Skip to main content

Launching a GPU Instance

We run our GPUs on virtual machines using customized Docker images. Launching a container is straightforward and can be done in several ways.

Through the Console​

  1. Log in to the Snowcell Console: Access the Snowcell Console through your web browser and log in with your credentials.

  2. Navigate to the Instances Page: Go to the 'Instances' section in the console menu.

  3. Create a New Instance:

    • Click on the 'Create Instance' button.
    • Select the desired GPU type from the available options.
    • Configure the instance settings, such as memory, storage, and network configurations.
    • Choose a customized AMI or provide your own Docker image. We have AMIs preinstalled with the most recent ML-libraries.

  4. Launch the Instance: Review your settings and click 'Launch'. Your GPU instance will be up and running within minutes.

Through the CLI​

  1. Install Snowcell CLI:

    • Ensure you have the Snowcell CLI installed on your local machine. If not, download and install it through either:
    • pip (PyPI): 
      pip install snc
    • Homebrew (macOS): 
      brew install snc
    • Binary Release: Download the latest release from the Snowcell GitHub repository and follow the installation instructions provided.
    • Linux (Apt-based systems): 
      wget -qO - https://your-snowcell-cli-repo-url/snowcell-cli.gpg.key | sudo apt-key add -
      echo "deb [arch=amd64] https://your-snowcell-cli-repo-url/ stable main" | sudo tee /etc/apt/sources.list.d/snowcell-cli.list
      sudo apt-get update
      sudo apt-get install snowcell-cli

  2. Authenticate the CLI:

    • Use the following command to authenticate the CLI with your Snowcell account: 
      snc login
    • Enter your credentials when prompted.

  3. List Available GPU Types:

    • To see the available GPU types, use the following command: 
      snc describe-instances

  4. Create a New Instance:

    • Use the following command to create a new GPU instance: 
      snc create-instance --type <GPU_TYPE> --memory <MEMORY_SIZE> --storage <STORAGE_SIZE> --network <NETWORK_CONFIG> --image <DOCKER_IMAGE>
    • Replace <GPU_TYPE>, <MEMORY_SIZE>, <STORAGE_SIZE>, <NETWORK_CONFIG>, and <DOCKER_IMAGE> with your desired configurations.

    Example:

    snc create-instance --type nvidia-v100 --memory 16GB --storage 500GB --network default --image my-custom-image:latest

  5. Launch the Instance:

    • Once configured, it will launch automatically. If for some reason the instance is not running, start it manually: 
      snc restart-instance --id <INSTANCE_ID>
    • Replace <INSTANCE_ID> with the ID of the instance you created.

Next Steps​

After launching your GPU instance, you can connect to it using SSH, deploy your workloads, and start utilizing the computational power for your tasks.