Application-level scheduling on distributed heterogeneous networks
Supercomputing '96 Proceedings of the 1996 ACM/IEEE conference on Supercomputing
Simulation Modeling and Analysis
Simulation Modeling and Analysis
Adaptive Computing on the Grid Using AppLeS
IEEE Transactions on Parallel and Distributed Systems
Heuristics for Scheduling Parameter Sweep Applications in Grid Environments
HCW '00 Proceedings of the 9th Heterogeneous Computing Workshop
HCW '00 Proceedings of the 9th Heterogeneous Computing Workshop
HINT: A new way to measure computer performance
HICSS '95 Proceedings of the 28th Hawaii International Conference on System Sciences
CCGRID '02 Proceedings of the 2nd IEEE/ACM International Symposium on Cluster Computing and the Grid
The Challenge of Volunteer Computing with Lengthy Climate Model Simulations
E-SCIENCE '05 Proceedings of the First International Conference on e-Science and Grid Computing
Hi-index | 0.00 |
Many simulation efforts in ecology and evolutionary biologyemploy individual-based models that are well suited for includingmany biological details. These models often pose seriouscomputational challenges if all biologically interesting parametercombinations are to be explored. The challenges are even greaterfor biologists who often lack supercomputing facilities and themanpower for implementing complex global computing systems such asSETI@home. Under such limiting conditions, evolution@home startedas a one-man effort to distribute simulations of Muller's ratchetto Internet-connected computers of participants from the generalpublic. This paper addresses experiences in low-effort globalcomputing made with evolution@home over more than four years. Itshows how allowing participants to choose the class ofcomputational complexity they want to contribute to can help todeal with the bewildering variety of computational complexitiesthat easily result from individual-based models. Results suggestthat, as a first rough approximation, participants' complexitychoices are distributed surprisingly even over all reasonableclasses of CPU-time and RAM requirements. More often than not,participants tend to finish the simulations they start, if they arecommitted enough to submit any results at all. Potential uses ofintermediate simulation results are discussed and the error ofmagnitude is introduced to help to deal with imprecise CPU-timepredictions. Experiences with the choices of over 300 users whohave contributed more than 100 000 simulations with a total of over80 years CPU time are reviewed. Copyright © 2007 John Wiley& Sons, Ltd.