Reševanje ugank Sudoku z genetskim algoritmom na grafičnem procesorju

diplomsko delo

Rok Brumec (Author), Janez Brest (Mentor), Borko Bošković (Co-mentor)

Abstract

V diplomski nalogi smo implementirali genetski algoritem za reševanje ugank Sudoku in izboljšali njegovo učinkovitost z izvajanjem na grafični procesni enoti. Za implementacijo genetskega algoritma na grafični procesni enoti smo uporabili ogrodje OpenCL. V diplomski nalogi najprej predstavimo Sudoku in genetske algoritme ter na kratko opišemo osnove paralelnega programiranja na grafični procesni enoti in opišemo ogrodje OpenCL. Na koncu opišemo našo implementacijo postopka reševanja Sudoka z genetskim algoritmom na grafični procesni enoti in predstavimo rezultate. Izvedli smo primerjavo z rezultati avtorjev iz literature.

Keywords

Sudoku;genetski algoritmi;OpenCL;grafična procesna enota;diplomske naloge;

Data

Language:	Slovenian
Year of publishing:	2016
Typology:	2.11 - Undergraduate Thesis
Organization:	UM FERI - Faculty of Electrical Engineering and Computer Science
Publisher:	R. Brumec
UDC:	004.421:794.5(043.2)
COBISS:	20085270
Views:	1645
Downloads:	100
Average score:	0 (0 votes)
Metadata:

Other data

Secondary language:	English
Secondary title:	SOLVING SUDOKU PUZZLES USING A GENETIC ALGORITHM ON GRAPHICS PROCESSOR
Secondary abstract:	In this diploma we implement a genetic algorithm for solving Sudoku puzzles and improve its efficiency by running it on a graphics processing unit. For the implementation of our genetic algorithm on graphics processing unit the OpenCL framework was used. Firstly we describe Sudoku and genetic algorithms, then the basics of parallel computing on a graphics processing unit are portrayed. The implementation of our Sudoku solving genetic algorithm on a graphics processing unit is presented, and the obtained results are compared to those found in literature.
Secondary keywords:	Sudoku;genetic algorithm;OpecCL;parallel computing;graphics processing unit;
URN:	URN:SI:UM:
Type (COBISS):	Bachelor thesis/paper
Thesis comment:	Univ. v Mariboru, Fak. za elektrotehniko, računalništvo in informatiko, Računalništvo in informacijske tehnologije
Pages:	X, 34 f.
ID:	9165877