Latency-Optimized Parallelization of the FMM Near-Field Computations

Authors:
Ivo Kabadshow;Bruno Lang
Affiliations:
John von Neumann Institute for Computing, Central Institute for Applied Mathematics, Research Centre Jülich, Germany;Applied Computer Science and Scientific Computing Group, Department of Mathematics, University of Wuppertal, Germany
Venue:
ICCS '07 Proceedings of the 7th international conference on Computational Science, Part I: ICCS 2007
Year:
2007

Citing 6
Cited 0

A fast algorithm for particle simulations

Journal of Computational Physics
Analysis of Multi-Dimensional Space-Filling Curves

Geoinformatica
Communications overlapping in fast multipole particle dynamics methods

Journal of Computational Physics
Massively parallel implementation of a fast multipole method for distributed memory machines

Journal of Parallel and Distributed Computing
Advances, Applications and Performance of the Global Arrays Shared Memory Programming Toolkit

International Journal of High Performance Computing Applications
High Performance Remote Memory Access Communication: The Armci Approach

International Journal of High Performance Computing Applications

Quantified Score

Hi-index	0.00

Visualization

Abstract

In this paper we present a new parallelization scheme for the FMM near-field. The parallelization is based on the Global Arrays Toolkit and uses one-sided communication with overlapping. It employs a purely static load-balancing approach to minimize the number of communication steps and benefits from a maximum utilization of data locality. In contrast to other implementations the communication is initiated by the process owning the data via a putcall, not the process receiving the data (via a getcall).