Tag: ‘Supercomputer’

Pi calculated to a record 2.5 trillion decimals

18 Aug 2009

Pi --

Researchers in Japan have calculated pi to over 2.5 trillion decimal places, more than double the previous world record set in 2002.

Led by University of Tsukuba professor Daisuke Takahashi, the research team performed the calculation using a massive parallel processing (MPP) supercomputer called the T2K Tsukuba System, which consists of 640 high-performance computers clustered together to achieve processing speeds of 95 teraflops (95 trillion floating-point operations per second). The supercomputer calculated pi to 2,576,980,377,524 decimal places in 73 hours 36 minutes.

By comparison, it took the previous record holders about 600 hours to perform their calculation (over 8 times longer than it took the T2K Tsukuba System). The previous record was set in 2002 when researchers from Hitachi and the University of Tokyo calculated pi to a little over 1.2 trillion decimal places.

The University of Tsukuba researchers, whose stated primary objective was to test the reliability and speed of their supercomputer, have submitted their results to the Guinness Book of World Records for official recognition.

[Sources: Asahi, Yomiuri]

Petaflops-level supercomputer to be unveiled

20 Jun 2006

MDGRAPE-3

On June 19, Japan's Institute of Physical and Chemical Research (Riken), SGI Japan and Intel announced the development of a supercomputer with a theoretical peak performance of 1 petaflops (one million billion floating point operations per second). Known as the MDGRAPE-3 (or the Protein Explorer), the computer system is designed to perform molecular dynamics simulations of such phenomena as non-bonding interactions between atoms.

The system consists of 201 units equipped with 24 of RIKEN?s MDGRAPE-3 LSI chips for molecular dynamics simulation (total of 4,808 chips), which are connected to 64 parallel servers equipped with 256 of Intel's Xeon 5000-series cores and 37 parallel servers equipped with 74 Xeon 3.2 GHz cores.

In the future, RIKEN plans to further upgrade the system with Xeon 5100-series processors (codenamed Woodcrest), and testing is now underway.

The LINPACK Benchmark, which is the standard for the Top 500 List, could not be performed on the system, so the performance cannot be compared directly with the world's other top supercomputers. However, the system's theoretical peak performance of 1 petaflops will set the computer firmly at the top of the list, with a speed about three times that of IBM's BlueGene/L at Lawrence Livermore National Laboratory (currently No.1 on the list).

The system will be unveiled to the public on June 24 at RIKEN's Yokohama laboratory.

[Source: IT Media]