Computational Computing Overview

The Economics computing cluster is a Linux based computational environment. It consists of several systems that are able to be logged into interactively, a set of systems to which computing jobs may be submitted, and, a back end filesystem shared amongst the interactive and batch systems.

Economic Computing Services divides computational computing resources into two main categories: storage and computational nodes. Storage refers to where you put data whereas computational nodes refer to what you use to manipulate data. This document deals with computational nodes. Computational nodes fall into three main categories:

Local Computation

This refers to using a local workstation, laptop or office computer to compile code or run programs such as Stata or Matlab, SAS, R or Stand RH programs to manipulate data. Local computation will be done in whatever environment you use for your desktop computing. There are a wide variety of options. The most well supported desktop OS in the department is Microsoft Windows. We have a growing support base for Apple’s OS X. Linux OS’s are also supported in a “best effort” manner, the most expertise in the group being with Red Hat derived systems.

Interactive Linux Cluster

Interactive nodes are usually accessed remotely via ssh and are shared resources to develop and test code, and run short jobs (like Stata or Matlab) to manipulate data. A list of interactive nodes is available here. Select any of the nodes that are generally green. If the node only has a red block associated with it, it may be offline. For instructions on accessing the interactive nodes, click the subheading above.

Batch nodes

Batch nodes are similar to interactive nodes, but once you have a compiled program to run, or have written code for programs such as Stata and Matlab, you submit the job to a queueing process which waits for available time in the batch cluster. For instructions on accessing the batch cluster, click the subheading above. At times using interactive nodes or the batch queue are both referred to as “using the cluster”, so when it is important to specify which of the two you are wanting to use or having difficulty with, it is best to say “I am using the interactive nodes” or “I have a job in the batch cluster” to provide clarity. The cluster, both interactive and batch nodes, are Red Hat based systems currently running Centos 5.10. A general overview of available Linux related services can be found here and a comprehensive list of available applications in the linux environment may be found here.