GPU Accelerated Stochastic Simulation
Department of Electrical Engineering and Computer Science, University of Tennessee, Knoxville, Tennessee, USA
Symposium on Application Accelerators in High Performance Computing, 2010
@article{jenkinsgpu,
title={GPU Accelerated Stochastic Simulation},
author={Jenkins, D.D. and Peterson, G.D.},
booktitle={Application Accelerators in High Performance Computing, 2010 Symposium, Papers},
year={2010}
}
Through computational methods, biologists are able learn more about molecular biology by building accurate models. These models represent and predict the reactions among species populations within a system. One popular method to develop predictive models is to use a stochastic, Monte Carlo method developed by Gillespie called the stochastic simulation algorithm (SSA). Since this algorithm is based on stochastic principles, large numbers of simulations are needed to provide quality statistical models of the species and their interactions, giving way to long runtimes for large systems. In this paper, we provide an implementation of SSA onto NIVIDA graphics processing units using CUDA to parallelize ensembles of simulations. With this implementation we are able to see up to 41.9x speedup over the best-known serial implementations.
February 16, 2011 by hgpu
