Softpedia
 

NEWS CATEGORIES:



NEWS ARCHIVE >>
SOFTPEDIA REVIEWS >>
MEET THE EDITORS >>
Home > News > Technology and Gadgets > Systems

September 29th, 2010, 14:07 GMT · By

Fujitsu Plans 10 PetaFLOPS Japanese Supercomputer

SHARE:

Adjust text size:


Fujitsu plans to make a 10 PetaFLOP supercomputer
Enlarge picture
Fujitsu has announced its intention to create the so-called “K computer', a HPC system intended to have a performance about five times that of the Jaguar, the currently most powerful supercomputer in the world.

Since solving the mysteries of our planet and the universe always involves very high computing capabilities, supercomputers periodically get created.

One of the latest projects was announced just a short time ago by Fujitsu, whose master plan is to create a supercomputer capable of reaching 10 PetaFLOPS.

Set to be a central part of the High-Performance Computing Infrastructure (HPCI) initiative, the conglomerate will be developed jointly with RIKEN, an independent organization under MEXT.

MEXT stands for Japan's ministry of education, culture, sports, science and technology, to whom belongs the HPCI initiative itself.

The system will be made up of over 800 computer racks and all of them will employ the company's SPARC64 VIIIfx central processing units.

Each such CPU has a computational performance of 128 gigaFLOPS, and over 80,000 such processors will be part of an interconnected network.

Pairing that with a world-class processing power of 2.2 gigaFLOPS per watt, that leaves the goal system performance at 10 PetaFLOPS, much higher than the 1.759 PetaFLOPS that Jaguar can brag about.

For those interested in more details, the supercomputer will use the so-called world's first six-dimensional mesh-torus topology that Fujitsu itself developed.

Basically, this will let multiple processors be flexibly allocated to groups, thus improving the efficiency of the installation.

As for heat management, water cooling will be employed for the processors and other hot components, leading to a higher mounting density, reduced failure rate and longer component life.

TELL US WHAT YOU THINK:

1,319 hits · Link to this article · Print article · Send to friend · Subscribe to news

MUST-READ RELATED ARTICLES:


Intel MIC Accelerators Get Detailed

Intel and HPC-Maker SGI Also Get Involved with DARPA

PhotoFast's Power Drive-LSI SSD Works at 1,500 MB/s

Analysts Say NVIDIA Needs x86 Technology in the Long Run

NVIDIA CUDA x86 Lets AMD and Intel CPUs Accelerate CUDA Applications

READER COMMENTS:



No user comments yet.
Be the first to express your opinion!
Copyright © 2001-2012 Softpedia. Contact/Tip us at

WindowsGamesDriversMacLinuxScriptsMobileHandheldNews

SUBMIT PROGRAM   |   ADVERTISE   |   GET HELP   |   SEND US FEEDBACK   |   RSS FEEDS   |   UPDATE YOUR SOFTWARE   |   ROMANIAN FORUM