A high performance parallel algorithm for 1-D FFT

Author

Agarwal, R.C. ; Gustavson, F.G. ; Zubair, M.

Author_Institution

IBM Thomas J. Watson Res. Center, Yorktown Heights, NY, USA

fYear

1994

fDate

14-18 Nov 1994

Firstpage

34

Lastpage

40

Abstract

Proposes a parallel high-performance fast Fourier transform (FFT) algorithm based on a multi-dimensional formulation. We use this to solve a commonly encountered FFT based kernel on a distributed memory parallel machine, the IBM scalable parallel system, SP1. The kernel requires a forward FFT computation of an input sequence, multiplication of the transformed data by a coefficient array, and finally an inverse FFT computation of the resultant data. We show that the multi-dimensional formulation helps in reducing the communication costs and also improves the single node performance by effectively utilizing the memory system of the node. We implemented this kernel on the IBM SP1 and observed a performance of 1.25 GFLOPS on a 64-node machine

Keywords

IBM computers; arrays; distributed memory systems; fast Fourier transforms; mathematics computing; parallel algorithms; performance evaluation; 1.25 GFLOPS; 1D FFT; 64-node machine; FFT based kernel; IBM SP1; coefficient array; communication costs; distributed memory parallel machine; fast Fourier transform; forward FFT computation; high performance parallel algorithm; inverse FFT computation; multi-dimensional formulation; multiplication; scalable parallel system; single node performance; Costs; Distributed computing; Fast Fourier transforms; Fourier transforms; Kernel; Parallel algorithms; Parallel machines; Partial differential equations; Signal processing; Signal processing algorithms;

fLanguage

English

Publisher

ieee

Conference_Titel

Supercomputing '94., Proceedings

Conference_Location

Washington, DC

Print_ISBN

0-8186-6605-6

Type

conf

DOI

10.1109/SUPERC.1994.344263

Filename

344263