Is this page helpful?

AI Workbench Container Runtimes#

NVIDIA AI Workbench will install or use an existing container runtime in location to build and run the project container for your Workbench Projects.

When you install AI Workbench, the installer provides a choice between Docker (CE or Desktop, depending) and Podman.

Docker#

Docker is a popular, all-in-one platform for creating, managing and deploying containers. Depending on your company size and revenue, you might need a paid license to use Docker. Docker uses a client-server architecture, with the docker daemon running as a persistent background process.

Podman#

Podman is an open-source container runtime and management tool originally developed by Red Hat. Podman employs a daemon-less architecture, and launches containers as child processes that do not require superuser privileges. AI Workbench uses rootless Podman.

Change the Container Runtime for Your Local Workbench#

The Desktop App does not require a local container runtime, however, if you want to enable local development and compute then you must do the full local install, which does require you to select either Docker or Podman.

Once you have done the full local install, you may want to switch from Docker to Podman or vice versa. You can do this directly from the AI Workbench Settings Window.

  1. Leave the Location Manager window open but close all location windows before changing the container runtime.

  2. Open the AI Workbench Settings Window from the Gear at the top of the Locations Manager or from the tool tray on your laptop.

  3. Select Settings > Container Runtime and then select whichever runtime you want to use (Docker left, Podman right) and then select Switch Runtime #. If you still have a location window open, you will be prompted to close it and try again.

  4. Select Restart AI Workbench in the modal and wait for the application to restart.

  5. When the Locations Manager reopens, select the local location that will have an Update badge on it. #. The full local installer will begin, and you will arrive at the Container Selection Screen.

  6. Select the desired container runtime and proceed.