The ABCI configuration will include Fujitsu's latest PRIMERGY x86 servers. The system will offer a maximum theoretical performance of 550 petaflops in half-precision floating-point operations (a critical metric in the field of AI) and 37 petaflops in double-precision floating-point operations, a conventional metric in simulation and other fields. Had this system been included in the TOP500 supercomputer ranking published in June 2017, it would have achieved first place in Japan and third place worldwide.
 
ABCI will be deployed at ABCI's new data center on the University of Tokyo's Kashiwa II campus and is scheduled to begin operations during fiscal year 2018 (fiscal years begin on April 1).
 
To accelerate cutting-edge research and development of AI technology, as well as practical applications in society and fields such as industry, AIST has been planning the implementation of ABCI as a global open innovation platform. This platform will enable high-speed AI processing by combining algorithms, big data, and computing power. As a cloud platform for AI applications, it offers world-class machine learning processing capabilities, high-performance computing power, and energy efficiency. ABCI is expected to create new applications in a variety of fields. Furthermore, the system is anticipated to promote the adoption of cutting-edge AI technology by industry, including the transfer of the latest technology from the public cloud platform through an open design.

More information