evolution@home demonstrates minimalist global computing
Right from the start evolution@home employed a minimalist approach to global computing due to a severe lack of resources. The manuscript describing experiences on the way is now available.
Global computing can be a very complex enterprise. In fact it is very easy to make it ever more complicated, not only for users, but especially for the developers behind the scenes. But nobody likes too much complexity. Therefore it is desirable to explore the simplest possible approaches that can be taken to global computing.
Exactly this approach was taken by evolution@home. The lack of extensive development resources forced simplicity of design, but evolution@home turned this around and deliberately explored the limits of simplicity in global computing.
The experiences of this experiment in simplicity have been described in a manuscript that is now available after having been reviewed by the Journal "Software Practice & Experience" (DOI link).
At its core this semi-automated system requires users to download a simulator executable and simulation work units, which are ready for computation. After completion the simulator produces a results file that is then emailed back for inclusion in a central repository. Such a design is possible, because each work unit needs to be computed many times in order to determine the mean and the standard deviation of each individual observation.
While the limitations of this current semi-automated system are clear and the future of evolution@home will be fully-automated, it is indeed very helpful to start with such a more basic system.
For more details, consult the full report here (DOI link).
A preliminary analysis of participation patterns was already published earlier (Loewe L (2002) "evolution@home: Experiences with work units that span more than 7 orders of magnitude in computational complexity", p. 425-431. 2nd International Workshop on Global and Peer-to-Peer Computing on Large Scale Distributed Systems at the 2nd IEEE/ACM International Symposium on Cluster Computing and the Grid (CCGrid2002), 21-24 May, Berlin, Germany, IEEE Computer Society. PDF | DOI | ConferenceProceedings).
This news item is part of the evolution@home results update season (added 22 Apr 2008).