Opendda: A novel high-performance computational framework for the discrete dipole approximation

James Mc Donald, Aaron Golden, S. Gerard Jennings

Research output: Contribution to journalArticle

15 Citations (Scopus)

Abstract

This work presents a highly optimized computational framework for the Discrete Dipole Approximation, a numerical method for calculating the optical properties associated with a target of arbitrary geometry that is widely used in atmospheric, astrophysical and industrial simulations. Core optimizations include the bit-fielding of integer data and iterative methods that complement a new Discrete Fourier Transform (DFT) kernel, which efficiently calculates the matrix' vector products required by these iterative solution schemes. The new kernel performs the requisite 3-D DFTs as ensembles of 1-D transforms, and by doing so, is able to reduce the number of constituent 1-D transforms by 60% and the memory by over 80%. The optimizations also facilitate the use of parallel techniques to further enhance the performance. Complete OpenMP-based shared-memory and MPI-based distributed-memory implementations have been created to take full advantage of the various architectures. Several benchmarks of the new framework indicate extremely favorable performance and scalability.

Original languageEnglish (US)
Pages (from-to)42-61
Number of pages20
JournalInternational Journal of High Performance Computing Applications
Volume23
Issue number1
DOIs
StatePublished - Mar 2009
Externally publishedYes

Fingerprint

Dipole
High Performance
Transform
kernel
Data storage equipment
Discrete Fourier transforms
Optimization
Cross product
Discrete Fourier transform
Matrix Product
OpenMP
Distributed Memory
Iterative Solution
Approximation
Shared Memory
Optical Properties
3D
Scalability
Ensemble
Complement

Keywords

  • CG-FFT
  • Discrete dipole approximation
  • Matrix-vector product
  • Optimization
  • Parallel algorithm

ASJC Scopus subject areas

  • Software
  • Hardware and Architecture
  • Theoretical Computer Science

Cite this

Opendda : A novel high-performance computational framework for the discrete dipole approximation. / Mc Donald, James; Golden, Aaron; Jennings, S. Gerard.

In: International Journal of High Performance Computing Applications, Vol. 23, No. 1, 03.2009, p. 42-61.

Research output: Contribution to journalArticle

@article{7ce329aad6bd4f16a68d9a62b2b6907d,
title = "Opendda: A novel high-performance computational framework for the discrete dipole approximation",
abstract = "This work presents a highly optimized computational framework for the Discrete Dipole Approximation, a numerical method for calculating the optical properties associated with a target of arbitrary geometry that is widely used in atmospheric, astrophysical and industrial simulations. Core optimizations include the bit-fielding of integer data and iterative methods that complement a new Discrete Fourier Transform (DFT) kernel, which efficiently calculates the matrix' vector products required by these iterative solution schemes. The new kernel performs the requisite 3-D DFTs as ensembles of 1-D transforms, and by doing so, is able to reduce the number of constituent 1-D transforms by 60{\%} and the memory by over 80{\%}. The optimizations also facilitate the use of parallel techniques to further enhance the performance. Complete OpenMP-based shared-memory and MPI-based distributed-memory implementations have been created to take full advantage of the various architectures. Several benchmarks of the new framework indicate extremely favorable performance and scalability.",
keywords = "CG-FFT, Discrete dipole approximation, Matrix-vector product, Optimization, Parallel algorithm",
author = "{Mc Donald}, James and Aaron Golden and Jennings, {S. Gerard}",
year = "2009",
month = "3",
doi = "10.1177/1094342008097914",
language = "English (US)",
volume = "23",
pages = "42--61",
journal = "International Journal of High Performance Computing Applications",
issn = "1094-3420",
publisher = "SAGE Publications Inc.",
number = "1",

}

TY - JOUR

T1 - Opendda

T2 - A novel high-performance computational framework for the discrete dipole approximation

AU - Mc Donald, James

AU - Golden, Aaron

AU - Jennings, S. Gerard

PY - 2009/3

Y1 - 2009/3

N2 - This work presents a highly optimized computational framework for the Discrete Dipole Approximation, a numerical method for calculating the optical properties associated with a target of arbitrary geometry that is widely used in atmospheric, astrophysical and industrial simulations. Core optimizations include the bit-fielding of integer data and iterative methods that complement a new Discrete Fourier Transform (DFT) kernel, which efficiently calculates the matrix' vector products required by these iterative solution schemes. The new kernel performs the requisite 3-D DFTs as ensembles of 1-D transforms, and by doing so, is able to reduce the number of constituent 1-D transforms by 60% and the memory by over 80%. The optimizations also facilitate the use of parallel techniques to further enhance the performance. Complete OpenMP-based shared-memory and MPI-based distributed-memory implementations have been created to take full advantage of the various architectures. Several benchmarks of the new framework indicate extremely favorable performance and scalability.

AB - This work presents a highly optimized computational framework for the Discrete Dipole Approximation, a numerical method for calculating the optical properties associated with a target of arbitrary geometry that is widely used in atmospheric, astrophysical and industrial simulations. Core optimizations include the bit-fielding of integer data and iterative methods that complement a new Discrete Fourier Transform (DFT) kernel, which efficiently calculates the matrix' vector products required by these iterative solution schemes. The new kernel performs the requisite 3-D DFTs as ensembles of 1-D transforms, and by doing so, is able to reduce the number of constituent 1-D transforms by 60% and the memory by over 80%. The optimizations also facilitate the use of parallel techniques to further enhance the performance. Complete OpenMP-based shared-memory and MPI-based distributed-memory implementations have been created to take full advantage of the various architectures. Several benchmarks of the new framework indicate extremely favorable performance and scalability.

KW - CG-FFT

KW - Discrete dipole approximation

KW - Matrix-vector product

KW - Optimization

KW - Parallel algorithm

UR - http://www.scopus.com/inward/record.url?scp=62249134423&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=62249134423&partnerID=8YFLogxK

U2 - 10.1177/1094342008097914

DO - 10.1177/1094342008097914

M3 - Article

AN - SCOPUS:62249134423

VL - 23

SP - 42

EP - 61

JO - International Journal of High Performance Computing Applications

JF - International Journal of High Performance Computing Applications

SN - 1094-3420

IS - 1

ER -