Two Approaches to Particle Simulation: OpenMPI and CUDA
Department of Computer Science and Engineering, University of California, San Diego
University of California, CSE260 Project Report, 2012
@article{seime2012two,
title={Two Approaches to Particle Simulation: OpenMPI and CUDA},
author={Seime, Bjorn Christian and Mannino, Matteo and Runde, Terje},
year={2012}
}
In this project, our goal is to make particle simulation run in parallel using both CUDA and MPI implementations. The scale of parallelism between the are two vastly different, and we wish to compare implementations to gain insight on the strengths and weaknesses between the two different paradigms of parallelization. Particle simulations are of huge interest in many scientific fields. Whether each particle represents an ion, a billiard ball or even entire galaxies, the structure of the problem is similar: Each particle in a system has initial position, velocity and acceleration given by the simulation characteristics, and affect each other with some gravitational or repulsive force. How to predict the motion of these particles with a discrete timestep is known as N-body simulations. [1] It is difficult to solve and justify the correctness of such simulations through pure data. Actually, for a particle count greater than 3 it is known to be impossible [2], and that is why approximate simulations are of particular interest. Visualization of simulations helps researchers explore the behaviour and find evidence for physical models from the real world.
January 15, 2013 by hgpu
