![]() |
Institute for Theoretical Physics | |||||
statistical mechanics group|
leonardo
|
||||||
|
||||||
What was our motivation to build such a big Beowulf cluster? Well, there are different reasons, here are some:
In science there is an increasing need for computing power. The classical way is to apply for an account on a supercomputer of a national computing center, for example Jülich. However, than you have to share this big, big computer with a lot of people. Your job has to wait a long time in the queue, until it is processed. With the aim of having supercomputer power in for our groups we planned Tina.
Beowulfs are the best way to get a lot of GFlops for your money if you can parallelize your code. But why are commercial supercomputers so expensive in contrast to the "cheap" beowulfs? If you buy one of these computers, you spend most of your money for an high-bandwidth/low-latency network. If you don't need this, then you should choose a beowulf.
If you build a beowulf cluster, you can optimize the design according to your needs. Memory, CPU, hard-disk, network, everything can be fit to your needs.
Until now, there are not too many big beowulf clusters like Tina or CLIC in Germany. Building and running of such a computer is a big challenge. Sharing information is one of the basic ideas of LINUX/Gnu/Beowulf, so we will share our experiences with the community.
We are offering lectures about parallel computing. Our students can collect some additional experiences, if they can "play around" on a big supercomputer.
Hey, it's fun, really. Get some old PCs, get a cheap hub, a few cable, a penguin and enjoy your private supercomputer.
URL: http://tina.cryptoweb.de
Sunday, January 11th 2026, 17:24:45 CET