0.0 Introduction - UNR-HPC/pronghorn GitHub Wiki

Getting started with Pronghorn

This guide has been written to help get you to to speed on what Pronghorn is and how to use it effectively. Pronghorn is a batch processing High Performance Computing (HPC) cluster. As such you will need to consider the following:

Your skill level

• Using an HPC cluster requires, at minimum, an intermediate to advanced understanding of the Linux operating system.
• The videos section of this guide contains essential Linux training videos which, once completed, will get you proficient in using Linux.

Ideal job characteristics for distributed computing

• Your research workflow can be split into independent jobs.
• Jobs can be run on a single processor, multiple processors, a compute node, or multiple compute nodes.
• Ideally your code, application, or complete software pipeline is “portable” or containerized.

Reasons for using a batch processing HPC

• Your computations access or produce large datasets.
• Your computations need or benefit from parallelism.
• Your computations need large numbers of cores simultaneously.
• Your computations require large amounts of memory.
• Your computations require a high performance shared filesystem.

Poorly suited job characteristics for distributed HPC computing

A batch processing system is not a good solution if:

• You need results immediately after submission (i.e. an interactive environment).
• Computing resources are accessed through a queue on a batch system and therefore suited for longer processing times, with job durations typically measured in hours, or multiple days.