The Full Local Install#

Downloading or installing AI Workbench implies consent to the NVIDIA AI Product Agreement.

By downloading or installing the NVIDIA AI Workbench software, you agree to the terms of the NVIDIA AI Product Agreement (EULA). If you do not agree to the terms of the EULA, you are not authorized to download, install, or use NVIDIA AI Workbench.

Overview#

The full local install is required to develop locally in AI Workbench.

This is optional but is required to use AI Workbench to manage Git repositories that are on your laptop.

The full local install installs or updates the required system dependencies.

For example, AI Workbench will check if Git, Git-LFS and a container runtime are installed, and if not install them.

If a given dependency is installed, AI Workbench may update it to get it to a minimum required version.

Full local install requires the following on all three operating systems: Windows, MacOS and Ubuntu.

The Desktop App: Full local install is initiated from within the Desktop App.
Admin privileges: Workbench installs into your system.
Internet connection: You will be downloading application binaries.
40 GB of free disk space: This is for deep learning framework containers - they are big.

The 40 GB of free disk space is to prevent you running out of space due to large containers.

The AI Workbench application itself is no more than 1 GB in size. However, deep learning containers can be quite large.

For example, the Pytorch container from NVIDIA is over 20 GBs in size.

If you don’t have that much space free, then just continue to work in remote only mode.

Key Concepts#

Container Runtime: The container engine software (Docker or Podman) required to manage containers locally. AI Workbench lets you select the runtime to use and will install it if not already installed.
Windows Subsystem for Linux (WSL2 on Windows): Linux virtualization provided by Windows. Required for the full local install on Windows machines.
NVIDIA-Workbench Distribution (WSL2 on Windows): The Ubuntu distribution that AI Workbench downloads and modifies during the Windows full local install. It is where the application binaries and application folder are installed.
CLI: The command line interface for AI Workbench. Gets installed during the full local install.

Full Local Install on Windows#

Prerequisites: AI Workbench will verify the following prior to install.

Windows User Account Controls configured: You will authorize AI Workbench through UAC modals.
BIOS Virtualization enabled: WSL2 requires virtualization enabled in the firmware.

Step One: Select the Full Local Install in the Locations Manager.

Click Locations Manager > Full Local Install
Click Add Local Modal > Next
Click Windows UAC Modal > Yes

Success: A new window opens showing the System Requirements Page.

Step Two: Resolve any required system configurations.

Systems Requirements may ask you to configure some things.
Follow the guidance to fix any issues. Find details in the reference here.
When you’ve configured the issue properly, click Retry Installation.

Success: Page shows Ready for full local installation

Click Next.

Step Three: Select your desired container runtime.

Select Docker Desktop or Podman > Next.

AI Workbench works identically with whichever you select.

Step Four: Wait for the installation to complete.

AI Workbench will install or update the selected container runtime, as well as any

Success: The locations manager opens showing the “local” location card.

If your system has an NVIDIA GPU, update the drivers before the full local install.

Full local will work with your drivers, but older drivers may break newer CUDA containers. Update and configure the drivers with the NVIDIA App.

Full Local Install on MacOS#

Step One: Select the Full Local Install in the Locations Manager.

Click Locations Manager > Full Local Install
Click Add Local Modal > Next

Success: A new window opens showing the System Requirements Page.

Step Two: Resolve any required system configurations.

Systems Requirements may ask you to configure some things.
Follow the guidance to fix any issues.
Click Retry Installation.

Success: Page shows Ready for full local installation

Click Next.

Step Three: Select your desired container runtime.

Select Docker Desktop or Podman > Next.

AI Workbench works identically with whichever you select.

Step Four: Wait for the installation to complete.

AI Workbench will install or update the selected container runtime, as well as any

Success: The locations manager opens showing the “local” location card.

Full Local Install on Ubuntu#

Step One: Select the Full Local Install in the Locations Manager.

Click Locations Manager > Full Local Install
Click Add Local Modal > Next

Success: A new window opens showing the System Requirements Page.

Step Two: Resolve any required system configurations.

Systems Requirements may ask you to configure some things.
Follow the guidance to fix any issues.
Click Retry Installation.

Success: Page shows Ready for full local installation

Click Next.

Step Three: Select your desired container runtime.

Select Docker Desktop or Podman > Next.

AI Workbench works identically with whichever you select.

Step Four: Wait for the installation to complete.

AI Workbench will install or update the selected container runtime, as well as any

Success: The locations manager opens showing the “local” location card.

If your system has an NVIDIA GPU but the drivers aren’t installed. AI Workbench can install them for you.

You will be prompted to allow AI Workbench to install them for you.