Evaluation of a Nearest-Neighbor Load Balancing Strategy for Parallel Molecular Simulations in MPI Environment
We evaluate the effectiveness of an iterative nearest-neighbor load balancing algorithm: Generalized Dimension Exchange (G.D.E.) method over Molecular Dynamics Simulations. Due to the overhead of the method, the simulation performance improves when the system tolerates some degree of unbalance. We p...
Gespeichert in:
Hauptverfasser: | , |
---|---|
Format: | Buchkapitel |
Sprache: | eng |
Schlagworte: | |
Online-Zugang: | Volltext |
Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
Zusammenfassung: | We evaluate the effectiveness of an iterative nearest-neighbor load balancing algorithm: Generalized Dimension Exchange (G.D.E.) method over Molecular Dynamics Simulations.
Due to the overhead of the method, the simulation performance improves when the system tolerates some degree of unbalance. We present a method to calibrate the best tolerance level for a simulation running on a homogeneous platform using MPI. |
---|---|
ISSN: | 0302-9743 1611-3349 |
DOI: | 10.1007/3-540-45825-5_38 |