Complex Architectures Used in Image Parallel Processing

Eng. Ştefan Mocanu, PhD
Prof. Eng.
Radu Dobrescu, PhD
Prof. Eng. Daniela Saru, PhD
PhD candidate, Junior Lect. Eng. Ramona Din
Eng. Andrei Grumăzescu

“Politehnica” University of Bucharest

Abstract: Training simulators, various design tools, science processes simulation and visualization, medical practice or „just” 3D games and animations, they all have in common the graphical card processing unit. Sustained by the technological progress and also due to the constant need for higher resolutions and speed for graphical applications which require HD (High Definition) performance, Graphical Processing Units (GPUs) are turning into increasingly complex architectures, sometimes even more powerful than the CPU. This paper presents Nvidia’s general purpose parallel computing architecture, code name CUDA, while a future article will develop case studies focused on the research of its potential in highly optimized parallel algorithms for image recognizing, classifying and processing.

Keywords: Graphical processing unit (GPU), CUDA, parallel computing, GPGPU, GUI, OpenCL.

View full text

Ştefan Mocanu, Radu Dobrescu, Daniela Saru, Ramona Din, Andrei Grumăzescu, Complex Architectures Used in Image Parallel Processing, Romanian Journal of Information Technology and Automatic Control, ISSN 1220-1758, vol. 20(1), pp. 97-105, 2010.