Distributed Computing

Modern physics research relies nowadays on three components, vertices of a virtuous triangle: experiments, theory and computing. In the long history of research, computing is the ``new kid on the block''. There are no experiment design, no data taking, no data analysis, no data interpretation or no theoretical prediction without computing. Computing has become ubiquitous in physics research.

But computing has also to deliver or it would be seen as an impediment to the progress of science. It has to fulfill the increasing science requirements that, in term of performance, have become gigantic, much larger than what ``Moore's law" allows. The only alternative is to parallelize or distribute the computing load over a large number of processing units.

Massive Parallel Systems, computer clusters, data/compute GRID architectures or dual/quad/multi-core processors have attracted most of the large investments both in human resources and financial support. If many aspects of the GRID projects were properly covered in this workshop, little was said about another approach to distributed computing: ``Public or Internet Distributed Computing" (PDC) like BOINC..

• Massive Parallel Systems
• Clusters
• GRID
• BOINC
• Multi-core CPU

-- DenisPerretGallix - 19 Oct 2005