Advanced Peregrine course

In this four hour course, which is a follow-up to the Peregrine introduction course, an overview of several parallelization and job handling methods will be given. This will help you in making a more effective use of the Peregrine cluster.

Contents

The following topics will be discussed:

• Some details of modern system architecture;
• Differences between shared and distributed memory parallelization;
• Introduction to bash scripting for writing advanced job scripts;
• Job arrays, submitting many jobs at the same time.

The course will focus on getting familiar with the concepts, not on a thorough explanation of all the details, as the latter would not fit in a single afternoon.

Course format

The course will be a mixture of lectures and hands-on exercises, in which you will work with the concepts explained.

Aim

After the course you should be able to understand how to work with shared memory or distributed memory applications on the cluster. You should be able to write more advanced job scripts that make it possible to automate job tasks. Finally you should be able to perform parameter sweeps using job arrays.

Prerequisites

For this course basic knowledge about the Linux command line, and the Peregrine cluster is assumed.