If you haven’t read it yet, you may want to review the Installation Overview.
Remote Ubuntu 22.04 Prerequisites
Access: SSH access as a non-root user with
sudo
privileges via a public/private key pair.The private key can’t be password-protected
Operating System: Ubuntu 22.04
16 GB of RAM
500 MB of disk space for the NVIDIA AI Workbench application
30 to 40 GB of disk space for containers
Before downloading, please read the NVIDIA AI Product Agreement, the NVIDIA AI Workbench Shared Security Model, and our Data Privacy Policy.
By downloading, installing, or using 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. Before downloading and using NVIDIA AI Workbench, please read our Shared Security Model.
Make sure your system satisfies the prerequisites above.
You need SSH access.
Through a local terminal, SSH to the remote.
You can use a password or private key for this step.
Run the following command. Do not run it as
root
.mkdir -p $HOME/.nvwb/bin && \ curl -L https://workbench.download.nvidia.com/stable/workbench-cli/$(curl -L -s https://workbench.download.nvidia.com/stable/workbench-cli/LATEST)/nvwb-cli-$(uname)-$(uname -m) --output $HOME/.nvwb/bin/nvwb-cli && \ chmod +x $HOME/.nvwb/bin/nvwb-cli && \ sudo -E $HOME/.nvwb/bin/nvwb-cli install
This will launch a Text-based User Interface (TUI) that will guide you through the remote installation.
You can advance through the installation by using the arrow keys to select different options and hitting
Enter
to select options.If the machine has an NVIDIA GPU but no drivers installed, AI Workbench will detect this and ask if you want it to install the drivers.
Selecting yes will install the drivers and require a reboot.
You will be prompted to choose Docker or Podman as a runtime.
Either selection will be installed by AI Workbench.
The selected container runtime and the other dependencies will be installed or updated, and you can follow the progress in the application window.
If you selected AI Workbench to install the drivers, they will be installed and you will need to reboot.
After that, the installation is finished
If drivers were installed, reboot the machine. Once it restarts you can connect it to your Local AI Workbench.
Otherwise, you can connect it to your Local AI Workbench right away.
Connecting a Remote Location to your Local AI Workbench is a one-time process. If not done so already, you must first generate a passwordless ssh key and pass the public key to the remote instance’s .ssh/authorized_keys
file. Then, add a new Remote Location via the AI Workbench application UI.
Open up your command terminal on the local system you are working in and execute the following depending on the operating system you are working in.
Replace the placeholder IP (
10.123.4.56
) with the location of the Launchpad instance you have been provided.Replace the user (
local-user
) with the username of your local system.Replace the user (
remote-user
) with the username of your remote system.
Windows Powershell
ssh-keygen -f "C:\Users\local-user\.ssh\id_rsa" -t rsa -N '""'
type$env:USERPROFILE\.ssh\id_rsa.pub | ssh remote-user@10.123.4.56 "cat >> .ssh/authorized_keys"
MacOS Terminal
ssh-keygen -f "/Users/local-user/.ssh/id_rsa" -t rsa -N ""
ssh-copy-id remote-user@10.123.4.56
Ubuntu 22.04 Local System
ssh-keygen -f "/home/local-user/.ssh/id_rsa" -t rsa -N ""
sh-copy-id remote-user@10.123.4.56
Next, open the NVIDIA AI Workbench Desktop Application on your local system. Click the Add Remote Location button.
Enter the following information of your Remote Location
Location Name
Description
Hostname or IP Address
SSH Port number, if not port 22
SSH Username to the Remote system
SSH Key File on the Local system (private key)
Location of the
.nvwb
directory on the Remote system, if not the default location.
Select Add Location. You should now be able to see the remote Launchpad instance show up as a location in your AI Workbench.