Java on CUDA architecture
Department of Software Engineering, Faculty of Information Technology, Czech Technical University in Prague
Czech Technical University in Prague, 2012
@article{strnad2012java,
title={Java on CUDA architecture},
author={Strnad, J.},
year={2012}
}
Traditional CPU is able to run only a few complex threads concurrently. On the other side, a GPU allows a concurrent execution of hundreds or thousands of simpler threads. The GPU was originally designed for a computer graphics, but nowadays it is being used for general-purpose calculations using a GPGPU technology. CUDA, one of the GPGPU technologies, is introduced in the thesis. Methods of cooperation between this technology and Java language are presented. Several algorithms from areas of linear algebra and image processing were implemented using JCuda library. The measured execution speed-up for matrix multiplication was 400 times over CPU. The execution speed-up for Canny edge detection varied between 5 to 30 times.
May 24, 2012 by hgpu
