Title :
Performance evaluations of PC clusters for large scale numerical analyses
Author :
Honda, Kohei ; Furukawa, Tatsuya ; Fukumoto, Hisao ; Ohchi, Masashi
Author_Institution :
Dept. of Electr. & Electron. Eng., Saga Univ., Saga, Japan
Abstract :
The PC cluster is a system that is connected two or more PC on the network and uses them as one parallel computer. Conventional numerical analyses can be executed comparatively low-cost by using PC cluster. In the authors´ laboratory, the PC cluster has been constructed. The author has made a finite element analysis parallel program and evaluated it. However, there is a possibility that the performance of the PC cluster will depend on clustering software. In addition, it has become able to build a higher-performance cluster more cheaply by the rapid progress of hardware performance. Therefore the authors have constructed the PC cluster newly using two kinds of clustering software on Windows and Linux and compared both of them. Ten PCs which equipped dual core CPU were used for PC cluster. The HPL benchmark test for the present cluster system has been conducted to evaluate the performance. As a result, the processing speed has not improved in Windows but it has improved well in Linux.
Keywords :
computer network performance evaluation; finite element analysis; mathematics computing; parallel programming; software performance evaluation; workstation clusters; HPL benchmark test; Linux; PC cluster; Windows; clustering software; finite element analysis parallel program; large scale numerical analyses; parallel computer; performance evaluation; processing speed; Benchmark testing; Computer networks; Concurrent computing; Finite element methods; Hardware; Large-scale systems; Linux; Numerical analysis; Personal communication networks; Software performance; Industrial Applications-Information Management Systems; Others-Others; System and Information-Systems Theory and Engineering;
Conference_Titel :
ICCAS-SICE, 2009
Conference_Location :
Fukuoka
Print_ISBN :
978-4-907764-34-0
Electronic_ISBN :
978-4-907764-33-3