Increasing efficiency of the sequential algorithms programs execution using CUDA
Liberios Vokorokos, Miroslav Hartinger, Norbert Ádám, Eva Chovancová, Ján Radušovský · 2014
The ambition to achieve higher computing performance, is based on using all the advantages of parallel approach. The devices, software development kits and other technologies are leaping forward. Nowadays, they allow programmer to use more effective and optimized template libraries. They empower him to acces functions of graphics processing unit through the highlevel language. This paper details the experience of using such optimization on basic and more complex algorithms and the measurement of overall effectiveness increase. Our experimental results demonstrate the overall acceleration on various types of GPU.