Up to 700k GPU cores, Kepler, and the Exascale future for simulations of star clusters around black holes
National Astronomical Observatories, Chinese Academy of Sciences, 20A Datun Rd., Chaoyang Distr., Beijing 100012, P.R. China
arXiv:1312.1789 [astro-ph.IM], (6 Dec 2013)
@article{2013arXiv1312.1789B,
author={Berczik}, P. and {Spurzem}, R. and {Wang}, L. and {Zhong}, S. and {Veles}, O. and {Zinchenko}, I. and {Huang}, S. and {Tsai}, M. and {Kennedy}, G. and {Li}, S. and {Naso}, L. and {Li}, C.},
title={"{Up to 700k GPU cores, Kepler, and the Exascale future for simulations of star clusters around black holes}"},
journal={ArXiv e-prints},
archivePrefix={"arXiv"},
eprint={1312.1789},
primaryClass={"astro-ph.IM"},
keywords={Astrophysics – Instrumentation and Methods for Astrophysics},
year={2013},
month={dec},
adsurl={http://adsabs.harvard.edu/abs/2013arXiv1312.1789B},
adsnote={Provided by the SAO/NASA Astrophysics Data System}
}
We present direct astrophysical N-body simulations with up to a few million bodies using our parallel MPI/CUDA code on large GPU clusters in China, Ukraine and Germany, with different kinds of GPU hardware. These clusters are directly linked under the Chinese Academy of Sciences special GPU cluster program in the cooperation of ICCS (International Center for Computational Science). We reach about the half the peak Kepler K20 GPU performance for our phi-GPU code [2], in a real application scenario with individual hierarchically block time-steps with the high (4th, 6th and 8th) order Hermite integration schemes and a real core-halo density structure of the modeled stellar systems. The code and hardware are mainly used to simulate star clusters [23, 24] and galactic nuclei with supermassive black holes [20], in which correlations between distant particles cannot be neglected.
December 9, 2013 by hgpu
