The supercomputer, delivering over 1.5 Petaflops, will provide the computer modeling and simulation capabilities needed for the ‘Broader Approach’ program, linked to the ITER initiative.
On March 19, Bull, the CEA, F4E and the JAEA – in the presence of representatives from the EU and the French Embassy in Japan, as well as numerous Japanese politicians (including MPs, the Governor of Aomori province and the mayor of Rokkasho) – have inaugurated the Helios supercomputer in Rokkasho (Japan): one of the world’s most powerful supercomputers, delivering peak performance of over 1.5 Petaflops. This is the third Petascale supercomputer designed and developed by Bull to go into operational production in the past 18 months.
In April 2011, the CEA (the French Alternative Energies and Atomic Energy Commission), acting on behalf of F4E (Fusion For Energy), chose Bull to equip the new data center being built at Rokkasho in Japan, under the auspices of the International Fusion Energy Research Center (IFERC). The data center is one of the key components of the ‘Broader Approach’; a research program designed to complement ITER and launched in November 2006 as part of a cooperation between Japan and Europe. F4E coordinates the European contribution to the program, while JAEA (the Japan Atomic Energy Agency) is responsible for the Japanese contribution.
Marking this occasion, Philippe Vannier, Bull’s Chairman and CEO, commented: “I am extremely proud that Bull is involved in this very large-scale project aimed at perfecting the use of this sustainable and environmentally-friendly energy source. Our teams have successfully demonstrated their technological and logistical prowess in installing this Petascale supercomputer in under six months, in an area affected by the tsunami a year ago. In the face of extremely difficult circumstances, our Japanese colleagues have demonstrated remarkable professionalism.”
High-density power for research
The new supercomputer has been designed to operate around the clock. Its peak power of over 1.5 Petaflops means that it ranks among the world’s most powerful systems. When it comes to computing power, it features 4,410 bullx® B510 processing nodes arranged in a cluster architecture, incorporating 8,820 Intel® Xeon® E5-2600 processors, with a total of 70,560 processing cores. Helios has a 280 Terabyte memory and a high-speed, 5.7 Petabyte storage system backed up by a secondary storage system designed to support up to 50 Petabytes. The interconnect network within the cluster is based on InfiniBand® technology. Supporting the processing nodes, six bullx® S Series and 80 bullx® R Series systems are used to administer the supercomputer’s operations: from managing the Lustre® file management system to user access. Bull has also supplied 32 bullx® R425 systems, featuring high-performance graphics cards, for pre/post-processing and visualization. The supercomputer is equipped with the Bull Supercomputer Suite Advanced Edition software, developed and optimized by Bull for Petaflops-class systems based on the Linux® operating system and featuring numerous Open Source components.
A wide range of services: from designing and fitting out computer suites to operational running
As part of this project, Bull has been responsible for designing and delivering the electrical infrastructure and chilled water supply for the liquid cooling systems used at the data center, with the physical infrastructure (the building, transformers, cooling units, etc) having been supplied by Japan. Bull has also installed the supercomputer, and will be looking after its maintenance and operation for a five-year period.
When it comes to delivering these services, Bull is drawing on the skills and resources of its local partner, SGI Japan, Ltd.
 The other main components of the R&D program are: the Tokamak JT-60SA, installed at Naka (Japan) and prototypes of future neutron sources at IFMIF (the International Fusion Materials Irradiation Facility) dedicated to studying fusion materials, including the accelerator.