PRESS RELEASE

Fujitsu announces Hokkaido University supercomputer win

Fujitsu has announced that it has received an order from the Information Initiative Center of Hokkaido University for an interdisciplinary, large-scale computing system with a theoretical peak performance of 4.0 petaflops, consisting of a supercomputer and a cloud system. The system is expected to begin operations in December 2018.

The system's supercomputer system is comprised of approximately 1,300 of Fujitsu's x86 servers, including the latest model, the Fujitsu Server PRIMERGY CX2550 M4. The cloud system is equipped with x86 servers installed at Hokkaido University as well as multiple remote locations throughout Japan, now from Hokkaido to Kyushu, that connects to the SINET5academic backbone network, making it a nationwide wide-area distributed cloud system. As such, it can be used for research relating to distributed systems, such as networks and the Internet of Things.

In addition to using the system for R&D relating to computational science and computer science, the Information Initiative Center of Hokkaido University will use it to provide computing resources to HPCI and JHPCN, thereby broadly providing the system to researchers throughout Japan, including those at private companies, and contributing to advancements in all kinds of fields. With Japan's top high-performance computing technology, Fujitsu will provide powerful support for the activities of the Information Initiative Center of Hokkaido University as well as other research endeavours in Japan.

Background to the Deployment of the New System

Hokkaido University, the largest university in Hokkaido, is a national university that is driving education and research. The Information Initiative Center of Hokkaido University provides an academic information platform as a national joint-use facility for academic research by university faculty members, graduate students, and researchers throughout Japan.

The Information Initiative Center of Hokkaido University currently operates the Hokkaido University Academic Cloud, which is comprised of an interdisciplinary large-scale computing system that includes a supercomputer system and a cloud system, and petabyte-class data science unified cloud storage.. To upgrade its interdisciplinary large-scale computing system to the Hokkaido University High-Performance Inter-Cloud, it is building both a supercomputer system that will improve the processing performance of the existing system by over 20 times, and a nationwide wide-area distributed cloud system that can also be used for research on clouds and networks. It will further advance uses and research in such existing areas as scientific and technological simulations, AI, big data, and data science, and will provide an environment that can meet an even broader range of needs as people are trained to use the large-scale computing system for groundbreaking research.

Features of the New System

Supercomputer

The new supercomputer system is a massively parallel computer system consisting of subsystem A, which is comprised of 1,004 PRIMERGY CX2550 M4 x86 servers that are equipped with the Intel® Xeon® Scalable family of processors, and subsystem B, which is comprised of 288 Fujitsu Server PRIMERGY CX1640 M1 x86 servers that are equipped with Intel® Xeon Phi™ processors. Each compute node is connected with Intel® Omni-Path Architecture, the latest high-speed interconnect, for highly parallel computing performance.

Cloud 

The new cloud system has 64 x86 servers, including the Fujitsu Server PRIMERGY RX2540 M4 with NVIDIA Tesla V100, the latest GPU computing card, deployed to Hokkaido University and seven more in remote sites in the Kanto, Kansai, and Kyushu regions of Japan. Connecting these servers to the SINET5 academic backbone network makes it a nationwide wide-area distributed cloud system that extends from Hokkaido to Kyushu.

The cloud platform uses the Mirantis Cloud Platform, with OpenStack at its core, delivering the latest research environment.

Designed to flexibly meet a wide variety of needs 

With Fujitsu Software Technical Computing Suite, Fujitsu's HPC middleware for large-scale systems, the system offers optimal controls and node occupancy for jobs, improving the convenience of a wide variety of anticipated operations. It also provides flexible electric power controls to conserve power by limiting the maximum electric power consumed by each user and turning the power for nodes on or off.

Comment from Dr Yoshiaki Takai, Director, Information Initiative Center, Hokkaido University

The Hokkaido University High-Performance Inter-Cloud is a transformational research promotion platform for advanced research that uses large-scale computational science simulations, artificial intelligence, big data, and the Internet of Things, leading to the creation of a super-smart society. The deployment of this system, which delivers the dramatically improved peak performance of 4 petaflops, will meet the expectations of universities and research institutions, as well as private sector researchers, throughout Japan, while promoting the further acceleration of academic research in Japan.

Company: 
Feature

Robert Roe reports on developments in AI that are helping to shape the future of high performance computing technology at the International Supercomputing Conference

Feature

James Reinders is a parallel programming and HPC expert with more than 27 years’ experience working for Intel until his retirement in 2017. In this article Reinders gives his take on the use of roofline estimation as a tool for code optimisation in HPC

Feature

Sophia Ktori concludes her two-part series exploring the use of laboratory informatics software in regulated industries.

Feature

As storage technology adapts to changing HPC workloads, Robert Roe looks at the technologies that could help to enhance performance and accessibility of
storage in HPC