Hybrid MPI/OpenMP optimization in Linpack benchmark on multi-core platforms

Author

Feng Wang ; Canqun Yang ; Juncheng Bai

Author_Institution

Sch. of Comput. Sci., Nat. Univ. of Defense Technol., Changsha, China

fYear

2013

fDate

26-28 April 2013

Firstpage

917

Lastpage

920

Abstract

With the increasing of the number of CPU cores, thousands of cores are used in the current supercomputers. The MPIJOpenMP hybrid programming model is popular in multicore systems. Some serial codes in the pure MPI programs turn to bottleneck and are easy to be neglected when these codes are ported to the MPIJOpenMP hybrid model. In the Linpack benchmark, we focus on the local swap algorithm and present an OpenMP optimization method to speedup the performance using multi-thread. On a cluster system with 36 multi-core CPUs, experiment results show that this method can decrease the time of lacal swap from 271.11 seconds to 202.53 seconds with the problem size 240, 000 and improve 5.23% of the whole Linpack program from 5.01 TeraFLOPS to 5.28 TeraFLOPS.

Keywords

application program interfaces; benchmark testing; mainframes; message passing; multi-threading; multiprocessing systems; pattern clustering; CPU cores; Linpack benchmark; Linpack program; MPI programs; MPI-OpenMP hybrid model; MPI-OpenMP hybrid programming model; TeraFLOPS; cluster system; hybrid MPI-OpenMP optimization method; local swap algorithm; multicore CPUs; multicore platforms; multicore systems; multithreading; supercomputers; Computers; Niobium; Programming; Standards; Cluster; Linpack; MPI; Multi-core; OpenMP;

fLanguage

English

Publisher

ieee

Conference_Titel

Computer Science & Education (ICCSE), 2013 8th International Conference on

Conference_Location

Colombo

Print_ISBN

978-1-4673-4464-7

Type

conf

DOI

10.1109/ICCSE.2013.6554042

Filename

6554042