The Amber Cluster
The mathematics department and the computer-science department share a 16-node parallel cluster called Amber.
The nodes communicate using special message-passing software that implements protocols such as the Message-Passing Interface (MPI) or Parallel Virtual Machine (PVM).
What Does It Do, and How Does It Work?
High-Performance Computing Clusters (HPCCs) are a way of using multiple relatively inexpensive (“commodity”) computers to emulate and even surpass the performance characteristics of much, much more expensive multiprocessor systems.
Our cluster is a “Class I Beowulf cluster”. It uses commodity hardware, in this case Dell PowerEdge 400SC computers, each with a 2.8 GHz Pentium 4 processor, 1 GB or 1.25 GB of RAM, and a gigabit Ethernet port.
The nodes communicate through a gigabit Ethernet switch using special “message-passing” software that runs on top of the regular network protocols.
To take advantage of the parallel-processing ability of the cluster, the algorithms used in software have to be inherently parallelizable and your code has to be written to use the message-passing libraries to allow communication between the nodes. You can't take just any program, run it on a cluster, and see parallel speedup.
For lots more background information about the Beowulf computing
concept, see beowulf.org and the
Beowulf FAQ.
Using the Amber Cluster
Getting access to the cluster requires you to have a mathematics department Linux account and special permission to be added to the necessary Unix group to allow you to log into the cluster machines.
The Amber Cluster Appropriate-Use Policy section explains the usage restrictions and provides a place to download the necessary Amber Account-Request Form.
Writing and Running Parallel Programs
More information about writing and running parallel programs is available in our parallel-software support section.
