International Journal of Computer Applications
Foundation of Computer Science (FCS), NY, USA
|
Volume 176 - Issue 36 |
Published: Jul 2020 |
Authors: Paulo S. S. De Souza, Arthur F. Lorenzon, Marcelo C. Luizelli, Fabio D. Rossi |
![]() |
Paulo S. S. De Souza, Arthur F. Lorenzon, Marcelo C. Luizelli, Fabio D. Rossi . Evaluating Embedded GPUs Performance via Computer Vision Applications. International Journal of Computer Applications. 176, 36 (Jul 2020), 7-11. DOI=10.5120/ijca2020920518
@article{ 10.5120/ijca2020920518, author = { Paulo S. S. De Souza,Arthur F. Lorenzon,Marcelo C. Luizelli,Fabio D. Rossi }, title = { Evaluating Embedded GPUs Performance via Computer Vision Applications }, journal = { International Journal of Computer Applications }, year = { 2020 }, volume = { 176 }, number = { 36 }, pages = { 7-11 }, doi = { 10.5120/ijca2020920518 }, publisher = { Foundation of Computer Science (FCS), NY, USA } }
%0 Journal Article %D 2020 %A Paulo S. S. De Souza %A Arthur F. Lorenzon %A Marcelo C. Luizelli %A Fabio D. Rossi %T Evaluating Embedded GPUs Performance via Computer Vision Applications%T %J International Journal of Computer Applications %V 176 %N 36 %P 7-11 %R 10.5120/ijca2020920518 %I Foundation of Computer Science (FCS), NY, USA
Computer vision applications usually present significant demand for computing resources, which limit its usage on embedded systems, since such devices typically have limited processing capacity. In this sense, hybrid embedded architectures are becoming more popular by offering higher levels of parallelism through Graphics Processing Units (GPUs). Despite the similarities with generalpurpose architectures that already exploit the benefits of GPUs, this new kind of embedded devices presents some architectural singularities, such as differences in memory access bandwidth. In this paper, we present an evaluation, considering how much these differences affect GPUs’ gains in the context of embedded systems. The results show that, despite the architectural limitations, using such devices can lead to a speed-up of 8 times compared to traditional embedded systems processing data only on CPUs.