2 Nov 2011 Tokyo - The "K computer", which is a supercomputer currently under the joint development of RIKEN and Fujitsu, has achieved a LINPACK benchmark performance of 10.51 Petaflop/s (10.51 quadrillion floating point operations per second).
The supercomputer system used for these tests is in its final configuration of 864 racks, comprising a total of 88,128 interconnected CPUs. With an execution efficiency of 93.2%, this system further exceeds its first place winning performance on the 37th TOP500 list international ranking of supercomputers published in June 2011, where it scored 93.0%.
Although development is still underway to adjust the system software, the K computer achieved the goal of a LINPACK score of 10 petaflops, a performance target set as a national core technology as part of Japan's 3rd Science and Technology Basic Plan.
RIKEN and Fujitsu have been working together to develop the K computer, with the aim of beginning shared use by November 2012, as a part of the High-Performance Computing Infrastructure (HPCI) initiative led by Japan's Ministry of Education, Culture, Sports, Science and Technology (MEXT). The K computer brings together leading-edge technologies, including ultrafast and energy-efficient CPUs and a network capable of an immense amount of interconnectivity, to create a high-performance, highly-reliable supercomputer.
In August 2011, the last of 864 racks was delivered and installed, bringing the system to its final configuration. For the purpose of conducting point basic operating tests and design-performance checks, the system was tested on October 7-8, using the LINPACK benchmarking program and measured processing speeds of 10.51 petaflops and 93.2% operating efficiency. These results have been submitted to the 38th TOP500 list, which will be presented at the International Conference for High Performance Computing, Networking, Storage and Analysis (SC11), being held in Seattle from November 12 to 18, 2011.
The central component of the K computer is its 88,128 CPUs, arrayed over 864 racks, which have a theoretical calculation speed of 11.28 petaflops. When the LINPACK benchmark programme measured the K computer in its final configuration, the supercomputer system achieved a speed of 10.51 petaflops exceeding its previous performance speed of 8.162 petaflops, which earned the system first place on the previous TOP500 list - published June 2011.
According to the latest LINPACK performance measurements, not only did the K computer excel on performance benchmarks, achieving its goal of 10 petaflops and exceeding its previous operating efficiency results, but it also demonstrated an extraordinary level of stability for one of the world's largest-scale systems, as the overall system comprised of 88,128 CPUs ran without a single failure for 29 hours and 28 minutes.
Work is still underway to further develop and adjust the operating system, compiler, and other system software before the June 2012 target date. In order to generate breakthroughs as quickly as possible in fields designated by the government of Japan as having strategic importance, a part of the K computer has been made available for those researchers participating in the "Grand Challenge" application software development project(4) and the HPCI strategic programmes as an environment for early access starting April 2011. While helping as much as possible to meet the needs of these researchers, the test environment has gradually expanded as work progresses on the K computer.
As the K computer represents a dramatic leap forward in terms of simulation precision and calculation speed, it can be applied to a variety of fields that use computational science. It holds the promise of contributing to the generation of world-class breakthroughs such as:
Having reached its performance target, the K computer is quickly moving toward completion. RIKEN and Fujitsu will be focusing their energies on developing and assessing system software for the large-scale K computer with the aims of system completion in June 2012 and the official beginning of its shared use in November 2012.
Ryoji Noyori, President, RIKEN, stated: "The K computer is a key national technology that will help lay the foundation for Japan's further progress. As such, I am delighted that it has achieved its major objective - a LINPACK performance of 10 petaflops - thus demonstrating our strong technical power. I would like to express my deep gratitude to the Fujitsu team members, who have devoted themselves wholeheartedly to building the system, and everybody else involved in the project for their extraordinary effort. At the same time, the system's advanced technical capabilities are a major source of pride. As we move forward to completing the system next June, we are committed to putting the K computer's ground-breaking computing performance of 10 petaflops to use to generate important results, as well as making it available for shared use by many researchers starting next fall."
Masami Yamamoto, President, Fujitsu Limited, stated: "I am thrilled that the K computer has achieved a LINPACK benchmark performance of 10 petaflops, while still being in development. I am also proud that we are one step closer to our goal of building a world-class supercomputer. At Fujitsu, we will continue to work tirelessly to deliver the K computer and its system software by June 2012, and enable it to be used by a large number of researchers throughout the world. We truly hope that the K computer's world-class performance will benefit and deliver value to both Japan and the entire world with leading-edge computing technology."