| Title | Optimizacija metode za kompresiju slikovnih podataka bez gubitaka pomoću grafičkog podsustava |
| Author | Luka Strižić |
| Mentor(s) | Josip Knezović (thesis advisor)
|
| Abstract | Energetska učinkovitost postaje sve važnije u potrošačkoj elektronici, zato je cilj ovog rada usporediti energetsku učinkovitost, kao i vrijeme izvršavanja, CBP implementacije na procesoru i grafičkom podsutavu, koristeći CUDA-u za potonje. Testovi su provedeni koristeći komponente srednje klase u stolnom računalu i komponente visoke klase u serveru. One su uključivale Intel i3-4130, Intel Xeon E5-2620, nVidia GTX 750 Ti i nVidia GTX 980. CUDA verzija se pokazala kao energetski učinkovitija i brža nasprem jednodretvene procesorske izvedbe, međutim, potrebna su daljnja testiranja koja bi usporedila optimiziranu CUDA i višedretvenu procesorsku implementaciju, da se pokrije cijeli spektar. U svakom slučaju, hardver srednje klase je energetski učinkovitiji od hardvera visoke klase.
|
| Keywords | General-Purpose Computing on Graphics Processing Units GPGPU Compute Unified Device Architecture CUDA Classification and Blending Prediction CBP Lossless Image Compresion |
| Parallel title (English) | Optimization of Lossless Image Compression Method for GPGPU |
| Committee Members | Josip Knezović (committee chairperson)
Igor Čavrak (committee member)
Ivana Bosnić (committee member)
|
| Granter | University of Zagreb
Faculty of Electrical Engineering and Computing |
| Place | Zagreb |
| State | Croatia |
| Scientific field, discipline, subdiscipline | TECHNICAL SCIENCES
Computing
|
| Study programme type | university |
| Study level | undergraduate |
| Study programme | Electrical Engineering and Information Technology |
| Academic title abbreviation | univ. bacc. ing. el. techn. inf. |
| Genre | undergraduate thesis |
| Language | English |
| Defense date | 2015-07-13 |
| Parallel abstract (English) | Power efficiency is becoming increasingly important in consumer electronics, which is why the aim of this thesis was to compare power efficiency, as well as execution time, of a Classification and Blending Prediction implementation on a CPU and a GPU, using CUDA for the latter. Test were conducted using mainstream components in a desktop PC and high end components in a server. They included Intel i3-4130, Intel Xeon E5-2620, nVidia GTX 750 Ti and nVidia GTX 980. The CUDA version proved to be both more efficient and faster than a singlethreaded CPU version, however, further tests should be done comparing optimized CUDA versus multithreaded CPU implementation to cover the whole spectrum. In any case, mainstream hardware is more energy efficient than high end hardware. |
| Parallel keywords (Croatian) | računalstvo opće namjene na grafičkom podsusavu GPGPU Compute Unified Device Architecture CUDA Classification and Blending Prediction CBP kompresija slika bey gubitaka |
| Resource type | text |
| Access condition | No public access |
| Terms of use |  |
| URN:NBN | https://urn.nsk.hr/urn:nbn:hr:168:971276 |
