2 mins read

Fujitsu Receives Order for Japan’s Fastest Supercomputer System for AI Applications

Fujitsu Leverages World's Fastest Supercomputer and AI Tech in Joint Field Trial for Safe and Efficient Tsunami Evacuations in Kawasaki City
Fujitsu Leverages World's Fastest Supercomputer and AI Tech in Joint Field Trial for Safe and Efficient Tsunami Evacuations in Kawasaki City
Fujitsu today announced that it has received an order from the National Institute of Advanced Industrial Science and Technology (AIST) for AI Bridging Cloud Infrastructure (ABCI), which is set to become the fastest supercomputer system in Japan.
Included in the ABCI configuration will be Fujitsu’s latest x86 servers, the Fujitsu Server PRIMERGY. The system will offer a theoretical peak performance in half-precision floating point operations (a critical measure in the AI field) of 550 petaflops(1), and 37 petaflops of double-precision floating point operation performance, a conventional measure in simulation and other fields. If this system had competed in the latest TOP500(2) ranking of supercomputers that took place in June 2017, it would have taken the top position in Japan and third place globally.
ABCI will be deployed to the new ABCI datacenter on the Kashiwa II campus of the University of Tokyo, and is scheduled to begin operations in fiscal 2018 (fiscal years start April 1).
Background
In order to accelerate cutting-edge research and development in AI technology, as well as practical applications in society in such fields as industry, AIST has been planning to deploy ABCI as a global open innovation platform that will enable high speed AI processing by combining algorithms, big data and computational power. As a cloud platform for AI applications offering the world’s top class machine learning processing capability, high performance computational capability, and energy efficiency, ABCI is expected to create new applications in a variety of fields. Furthermore, the system is foreseen to promote the utilization of cutting-edge AI technology by industry, including transfer of the latest cloud platform technology to the public through an open design.
Fujitsu has now received the order for this system due to the company’s proposal for a large-scale computational system using the technology acquired through HPC development, as well as AI focused software acceleration technology from Fujitsu Laboratories Ltd.
Overall System Summary
ABCI will feature hardware including a high-performance computational system, a high-capacity storage system, and a variety of networking technology, as well as a group of software that makes the best use of the system.
As the core of ABCI, the high-performance computational system will consist of 1,088 PRIMERGY CX2570 M4 servers, mounted in Fujitsu’s PRIMERGY CX400 M4 multi-node servers. Each server will feature the latest components, including two Intel Xeon Gold processor CPUs (a total of 2,176 CPUs) and four NVIDIA Tesla V100 GPU computing cards (a total of 4,352 GPUs), as well as Intel SSD DC P4600 series based on an NVMe standard, as local storage.
Moreover, the 2U size chassis PRIMERGY CX400 M4 can each mount two PRIMERGY CX2570 M4 server nodes with GPU computing cards, offering high installation density. In addition, by utilizing “hot water cooling” for its servers, this system can also realize significant power savings.
Future Plans
Applying its AI and HPC technology to AIST’s high system requirement standards for both hardware and software, Fujitsu will construct the system with the goal of commencing operations in fiscal 2018. After the launch, Fujitsu will support with the operation of ABCI, and with researchers both inside and outside AIST, through R&D and practical applications in society will contribute to the acceleration of AI utilization.
Fujitsu will also adapt the various techniques, insight, and experience gained from building and support in the operation of ABCI to Fujitsu Human Centric AI Zinrai, its approach to AI technology, as well as to its HPC technology, improving the company’s products, services and solutions while promoting its business.
(1) Petaflops
Quadrillion floating point operations per second.
(2) TOP500
The TOP500 list is a project that periodically ranks and evaluates the 500 computer systems around the world with the highest processing speed for simultaneous linear equations. Begun in 1993, it publishes its supercomputer lists twice per year.