Authors: Gajić, Dušan
Stanković, Radomir 
Title: Computation of the vilenkin-Chrestenson transform on a GPU
Journal: Journal of Multiple-Valued Logic and Soft Computing
Volume: 24
Issue: 1-4
First page: 317
Last page: 340
Issue Date: 1-Jan-2014
Rank: M21a
ISSN: 1542-3980
The Vilenkin-Chrestenson transform on finite Abelian groups is a useful mathematical tool for the analysis, synthesis, and optimization of multiple-valued functions. This paper proposes techniques for the efficient computation of the Vilenkin-Chrestenson transform using graphics processing units (GPUs). The development of the method is motivated by certain computationally demanding problems in multiplevalued logic (MVL), such as the spectral analysis of mosaics and the design and analysis of MVL circuits. The paper presents mappings of two distinct fast Fourier transform (FFT)-like algorithms, the Cooley- Tukey and the constant geometry algorithms, to the GPU computing model. The proposed solution implements each of the algorithms through a single kernel which permits the computation of the Vilenkin- Chrestenson spectrum of a p-valued function for an arbitrary value of p. The paper also discusses GPU implementation issues specific for the considered algorithms, such as their computational requirements, memory optimizations, and the use of compiler options in overcoming certain restrictions in GPU programming. Experimental results are included in order to verify the validity of the approach and examine its potential for applications in MVL and other areas.
Keywords: Fourier transform | GPGPU | GPU computing | Multiple-valued logic | Spectral methods | Vilenkin-Chrestenson transform
Publisher: Old City Publishing

Show full item record


checked on May 24, 2024

Page view(s)

checked on May 9, 2024

Google ScholarTM


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.