BuckAI HPC Practicum

A 1-credit-hour, self-guided, asynchronous practicum that takes grad students in Earth Sciences (and the broader College of Arts & Sciences) from "I have a Unity account and don't know what to do with it" to "I can run reproducible HPC jobs and analyze the results."

👋 New to HPC? Not sure if this course is for you?

Read Is this course for you? first — a plain-English explanation of what an HPC cluster is, who needs one (and who doesn’t), and what Windows / Mac / Linux users each have to do to participate. Especially recommended if terms like “HPC”, “Unity”, “SSH”, or “Slurm” don’t yet mean anything to you.

A 1-credit, self-guided practicum in real-world HPC

This course is for grad students who want to become comfortable running research code on a real HPC cluster — but who don’t have time for a full semester of system-administration theory. Each week takes about 1 hour and produces a concrete artifact (a config file, a Slurm job log, a seff efficiency report, a screenshot, a reproducible repo).

By the end of the semester you will have:

A working SSH + VS Code Remote-SSH setup against OSU’s Unity HPC cluster, with one-Duo-per-session multiplexing
A productive shell environment (.bashrc, aliases, umask 002, group/partition awareness)
A persistent-session workflow (tmux + livenode) that survives laptop sleeps and reboots
Fluency with mamba environments, including the dangers of mixing pip
A working Jupyter-on-the-cluster setup, both via OnDemand and via SSH tunneling
The instincts of a polite HPC user: right-sizing memory, CPU, and walltime requests so your jobs run sooner and don’t block others
A small capstone project packaged as a reproducible GitHub repo, run end-to-end on Unity or OSC

Read

Each lab is grounded in chapters from the BuckAI HPC Handbook, our open-source reference. You’ll read 1–2 chapters per week before doing the lab.

Do

Twelve hands-on labs over the first 12 weeks of the semester, plus a 3-week capstone project. Most labs produce a verifiable artifact you can self-check.

Start here

Syllabus — week-by-week structure, learning outcomes, expectations, and how to get help
Labs overview — one-line summary of all 12 labs
Lab 1 — SSH config and multiplexing — get connected; takes ~1 hour
Capstone project guide — what the final 3 weeks look like

About this course

Format: Self-guided / asynchronous. No lectures, no fixed meeting time.
Time commitment: ~1 hour/week for 15 weeks (with some labs running longer if you hit a snag — HPC is like that).
Audience: Grad students in Earth Sciences and related departments in the OSU College of Arts and Sciences. No prior HPC experience required, but comfort with Python and a Unix-style terminal is assumed.
Platform: Primary target is OSU’s Unity HPC cluster at ASC, with optional bonus tracks on OSC (Pitzer, Cardinal, Owens) — getting an account on either is straightforward for OSU students.
Source: This practicum and its reading materials are open-source. Corrections, additions, and pull requests welcome.

Acknowledgments

Built at the BuckAI Observatory at Ohio State University. Underlying reading material is the BuckAI HPC Handbook. Site built with Quarto.