Detail publikačního výsledku

Optimizing Code For Graphics Processing Units

SCHWARZ, J.; POSPÍCHAL, P.

Originální název

Optimizing Code For Graphics Processing Units

Anglický název

Optimizing Code For Graphics Processing Units

Druh

Stať ve sborníku mimo WoS a Scopus

Originální abstrakt

Graphics processing units (GPUs) are suitable for acceleration of many difficult scientific tasks. Several high-level languages for GPU have become available such as CUDA and OpenCL and GPU technology is rapidly advancing. This paper presents methods for the code optimization for general purpose computation on GPUs (GPGPU). Architecture of the GPU is briefly introduced as well as experiments leading to conclusions about factors mostly affecting their performance. Paper is concluded with the aim of PhD thesis.

Anglický abstrakt

Graphics processing units (GPUs) are suitable for acceleration of many difficult scientific tasks. Several high-level languages for GPU have become available such as CUDA and OpenCL and GPU technology is rapidly advancing. This paper presents methods for the code optimization for general purpose computation on GPUs (GPGPU). Architecture of the GPU is briefly introduced as well as experiments leading to conclusions about factors mostly affecting their performance. Paper is concluded with the aim of PhD thesis.

Klíčová slova

GPU, CUDA, Speedup, OpenCL

Klíčová slova v angličtině

GPU, CUDA, Speedup, OpenCL

Autoři

SCHWARZ, J.; POSPÍCHAL, P.

Rok RIV

2012

Vydáno

15.09.2011

Nakladatel

Slovak University of Technology in Bratislava

Místo

Bratislava

ISBN

978-80-227-3552-0

Kniha

Počítačové architektury a diagnostika 2011

Strany od

79

Strany do

84

Strany počet

6

BibTex

@inproceedings{BUT76455,
  author="Josef {Schwarz} and Petr {Pospíchal}",
  title="Optimizing Code For Graphics Processing Units",
  booktitle="Počítačové architektury a diagnostika 2011",
  year="2011",
  pages="79--84",
  publisher="Slovak University of Technology in Bratislava",
  address="Bratislava",
  isbn="978-80-227-3552-0"
}