Accelerating a three-dimensional MOC calculation using GPU with CUDA and two-level GCMFD method

被引：15

作者：

Zhang, ZhiZhu ^{[1
,2
]}

Wang, Kan ^{[1
]}

Li, Qing ^{[2
]}

机构：

[1] Tsinghua Univ, Dept Engn Phys, Beijing 100084, Peoples R China

[2] Nucl Power Inst China, Sci & Technol Reactor Syst Design Technol Lab, Chengdu 610041, Peoples R China

来源：

ANNALS OF NUCLEAR ENERGY | 2013年 / 62卷

关键词：

MOC; GPU; CUDA; TCM; Two-level GCMFD acceleration;

D O I：

10.1016/j.anucene.2013.06.039

中图分类号：

TL [原子能技术]; O571 [原子核物理学];

学科分类号：

0827 ; 082701 ;

摘要：

The method of characteristics (MOC) is a promising method that can solve neutron transport equation in three-dimensional complex geometry accurately. But the extremely long computation time limits its application. In order to make it acceptable, we studied the acceleration of a three-dimensional MOC code, TCM, with Graphics Processing Unit (GPU) computing which is considered as one of the most high performance computing techniques. With the application of Computer Unified Device Architecture (CUDA) architecture, the conversion from Central Processing Unit (CPU) to GPU is much simpler. Furthermore, the two-level General Coarse Mesh Finite Difference (GCMFD) acceleration method is employed to accelerate the MOC code. Numerical results show that both GPU and two-level GCMFD acceleration are effective for reducing the computational time. (C) 2013 Elsevier Ltd. All rights reserved.

引用

页码：445 / 451

页数：7