DocumentCode :
1829202
Title :
Performance of Large Scale MHD Simulation of Global Planetary Magnetosphere with Massively Parallel Scalar Type Supercomputer Including Post Processing
Author :
Fukazawa, Keiichiro ; Nanri, Takeshi
Author_Institution :
Res. Inst. for Inf. Technol., Kyushu Univ., Fukuoka, Japan
fYear :
2012
fDate :
25-27 June 2012
Firstpage :
976
Lastpage :
982
Abstract :
To use the massively parallel scalar type supercomputer for the large scale numerical magnetospheric simulation, the computational performance of magneto-hydro-dynamic (MHD) code is evaluated. We have made performance tuning of a three-dimensional MHD code for space plasma simulations on the HA8000 with 1024 cores at the University of Tokyo. For parallelization of the MHD code, we use four different methods, i.e., regular one-dimensional, two-dimensional, three-dimensional domain decomposition methods and a cache-hit type of three-dimensional domain decomposition method. We found that the regular two and three-dimensional decompositions are suitable for HA8000 system. In addition we found the three problems to run the large scale simulation, data I/O, storage system and post processing. To improve them, we can simulate the high resolution Saturn´s magnetosphere.
Keywords :
Saturn; astronomical techniques; astronomy computing; magnetohydrodynamics; parallel machines; planetary atmospheres; plasma simulation; virtual machines; HA8000 system; University of Tokyo; cache-hit type; computational performance; data I/O; global planetary magnetosphere; high resolution Saturn magnetosphere; large scale 3D MHD simulation; large scale numerical magnetospheric simulation; magnetohydrodynamic code; massively parallel scalar type supercomputer; performance tuning; post processing; regular 1D domain decomposition method; regular 2D domain decomposition method; regular 3D domain decomposition method; space plasma simulations; storage system; Arrays; Computational modeling; Computers; Equations; Magnetohydrodynamics; Mathematical model; Plasmas; MHD simulation; high performance computing; massively parallel computing; performance evaluation;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
High Performance Computing and Communication & 2012 IEEE 9th International Conference on Embedded Software and Systems (HPCC-ICESS), 2012 IEEE 14th International Conference on
Conference_Location :
Liverpool
Print_ISBN :
978-1-4673-2164-8
Type :
conf
DOI :
10.1109/HPCC.2012.142
Filename :
6332278
Link To Document :
بازگشت