Presenter/Author Information

Jiye Zeng
Tsuneo Matsunaga
Hitoshi Mukai

Keywords

lagrangian particle dispersion model, nvidia gpu, parallel computing

Start Date

1-7-2010 12:00 AM

Abstract

A Lagrangian particle dispersion model computes trajectories for a large number of particles and is often used for describing the transport and diffusion of tracers in the atmosphere. In a typical application, computing the position changes of particles is the bottleneck that controls the total execution time. Such a model presents a good case for investigating the use of NVIDIA’s Graphic Processor Unit or GPU. We developed a Lagrangian particle dispersion model based on FLEXPART to take the advantage of the GPU’s parallel computing power. Preliminary tests show that using a GPU can improve the performance by as much as 10 times.

COinS
 
Jul 1st, 12:00 AM

Using NVIDIA GPU for Modelling the Lagrangian Particle Dispersion in the Atmosphere

A Lagrangian particle dispersion model computes trajectories for a large number of particles and is often used for describing the transport and diffusion of tracers in the atmosphere. In a typical application, computing the position changes of particles is the bottleneck that controls the total execution time. Such a model presents a good case for investigating the use of NVIDIA’s Graphic Processor Unit or GPU. We developed a Lagrangian particle dispersion model based on FLEXPART to take the advantage of the GPU’s parallel computing power. Preliminary tests show that using a GPU can improve the performance by as much as 10 times.