izpis_h1_title_alt

Stiskanje podatkov na grafični procesni enoti
ID Ciglarič, Tadej (Author), ID Lotrič, Uroš (Mentor) More about this mentor... This link opens in a new window

.pdfPDF - Presentation file, Download (289,50 KB)
MD5: 074A7D4F10E4812ED67FA2AAC2E60DB5
PID: 20.500.12556/rul/64ffac98-b850-4f35-b051-1b4b1ea73bfe

Abstract
Učinkoviti algoritmi za stiskanje podatkov so lahko dokaj počasni. Namen tega dela je poizkus paralelizacije kompresijskega algoritma za izvajanje na grafičnih karticah. Ker grafične kartice vsebujejo zmogljivo vzporedno računsko enoto, bi pričakovali, da je mogoče izvajanje algoritma močno pohitriti. Zato smo v tem delu pregledali algoritem deflate in obstoječe paralelne implementacije za izvajanje tega algoritma na grafičnih procesnih enotah. Implementirali smo sekvenčni algoritem in ga na dva načina paralelizirali z uporabo ogrodja OpenCL. Vse implementacije smo preizkusili na korpusu datotek za testiranje algoritmov za stiskanje podatkov. Primerjali smo rezultate naših implementacij z obstoječimi paralelnimi in sekvenčnimi implementacijami.

Language:Slovenian
Keywords:GPGPU, OpenCL, stiskanje podatkov, algoritem deflate, paralelizacija
Work type:Bachelor thesis/paper
Organization:FRI - Faculty of Computer and Information Science
Year:2016
PID:20.500.12556/RUL-91237 This link opens in a new window
Publication date in RUL:27.03.2017
Views:2392
Downloads:416
Metadata:XML DC-XML DC-RDF
:
Copy citation
Share:Bookmark and Share

Secondary language

Language:English
Title:Data Compression on Graphics Processing Unit
Abstract:
Efficient data compression algorithms can be slow. The purpose of this work is an attempt of efficient parallelization of compression algorithm for execution on graphics processing units. Since graphics processing units contain an efficient parallel computing unit, it is reasonable to expect speedup from such parallelization of the algorithm. This work contains an overview of deflate algorithm and its existing parallel implementations intended for graphics processing units. We sequentially implemented the algorithm and parallelized it in two different ways using OpenCL framework. The implementations were tested on a corpus of files, intended for testing of compression algorithms. We compared the results with existing sequential and parallel implementations.

Keywords:GPGPU, OpenCL, data compression, deflate algorithm, paralellization

Similar documents

Similar works from RUL:
Similar works from other Slovenian collections:

Back