A shared memory parallel algorithm for data reduction using the singular value decomposition

Authors:
Rhonda D. Phillips;Layne T. Watson;Randolph H. Wynne
Affiliations:
Virginia Polytechnic Institute and State University, Blacksburg, Virginia;Virginia Polytechnic Institute and State University, Blacksburg, Virginia;Virginia Polytechnic Institute and State University, Blacksburg, Virginia
Venue:
Proceedings of the 2008 Spring simulation multiconference
Year:
2008

Citing 6
Cited 0

LAPACK's user's guide

LAPACK's user's guide
Applied numerical linear algebra

Applied numerical linear algebra
Remote Sensing Digital Image Analysis: An Introduction

Remote Sensing Digital Image Analysis: An Introduction
Hybrid image classification and parameter selection using a shared memory parallel algorithm

Computers & Geosciences
Examining pine spectral separability using hyperspectral data from an airborne sensor: An extension of field-based results

International Journal of Remote Sensing
Validity of the single processor approach to achieving large scale computing capabilities

AFIPS '67 (Spring) Proceedings of the April 18-20, 1967, spring joint computer conference

Quantified Score

Hi-index	0.00

Visualization

Abstract

This paper presents a parallel version of a singular value decomposition (SVD) based data reduction method for remotely sensed data that uses a training data set to find good basis vectors for the reduced dimension data. The parallel algorithm is implemented using Fortran 95, OpenMP, and LAPACK, and speedup results are given for up to 128 1.6 GHz processors of an SGI Altix 3700 with 512 MB of RAM. Performance is evaluated for various parallel options including scheduling strategies, data initialization, use of dplace, variable storage options, and cache optimization. Parallel speedup (without including time for input/output) using 128 processors reaches 75 with static scheduling and 190 with dynamic (guided) scheduling.