Training

Guides we’ve written for this cluster, plus the external tutorials we recommend for SLURM, Python environments, GPUs, and parallel computing.

Self-Study Resources

Internal

Quick Start

A 15-minute walkthrough, from logging in to your first submitted job.

Quick Start Guide

Example Jobs

Ready-to-run SBATCH templates for serial, MPI, GPU, array, and application jobs.

Example Job Scripts

GPU Computing

Working with the cluster’s GPUs — H100/H200, CUDA, and deep-learning workflows.

GPU Computing

Best Practices

Right-sizing your requests, checkpointing long runs, and keeping I/O fast.

Best Practices

External (recommended)

New to the cluster? Start with SLURM and Python environments, then reach for the rest as your work grows.

SLURM — the job scheduler

Quick Start User Guide — slurm.schedmd.com/quickstart — the official introduction to submitting and managing jobs, from the team that builds SLURM.

Python environments — conda, mamba & venv

Conda: Managing environments — docs.conda.io — the official guide to creating, activating, exporting, and sharing environments.
Miniforge (conda + mamba) — github.com/conda-forge/miniforge — the recommended way to install conda and the much faster mamba solver, with the conda-forge channel preconfigured.
Python venv & pip — docs.python.org — the lightweight, standard-library option for when you just need a pure-Python environment without conda.

Broader HPC, GPUs & parallel computing

NERSC Training — nersc.gov/users/training — free recordings on parallel programming, GPU computing, and AI/ML at scale.
NVIDIA Deep Learning Institute — nvidia.com/training — free self-paced courses on CUDA, RAPIDS, and modern AI workflows.
Software Carpentry — software-carpentry.org/lessons — the Unix shell, Python, R, and Git, from the ground up.
HPC Carpentry — hpc-carpentry.org — entry-level HPC concepts: schedulers, parallelism, and file systems.
OpenMP & MPI tutorials — hpc-tutorials.llnl.gov — LLNL’s long-standing OpenMP and MPI reference tutorials.

Announcements

Maintenance windows and service changes go out on the central-hpc-users@caltech.edu mailing list. Email help-hpc@caltech.edu and we’ll add you — it’s how you’ll hear about downtime before it affects your jobs.