The K Computer and its Application

Author

Yokokawa, M.

Author_Institution

Oper. & Comput. Syst. Div., RIKEN Adv. Inst. for Comput. Sci., Kobe, Japan

fYear

2012

fDate

5-7 Dec. 2012

Firstpage

21

Lastpage

22

Abstract

The K computer is a distributed memory super-computer system with 82,944 compute nodes and 5,184 I/O nodes that was jointly developed by RIKEN and Fujitsu as a Japanese national project. Its development began in 2006 and was completed in June, 2012. By achieving the LINPACK performance of 10.51 peta-FLOPS, the K computer ranked first for two consecutive TOP500 list in June and November 2011. During its adjustment, part of the K computer was served, gradually increasing computing resources, to experts in computational science as a trial and was used for performance optimization of users´ application codes.

Keywords

distributed memory systems; optimisation; parallel machines; Fujitsu; Japanese national project; K Computer; LINPACK performance; RIKEN; computational science; compute nodes; computing resources; distributed memory super-computer system; performance optimization; peta-FLOPS; Computer architecture; File systems; Large scale integration; Parallel processing; Scientific computing; Supercomputers; 10 peta-FLOPS; K computer; SPARCK64 VIIIfx; Supercomputer; Tofu interconnect network; six-demensional network;

fLanguage

English

Publisher

ieee

Conference_Titel

Networking and Computing (ICNC), 2012 Third International Conference on

Conference_Location

Okinawa

Print_ISBN

978-1-4673-4624-5

Type

conf

DOI

10.1109/ICNC.2012.13

Filename

6424539