RT info:eu-repo/semantics/article
T1 A dynamic multi¿Objective approach for dynamic load balancing in heterogeneous systems
A1 Almeida Rodríguez, Francisco Carmelo
A1 Cabrera Pérez, Alberto
A1 Acosta, Alejandro
A1 Blanco Pérez, Vicente José
K1 Dynamic load balancing
K1 Energy efficiency
K1 Iterative algorithms
K1 Heterogeneous computing
AB Modern standards in High Performance Computing (HPC) have started to consider energy consumption and power draw asa limiting factor. New and more complex architectures have been introduced in HPC systems to afford these new restrictions, andinclude coprocessors such as GPGPUs for intensive computational tasks. As systems increase in heterogeneity, workload distributionbecomes a more core problem to achieve the maximum efficiency in every computational component. We present a Multi-ObjectiveDynamic Load Balancing (DLB) approach where several objectives can be applied to tune an application. These objectives can bedynamically exchanged during the execution of an algorithm to better adapt to the resources available in a system. We haveimplemented the Multi–Objective DLB together with a generic heuristic engine, designed to perform multiple strategies for DLB initerative problems. We also present Ull Multiobjective Framework (UllMF), an open–source tool that implements the Multi-Objectivegeneric approach. UllMF separates metric gathering, objective functions to be optimized and load balancing algorithms, and improvescode portability using a simple interface to reduce the costs of new implementations. We illustrate how performance and energyconsumption are improved for the implemented techniques, and analyze their quality using different DLB techniques from the literature.
SN 1045-9219
YR 2020
FD 2020
LK http://riull.ull.es/xmlui/handle/915/35147
UL http://riull.ull.es/xmlui/handle/915/35147
LA en
DS Repositorio institucional de la Universidad de La Laguna
RD 12-may-2024