Exploratory research on embedding CUDA code into hetrogeneous MP-SOC achitectures programmed with the Daedalus framework
Leiden Institute of Advanced Computer Science (LIACS), Leiden University, Niels Bohrweg 1, 2333 CA Leiden, The Netherlands
Leiden University, 2011
@article{van2011universiteit,
title={Exploratory research on embedding CUDA code into hetrogeneous MP-SOC achitectures programmed with the Daedalus framework},
author={van der Zwet, R.},
year={2011}
}
The objective of this Bachelor Thesis is to explore the possibilities of using NVIDIA CUDA enabled GPU Processors within the HDPC framework. The HDPC framework is one of the heterogeneous MP-SoC architectures programmed with the Daedalus framework. This paper will focus on the transfer overhead introduced by using the GPU and how to best cope with this introduced latency. In this study it was found that using the GPU instead of the CPU , the overall execution times will decrease if the execution pattern has specific characteristics with regards to token size and processing complexity.
January 23, 2012 by hgpu
