18932

Temporospatial Epidemic Simulations Using Heterogeneous Computing

Dhananjai M. Rao
CSE Department, Miami University, Oxford, OH 45056, USA
32nd European Modelling and Simulation Conference (ESM’18), 2018

@article{raotemporospatial,

   title={Temporospatial Epidemic Simulations Using Heterogeneous Computing},

   author={Rao, Dhananjai M},

   year={2018}

}

Discrete Event Simulation (DES) is widely used for analysis of complex temporospatial epidemic models. In such simulations, a conspicuous fraction (50%-90%) of simulation runtime is typically spent in solving equations used to model epidemic progression. General Purpose Graphics Processing Units (GPGPUs) hold considerable potential to reduce time for solving epidemic equations. However, the significant differences in hardware and programming models of GPGPUs and CPUs hinder their effective use, particularly by epidemiologists and public health experts. Consequently, we have developed an epidemic modeling and simulation environment called MUSE-HC. In MUSE-HC, discrete event processing is performed on the CPU while epidemic equation processing is performed on a GPGPU. MUSE-HC provides a domain-specific modeling language called Epidemic Description Language (EDL) to streamline modeling for noncomputing experts. The EDL model description is compiled and transformed to heterogeneous computing source code based on OpenCL. The generated code is compiled and executed on a workstation equipped with a GPGPU for simulation and analyses. Our experiments conducted using synthetic benchmarks show that our heterogeneous approach can improve simulation performance by up to 16x for certain temporospatial epidemic models.
No votes yet.
Please wait...

Recent source codes

* * *

* * *

HGPU group © 2010-2019 hgpu.org

All rights belong to the respective authors

Contact us: