COS Compute Handbook
Preface
This handbook covers using the HPC cluster for computational research, with a focus on bioinformatics workflows. It walks through everything from first login to optimizing multi-sample parallel jobs.
What You Will Learn
| Chapter | Topic |
|---|---|
| HPC Terminology | Clusters, nodes, cores, memory vs. storage |
| Getting Started | Account setup, login, file transfer |
| Running Jobs | SLURM job scripts, monitoring, common errors |
| Best Practices for Job Scripts | Config + job-script split, strict bash, validation, reproducibility |
| Job Arrays | Scaling to hundreds of samples |
| Job Performance | Resource estimation and optimization |
| Globus | High-speed data transfers |
| Downloading Data | Pulling data onto Hazel from public sources |
| GNU Parallel | Controlled parallelism on login nodes |
| Git and GitHub on Hazel | SSH-key auth, develop-locally workflow |
Customizing Your .bashrc |
Aliases, functions, directory shortcuts, and PATH/env tweaks |
Conventions
Shell commands are shown in code blocks. Commands you type at a prompt begin with $; omit the $ when copying. The intended output of the command is shown below the prompt, without a proceeding $.
$ echo "hello world!"
hello world!
Important
Never run computationally intensive work on login nodes. Always submit jobs through SLURM.
Support
- BRC support: ncsate_brc@ncsu.edu
- OIT HPC help: oit_hpc@help.ncsu.edu | 919-515-HELP