Astrophysical particle simulations with large custom GPU clusters on three continents
Astrophysical particle simulations with large custom GPU clusters on three continents
复制标题
DOI：
10.1007/s00450-011-0173-1
复制
发表时间：
2011-06
期刊：
影响因子：
--
通讯作者：
R. Spurzem;P. Berczik;I. Berentzen;Keigo Nitadori;T. Hamada;G. M. Martinez;A. Kugel;R. Männer;J. Fiestas;R. Banerjee;R. Klessen
中科院分区：
文献类型：
--
作者：
R. Spurzem;P. Berczik;I. Berentzen;Keigo Nitadori;T. Hamada;G. M. Martinez;A. Kugel;R. Männer;J. Fiestas;R. Banerjee;R. Klessen
我们提出了直接的天体物理学N体模拟高达600万机构使用我们的并行MPI-CUDA代码在北京，伯克利和海德堡的大型GPU集群，不同种类的GPU硬件。这些集群在ICCS（国际计算科学中心）的合作下相互联系。我们达到约三分之一的峰值性能为这个代码，在一个真实的应用程序的情况下，分层块时间步长和核心晕密度结构的恒星系统。该程序和硬件被用来模拟密集的星星集群与许多双星和星系核与超大质量黑洞，其中遥远的粒子之间的相关性不能忽略。
We present direct astrophysicalN-body simulations with up to six million bodies using our parallel MPI-CUDA code on large GPU clusters in Beijing, Berkeley, and Heidelberg, with different kinds of GPU hardware. The clusters are linked in the cooperation of ICCS (International Center for Computational Science). We reach about one third of the peak performance for this code, in a real application scenario with hierarchically block time-steps and a core-halo density structure of the stellar system. The code and hardware is used to simulate dense star clusters with many binaries and galactic nuclei with supermassive black holes, in which correlations between distant particles cannot be neglected.