Quick Start

This is the shortest path to a local single-node AUP Learning Cloud deployment using auplc-installer, the Python-based installer for this repository.

The installer can run as a regular user. When a step needs elevated privileges (for example K3s installation), it prompts for your password once and keeps the sudo credential cache fresh—you do not need to prefix every command with sudo. Running with sudo ./auplc-installer ... is still supported and skips the password prompt.

Prerequisites

• Ubuntu 24.04

• sudo access

• Supported Ryzen AI 300 series and above APUs or Radeon 9000 series PCIe GPUs (ROCm)

• Docker available for the default install path

Install basic host dependencies:

sudo apt install build-essential

If Docker is not installed yet:

curl -fsSL https://get.docker.com | sh
sudo usermod -aG docker $USER
newgrp docker
docker --version

TUI dependencies (required for interactive install)

The recommended ./auplc-installer wizard requires questionary and prompt_toolkit.

System Python users (apt):

sudo apt install python3-questionary python3-prompt-toolkit

Conda or virtualenv users: install with pip inside your active environment instead of the apt packages:

pip install questionary prompt_toolkit

Non-interactive ./auplc-installer install does not need these packages.

Recommended: Interactive Install

Clone the repository and launch the wizard from the repo root:

git clone https://github.com/AMDResearch/aup-learning-cloud.git
cd aup-learning-cloud
./auplc-installer

In a real terminal, invoking the installer with no subcommand also opens the TUI automatically (same as tui).

For a first-time install, pick Install and press Enter at each prompt to accept the defaults. Before installation starts, you should see a summary like this:

Configuration summary
  GPU              : auto-detect
  K3s runtime      : Docker
  Image source     : pull
  Image registry   : ghcr.io/amdresearch
  Image tag        : latest
  Registry mirror  : (none)
  PyPI mirror      : (default)
  npm mirror       : (default)
  Courses          : cpu, gpu, Course-CV, Course-DL, Course-LLM, Course-PhySim

The installer requests your sudo password once, then runs eight stages: GPU detection, initial values overlay, helm + k9s install, K3s install, image pull (or build), ROCm device plugin + node labeller, values overlay refresh, and JupyterHub runtime deploy. When it finishes, open http://localhost:30890.

For a prompt-by-prompt explanation of every TUI choice, see Interactive TUI reference in the single-node guide.

Command-Line Install

Non-interactive install with the same defaults as the TUI (pull, latest tag, Docker runtime, all courses):

git clone https://github.com/AMDResearch/aup-learning-cloud.git
cd aup-learning-cloud
./auplc-installer install

Preview the plan without installing:

./auplc-installer install --dry-run

To override GPU detection, pick your GPU below—the install command updates to match your selection:

AMD device family

Radeon PRO

Radeon

Ryzen AI

Radeon PRO GPU

AI PRO R9700

AI PRO R9600D

Radeon GPU

RX 9070 XT

RX 9070 GRE

RX 9070

RX 9060 XT LP

RX 9060 XT

RX 9060

Ryzen AI APU

Max+ PRO 395

Max PRO 390

Max PRO 385

Max PRO 380

Max+ 395

Max 390

Max 385

9 HX 375

9 HX 370

9 365

git clone https://github.com/AMDResearch/aup-learning-cloud.git
cd aup-learning-cloud

./auplc-installer install --gpu=rdna4

After installation completes, open http://localhost:30890.

The checked-in default values in this repository use:

• custom.authMode: auto-login

• proxy.service.type: NodePort

• proxy.service.nodePorts.http: 30890

• local-path storage

So the default local experience is a simple HTTP NodePort deployment.

Common Variants

# Default non-interactive path (pull from registry, latest tag)
./auplc-installer install

# Preview configuration without sudo or system changes
./auplc-installer install --dry-run

# Use a different image tag prefix (GPU suffix appended automatically)
./auplc-installer install --image-tag=develop

# Build images locally instead of pull
./auplc-installer install --image-source=build

# Override GPU detection
./auplc-installer install --gpu=strix-halo

# Install only Hub plus CPU/GPU base environments
./auplc-installer install --courses=basic

# Use containerd mode for more portable/offline-oriented operation
./auplc-installer install --runtime=containerd

# Use registry and package mirrors
./auplc-installer install \
  --mirror=mirror.example.com \
  --mirror-pip=https://pypi.tuna.tsinghua.edu.cn/simple

Uninstall

./auplc-installer uninstall

Next Steps

• For all installer subcommands, TUI prompts, and runtime workflows, see Single-Node Deployment

• To add a course or change deployment settings after install, see Customizing a Single-Node Deployment

• For auth and resource configuration, see JupyterHub Configuration