ARCHER Quick Start

This screencast is an demonstration walkthrough of the ARCHER Quick Start Guide at:

http://www.archer.ac.uk/documentation/getting-started/

It covers: compiling a simple parallel (MPI) program, creating a job submission script for PBSPro, submitting the job to the batch scheduling system and checking the output.

Logging on to ARCHER from Windows using PuTTY

How to access ARCHER via SSH from Windows using the PuTTY SSH client. For full description see Interactive Access http://www.archer.ac.uk/documentation/user-guide/connecting.php#sec-2.1

Logging in to ARCHER using SSH keys

How to use SSH keys instead of passwords for logging in to ARCHER - it's more secure and also easier for you as you don't need to remember your password every time! Full documentation at:

http://www.archer.ac.uk/documentation/user-guide/connecting.php#sec-2.3

Running Interactive jobs on ARCHER

Interactive jobs can be particularly useful for testing and debugging as they allow instant feedback on commands launched on the compute nodes. This screencast demonstrates how to launch interactive jobs on ARCHER and how to run programs on the compute node from within an interactive job. Full documentation on running Interactive jobs on ARCHER can be found in the User Guide at:

http://www.archer.ac.uk/documentation/user-guide/batch.php#sec-5.4.8

Finding help and documentation from the command line (man pages etc.)

Did you forget that really convenient option for a Linux shell command you use all the time? Is there a command you use less frequently and need a reminder about? Would you like to find out more information about what compiler flags actually do, what the options for qsub are and what all those aprun options actually mean?
This screencast will help you find out. You will be shown three different ways to get help directly from the command line, including how to access and navigate man (manual) pages. As you will see these mechanisms give you quick, direct access to relevant documentation. You will be able to use this to improve not only your use of ARCHER but of any Linux and indeed Unix-variant system.

Using STAT for parallel debugging

This video demonstrates how to use the STAT tool (from University of Wisconsin-Madison), for debugging deadlocked or none-progressing parallel programs, on ARCHER. Further details on STAT can be found at the ARCHER website here:

http://www.archer.ac.uk/documentation/best-practice-guide/debug.php#sec-7.3

Using Reveal to Improve Parallel Performance

Reveal is a performance analysis tool supported by Cray that guides you in improving the runtime performance of a Cray-compiled application code. Further details on Reveal can be found at the ARCHER website here:

http://www.archer.ac.uk/documentation/best-practice-guide/performance.php#sec-5.3