Exploring dynamic load imbalance solutions with the CoMD proxy application

Proxy applications are developed to simplify studying parallel performance of scientific simulations and to test potential solutions for performance problems. However, proxy applications are typically too simple to allow work migration or to represent the load imbalance of their parent applications....

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Future generation computer systems 2019-03, Vol.92 (C), p.920-932
Hauptverfasser: Pearce, Olga, Ahmed, Hadia, Larsen, Rasmus W., Pirkelbauer, Peter, Richards, David F.
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:Proxy applications are developed to simplify studying parallel performance of scientific simulations and to test potential solutions for performance problems. However, proxy applications are typically too simple to allow work migration or to represent the load imbalance of their parent applications. To study the ability of load balancing solutions to balance work effectively, we enable work migration in one of the Exascale Co-design Center for Materials in Extreme Environments (ExMatEx) [[1]] applications, CoMD. We design a methodology to parameterize three key aspects necessary for studying load imbalance correction: (1) the granularity with which work can be migrated; (2) the initial load imbalance; (3) the dynamic load imbalance (how quickly the load changes over time). We present a study of the impact of flexibility in work migration in CoMD on load balance and the associated rebalancing costs for a wide range of initial and dynamic load imbalance scenarios. •A proxy application that allows us to study load imbalance and potential solutions.•Ability to control initial and dynamic load imbalance, flexibility in work migration.•An evaluation of a load balance algorithm performance on a wide range of scenarios.
ISSN:0167-739X
1872-7115
DOI:10.1016/j.future.2017.12.010