A Many-core Machine Model for Designing Algorithms with Minimum Parallelism Overheads
Department of Computer Science, University of Western Ontario, ON, Canada N6A 5B7
@article{haquemany,
title={A Many-core Machine Model for Designing Algorithms with Minimum Parallelism Overheads},
author={Haque, Sardar Anisul and Maza, Marc Moreno and Xie, Ning}
}
We propose a model of computations which aims at capturing parallelism overheads (such as communication and synchronization costs) of programs written for modern GPU architectures. We establish a Graham-Brent theorem for this model so as to estimate running time of programs running on p streaming multiprocessors. We evaluate the benefits of our model with three applications. In each case, our model is used to optimize a program parameter controlling overhead.
April 15, 2013 by hgpu