 /" > /etc/apt/sources.list.d/cuda.list'
   

5. Install NVIDIA CUDA Toolkit

   sudo apt-get update
   sudo apt-get install cuda-toolkit-11-0 cuda-toolkit-10-2
   # 위 방법을 써야지만 Driver가 dependency로 변경되지 않음
   (X) sudo apt-get install cuda-11-0 -> Driver를 변경하게 만듬
   

6. Add NVIDIA to the Source List Directory

   release="ubuntu"$(lsb_release -sr | sed -e "s/\.//g")
   echo $release
   sudo sh -c 'echo "deb http://developer.download.nvidia.com/compute/machine-learning/repos/"$release"/x86_64 /" > /etc/apt/sources.list.d/nvidia-machine-learning.list'
   

7. Install CUDA and CUDNN

   sudo apt-get update
   sudo apt-get install --no-install-recommends cuda-11-0 libcudnn8=8.0.5.39-1+cuda11.0 libcudnn8-dev=8.0.5.39-1+cuda11.0
   

---

Deprecated

1. Open the terminal and remove any NVIDIA traces that you may have on your system.

   sudo rm /etc/apt/sources.list.d/cuda*
   sudo apt remove --autoremove nvidia-cuda-toolkit
   sudo apt remove --autoremove nvidia-*
   sudo apt-get --purge remove "*cublus*" "*cuda*"
   sudo apt-get --purge remove "*nvidia*"
   

2. Setup the correct CUDA PPA on your system.

   sudo apt update
   
   release="ubuntu"$(lsb_release -sr | sed -e "s/\.//g")
   echo $release
   
   sudo apt install sudo gnupg
   sudo apt update
   sudo add-apt-repository ppa:graphics-drivers
   sudo apt-key adv --fetch-keys "http://developer.download.nvidia.com/compute/cuda/repos/"$release"/x86_64/7fa2af80.pub"
   sudo sh -c 'echo "deb http://developer.download.nvidia.com/compute/cuda/repos/'$release'/x86_64 /" > /etc/apt/sources.list.d/nvidia-cuda.list'
   sudo sh -c 'echo "deb http://developer.download.nvidia.com/compute/machine-learning/repos/'$release'/x86_64 /" > /etc/apt/sources.list.d/nvidia-machine-learning.list'
   sudo apt update
   

3. Install nvidia-driver

   sudo apt update
   sudo apt-get install nvidia-driver-460
   sudo apt-get install dkms nvidia-modprobe
   sudo reboot
   nvidia-smi
   

4. Install CUDA packages.

   sudo apt update
   (X) sudo apt-get install cuda-11-0 -> Driver를 변경하게 만듬
   # 아래 방법을 써야지만 Driver가 dependency로 변경되지 않음
   sudo apt install cuda-toolkit-11-0
   sudo reboot
   cat /usr/local/cuda/version.txt
   

5. Install CUDNN pacakges

   release="ubuntu"$(lsb_release -sr | sed -e "s/\.//g")
   echo $release
   
   wget https://developer.download.nvidia.com/compute/cuda/repos/$release/x86_64/cuda-$release.pin
   sudo mv cuda-$release.pin /etc/apt/preferences.d/cuda-repository-pin-600
   sudo apt-key adv --fetch-keys https://developer.download.nvidia.com/compute/cuda/repos/$release/x86_64/7fa2af80.pub
   sudo add-apt-repository "deb https://developer.download.nvidia.com/compute/cuda/repos/$release/x86_64/ /"
   sudo apt-get update
   
   sudo apt update
   sudo apt-get install libcudnn8=${cudnn_version}-1+${cuda_version}
   sudo apt-get install libcudnn8-dev=${cudnn_version}-1+${cuda_version}
   
   # ${cudnn_version} is 8.1.1.*
   # ${cuda_version} is cuda10.2 or cuda11.2