Regensburg 2013 – wissenschaftliches Programm
Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
CPP: Fachverband Chemische Physik und Polymerphysik
CPP 18: Poster: Polymer Dynamics
CPP 18.16: Poster
Dienstag, 12. März 2013, 18:15–20:15, Poster C
POCLMD: A Flexible GPU Accelerated Molecular Dynamics Code — •Chuanfu Luo1 and Jens-Uwe Sommer1,2 — 1Leibniz-Institut für Polymerforschung Dresden, Germany — 2Technische Universität Dresden, Germany
Recently, general-purpose computing on Graphics Processing Units (GPU) or coprocessor cards becomes a trend in high performance computing. Many newly built super-computers are equipped with powerful GPUs or coprocessors. Currently, several companies provide high performance accelerating cards and all of them support the open standard OpenCL programing. The code writing in OpenCL can run on multiple-core CPUs, GPUs, and coprocessors without any changes. POCLMD is designed to be a fast, flexible and easy-to-use MD code, which is programed in Python and OpenCL through PyOpenCL [1]. The kernel code written in OpenCL is compiled and loaded on the fly during the running, thus the executing kernel of POCLMD is very small and efficient. Benefited by the powerful template programing of Python, POCLMD can support an easy and flexible way to add some custom potentials without asking users to modify the codes. The first benchmark of LJ liquids shows that the performance of POCLMD can catch up 90% of HOOMD-blue [2] (the fastest CUDA based MD code) on the same NVIDIA's GPUs. The best performance is achieved on an AMD's HD7970 card, on which POCLMD runs over 80 times faster than LAMMPS on a single core Q6600 CPU (2.4GHz). [1] http://mathema.tician.de/software/pyopencl [2] http://codeblue.umich.edu/hoomd-blue/