[en] This article introduces an efficient multithreading local search algorithm for solving the multiobjective scheduling problem in heterogeneous computing systems considering the makespan and energy consumption objectives. The proposed method follows a fully multiobjective approach using a Pareto-based dominance search executed in parallel. The experimental analysis demonstrates that the new multithreading algorithm outperforms a set of deterministic heuristics based on Min-Min. The new method is able to achieve significant improvements in both objectives in reduced execution times for a broad set of testbed instances.
Centre de recherche :
ULHPC - University of Luxembourg: High Performance Computing
Disciplines :
Sciences informatiques
Identifiants :
UNILU:UL-CONFERENCE-2012-398
Auteur, co-auteur :
Iturriaga, Santiago; Universidad de la República, Uruguay
Nesmachnow, Sergio; Universidad de la República, Uruguay
DORRONSORO, Bernabé ; University of Luxembourg > Interdisciplinary Centre for Security, Reliability and Trust (SNT)
Langue du document :
Anglais
Titre :
A Multithreading Local Search For Multiobjective Energy-Aware Scheduling In Heterogeneous Computing Systems
Date de publication/diffusion :
2012
Nom de la manifestation :
European Conference on Modelling and Simulation (ECMS)
Lieu de la manifestation :
Koblenz, Allemagne
Date de la manifestation :
from 29-05-2012 to 1-06-2012
Titre de l'ouvrage principal :
European Conference on Modelling and Simulation (ECMS)
ISBN/EAN :
978-0-9564944-4-3
Pagination :
1 - 7
Peer reviewed :
Peer reviewed
Commentaire :
26th European Conference on Modelling and Simulation (ECMS)
