Masivně paralelní implementace nedeterministických konečných automatů

Abstract

This thesis focuses on implementations of various algorithms using graphics cards with CUDA technology. We have explored processing of sparse matrices and execution of nondeterministic finite automata. From area of processing sparse matrices we have focused on multiplication of sparse matrix with dense vector, multiplication of two sparse matrices and computation of similarity matrix. We have found, that not all of the algorithms are suitable for CUDA technology. Finally we have experimentally verified all implemented algorithms.