(405c) GPU-Based High Performance Molecular Dynamics Simulation of Highly Heterogeneous Molecular Systems: Application for Three-Phase Model of Polymer Electrolyte Membrane Fuel Cell

Ji Il Choi, Giuseppe F. Brunello, and Seung Soon Jang

Computational NanoBio Technology Laboratory, School of Materials Science and Engineering, Georgia Institute of Technology, 771 Ferst Drive NW, Atlanta, GA 30332-0245

Molecular dynamics (MD) simulation softwares such as NAMD, HOOMD, LAMMPS etc. have been successfully ported onto the GPU-based massively parallelized hardwares. In this study, we perform MD simulations of highly heterogeneous three-phase model for polymer electrolyte membrane fuel cell using GPU-version of LAMMPS in order to analyze the computational efficiency and scaling of the GPU-based parallelization for our specific model in which various type of materials such as polymer, water, proton, oxygen, platinum and graphite are integrated to describe the complicated interfacial system in polymer electrolyte membrane fuel cell system using multiple force fields such as Dreiding, F3C, EAM, and graphite force fields.

We found that our GPU-based MD simulation shows less performance for heterogeneous system in comparison to homogeneous system. Nevertheless, GPU packages in LAMMPS still show greater performance compared to CPU based cluster computations.