Presentation 2011-07-29
The proposal of performance tuning tool for Cell broadband Engine cluster
Tetsuya NAKAHAMA, Masahiro YAMADA, Masato YOSHIMI, Hideharu AMANO,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) The productivity of parallel programming has been received attention with the rapid advance of multi-core and many-core processors. Especially, the programming environment of an economical small or middle scale clusters using accelerators must be improved. Here, a tool for extending a parallel program for a single Cell Broadband Engine(Cell/B.E.) to that for small scale clusters with Cell/B.E. is proposed. Since the Cell APIs must be called in a certain order, the tool inserts directive and MPI library calls automatically. The developed tool successfully extends an application using Monte-Carlo method for Cell/B.E. clusters. When 16 nodes each of which uses 8 SPE are used, the performance of automatically generated code was 0.74 times that of manually optimized code. Although the proposed tool has a room of improvement on the distribution of job between PPE and SPE, it can generate the practical program only with a simple configuration file.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Cell Broadband Engine / Cell/B.E. cluster / source code conversion / auto tuning
Paper # CPSY2011-18
Date of Issue

Conference Information
Committee CPSY
Conference Date 2011/7/21(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Computer Systems (CPSY)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) The proposal of performance tuning tool for Cell broadband Engine cluster
Sub Title (in English)
Keyword(1) Cell Broadband Engine
Keyword(2) Cell/B.E. cluster
Keyword(3) source code conversion
Keyword(4) auto tuning
1st Author's Name Tetsuya NAKAHAMA
1st Author's Affiliation Graduate School of Science and Technology, Keio University()
2nd Author's Name Masahiro YAMADA
2nd Author's Affiliation Graduate School of Science and Technology, Keio University
3rd Author's Name Masato YOSHIMI
3rd Author's Affiliation Department of Intelligent Information Engineering and Science, Doshisha University
4th Author's Name Hideharu AMANO
4th Author's Affiliation Graduate School of Science and Technology, Keio University
Date 2011-07-29
Paper # CPSY2011-18
Volume (vol) vol.111
Number (no) 163
Page pp.pp.-
#Pages 6
Date of Issue