Analýza škálovatelnosti a přesnosti paralelních FFT algoritmů při násobení obrovských celých čísel

Abstract

This Thesis is focused on a technigue of multiplication of huge integers using polynomials, conversion of numbers to polynomials with different numerical bases and FFT multiplication, which is also compared with the normal multiplication. A reader will know FFT fundations and algorithms Radix-2, Radix-4, Split-Radix. There is also a comparison of various FFT implementation for multiplication of huge numbers including a scalability by parallel FFT libraries using shared and distributed memory and tested on Ostrava supercomputer. This Thesis meaning has been to point out alternative ways how to multiply huge integers not only because we are limited by number of digits in case of normal multiplication but also we are able to achieve results more effective.