Application of the OpenCL API for Implementation of the NIPALS Algorithm for Principal Component Analysis of Large Data Sets

Author

Bowden, Joshua C.

Author_Institution

CSIRO Inf. Manage. & Technol., Clayton, VIC, Australia

fYear

2010

fDate

7-10 Dec. 2010

Firstpage

25

Lastpage

30

Abstract

An implementation of the nonlinear iterative partial least squares algorithm (NIPALS) was used as a test case for use of OpenCL for computation on a general purpose graphics processing unit (GPGPU) cluster using MPI. Timing results are shown along with results of a model of time required per iteration for defined problem sizes. Various steps in optimization of the code are discussed, moving from use of a single GPU, to multiple GPUs on a single node, to multiple GPUs on multiple nodes. Comparison of performance between OpenCL and BLAS implementations, modern CPU architectures and NVidia Tesla and Fermi class GPU systems are given.

Keywords

application program interfaces; computer graphic equipment; coprocessors; iterative methods; least squares approximations; principal component analysis; BLAS; CPU architectures; Fermi class GPU systems; GPU; MPI; NIPALS; NVidia Tesla; OpenCL API; general purpose graphics processing unit; nonlinear iterative partial least squares algorithm; optimization; principal component analysis; Algorithm design and analysis; Graphics processing unit; Instruction sets; Memory management; Pixel; Timing; Workstations; PCA; principal component analysis;

fLanguage

English

Publisher

ieee

Conference_Titel

e-Science Workshops, 2010 Sixth IEEE International Conference on

Conference_Location

Brisbane, QLD

Print_ISBN

978-1-4244-8988-6

Electronic_ISBN

978-0-7695-4295-9

Type

conf

DOI

10.1109/eScienceW.2010.14

Filename

5693137