Welcome to the Tillicum Onboarding Tutorial — a guided introduction to the Tillicum GPU cluster at the University of Washington. This repository provides step-by-step materials to help new users get started with Tillicum, including examples, commands, and best practices.
Tillicum is a GPU-accelerated research computing cluster supporting AI, data science, and HPC workloads. This tutorial covers everything you need to start using Tillicum effectively — from logging in to running interactive notebooks on GPUs.
By completing this tutorial, you’ll learn how to:
- Connect to the UW VPN
- Log in and access Tillicum resources
- Navigate the filesystem and manage data
- Transfer data with scp and Globus
- Submit and monitor jobs with Slurm
- Load and manage software modules
- Use Conda computing environments
- Launch Jupyter notebooks through Open OnDemand
Each topic in this tutorial is contained in its own Markdown file for easy navigation:
| Section | Description |
|---|---|
| 01-vpn.md | Connecting to the UW VPN (required for Open OnDemand) |
| 02-logging-in.md | How to log in via SSH |
| 03-filesystem.md | Understanding Tillicum filesystem, storage, and best practices |
| 04-file-transfer.md | Transferring data using Globus (installing Globus Connect Personal) |
| 05-slurm.md | Submitting, monitoring, and managing jobs |
| 06-modules.md | Using LMOD to load software modules |
| 07-computing-env.md | Setting up Conda environments |
| 08-ood-jupyter.md | Running Jupyter notebooks via Open OnDemand |
| 09-task.md | Hands-on exercise |
Link to introduction video (recorded on October 10), which reviews the materials and concepts: https://youtu.be/aiMAre1GS3U
We’d love your feedback to help improve this tutorial and future Tillicum trainings. After completing the tutorial or attending the workshop, please fill out our feedback form.
