Docker EE GPU for Apt Installation Recipe

Preparation

Prepare your host by installing NVIDIA drivers, Docker, and NVIDIA runtime.

Install NVIDIA Drivers

To install NVIDIA drivers, open a terminal window on the host. Run apt update and apt upgrade to ensure that you are using the latest operating system software.

sudo apt update
sudo apt upgrade

Use apt-get to install required utilities.

sudo apt-get install \
    apt-transport-https \
    ca-certificates \
    curl \
    gnupg-agent \
    software-properties-common

Reboot your system to activate all of your changes.

sudo reboot

Install CUDA

Verify that the gcc compiler is installed with the following command.

gcc --version

If no version information returns, run the following command.

sudo apt install gcc

To install the CUDA package:

1. Go to https://developer.nvidia.com/cuda-downloads.
2. Select the target platform by selecting the operating system (Linux), architecture (based on your environment), distribution (Ubuntu), version (based on your environment), and installer type (OmniSci recommends deb (network)).
3. Install the CUDA drivers per instructions on the NVIDIA website.
4. Reboot your system to ensure that all changes are active:

   sudo reboot
   

Install Docker

Remove any existing Docker installs and the legacy NVIDIA docker runtime.

sudo docker volume ls -q -f driver=nvidia-docker | xargs -r -I{} -n1 docker ps -q -a -f volume={} | xargs -r docker rm -f
sudo apt-get purge nvidia-docker

Remove Docker.

sudo apt-get remove docker docker-engine docker.io containerd runc

Update with apt-get.

sudo apt-get update

Use curl to download the latest Docker version.

sudo curl -fsSL https://download.docker.com/linux/ubuntu/gpg | sudo apt-key add -

Add Docker to your Apt repository.

sudo add-apt-repository \
   "deb [arch=amd64] https://download.docker.com/linux/ubuntu \
   $(lsb_release -cs) \
   stable"

Update your repository.

sudo apt-get update

Install Docker, the command line interface, and the container runtime.

sudo apt-get install docker-ce docker-ce-cli containerd.io

Optional: Run the following usermod command so that docker command execution does not require sudo privilege. Log out and log back in for the changes to take effect.

sudo usermod -aG docker $USER

Checkpoint

Verify your Docker installation.

sudo docker run hello-world

Install NVIDIA Docker Runtime

Use curl to add a gpg key:

curl -s -L https://nvidia.github.io/nvidia-container-runtime/gpgkey | \
  sudo apt-key add -
  distribution=$(. /etc/os-release;echo $ID$VERSION_ID)

Update your sources list:

curl -s -L https://nvidia.github.io/nvidia-container-runtime/$distribution/nvidia-container-runtime.list | \
  sudo tee /etc/apt/sources.list.d/nvidia-container-runtime.list

Update apt-get and install nvidia-container-runtime:

sudo apt-get update
sudo apt-get install -y nvidia-container-runtime

Edit /etc/docker/daemon.json to add the following, and save the changes:

{
 "default-runtime": "nvidia",
  "runtimes": {
     "nvidia": {
         "path": "/usr/bin/nvidia-container-runtime",
         "runtimeArgs": []
     }
 }
}

Restart the Docker daemon:

sudo pkill -SIGHUP dockerd

Checkpoint

Verify docker and NVIDIA runtime work together.

sudo docker run --runtime=nvidia --rm nvidia/cuda:10.0-runtime-ubuntu18.04 nvidia-smi

Standard NVIDIA-SMI output shows the GPUs in your instance.

Installation

Download OmniSci from DockerHub and Start OmniSci in Docker.

sudo docker run --runtime=nvidia \
  -d --runtime=nvidia \
  -v /var/lib/omnisci:/omnisci-storage \
  -p 6273-6280:6273-6280 \
  omnisci/omnisci-ee-cuda:v5.3.0

See also the note regarding the CUDA JIT Cache in Optimizing Performance.

Ensure that you have sufficient storage on the drive you choose for /var/lib/omnisci.

config file does not exist, ignoring --config /omnisci-storage/omnisci.conf

For more information on CUDA driver installation, see the CUDA Installation Guide.

For more information on Docker installation, see the Docker Installation Guide.