Bikov, Dusan and Pashinska, Maria (2020) Parallel Programming Strategies for Computing Walsh Spectra of Boolean Functions. In: ICT Innovations 2020. Machine Learning and Applications. Communications in Computer and Information Science, 1316 . Springer International Publishing, Cham, pp. 138-152. ISBN 978-3-030-62098-1
Full text not available from this repository.Abstract
The utilization of all computational resources is significant to achieve efficient computation. In order to exploit the available computation resources, we combine two parallel programming models such as MPI and CUDA. Combining of these two programming models ensures usage of the whole computation resources available in one computer system (CPU and GPU). In this paper, we present a way to use the available parallel processing resources to their full potential utilizing different strategies and techniques regarding data transfer. We perform experimental computation of well know algorithm for computing Walsh spectra of Boolean functions by combining these two parallel programming models. Experiments are performed on two different class of parallel processing capability hardware. Randomly generated Boolean functions of fourteen, sixteen, eighteen and twenty variables represent the used data set for experiments evaluations. Performed experiments show how the growth of the data size results in gaining more parallelization and therefore accelerate the execution.
Item Type: | Book Section |
---|---|
Subjects: | Natural sciences > Computer and information sciences Natural sciences > Matematics |
Divisions: | Faculty of Computer Science |
Depositing User: | Dusan Bikov |
Date Deposited: | 09 Nov 2020 09:40 |
Last Modified: | 09 Nov 2020 09:40 |
URI: | https://eprints.ugd.edu.mk/id/eprint/26707 |
Actions (login required)
![]() |
View Item |