Energy efficient scheduling in heterogeneous systems with a parallel multiobjective local search

Santiago Iturriaga (1) , Sergio Nesmachnow (1) , Dorronsoro Bernabe (2) , Pascal Bouvry (3)

1 INCO - Instituto de Computacion [Montevideo]
2 DOLPHIN - Parallel Cooperative Multi-criteria Optimization
3 Uni.lu - Université du Luxembourg

This article introduces ME-MLS, an e cient multithreading local search algorithm for solving the multiobjective scheduling problem in heterogeneous com- puting systems. We consider the minimization of both the makespan and energy consumption objectives. The proposed method follows a fully multiobjective ap- proach, applying a Pareto-based dominance search that is executed in parallel by using several threads. The experimental analysis demonstrates that the new multi- threading algorithm outperforms a set of fast and accurate two-phases deterministic heuristics based on the traditional MinMin. The new ME-MLS method is able to achieve signi cant improvements in both makespan and energy consumption objec- tives in reduced execution times for a large set of testbed instances, while exhibiting a near linear speedup behavior when using up to 24 threads.

Domaines: Intelligence artificielle [cs.AI]

Santiago Iturriaga, Sergio Nesmachnow, Dorronsoro Bernabe, Pascal Bouvry. Energy efficient scheduling in heterogeneous systems with a parallel multiobjective local search. Computing and Informatics, 2013, 32 (2), pp.273-294. ⟨hal-00872333⟩