Fujitsu Limited will begin global salesof the PRIMEHPC FX1000 and PRIMEHPC FX700 models from the "Fujitsu Supercomputer PRIMEHPC" series, which utilize the technology of supercomputer Fugaku jointly developed by RIKEN and Fujitsu.

In related news, supercomputer leader Cray and Fujitsu partner to develop the first commercial supercomputer powered by the Fujitsu A64FX Arm-based processor with high-memory bandwidth (HBM).

PRIMEHPC Supercomputers

Scheduled to start shipping in March 2020, the new PRIMEHPC products are equipped with the first CPU A64FX adopting Scalable Vector Extension (SVE) which is an extension of the Armv8-A architecture for supercomputers.

Fujitsu says that the CPU not only achieves a high memory bandwidth by using HBM2, a high-performance die-stacked memory, but also can handle half-precision arithmetic and multiply-add, which are important in such technologies as deep learning.

With support for the same Tofu interconnect D as Fugaku for maximum scalability in large configurations, and ultra-dense implementation of up to 384 nodes per rack, FX1000 can efficiently build large systems that deliver theoretical computing performance in excess of 1.3 exaflops. This model uses a water-cooling system.

The FX700 is based on the standard technologies for supercomputer systems, supporting InfiniBand as an interconnect and using open source software (OSS) for HPC middleware. By adopting an air-cooling system and a chassis that can be mounted in a standard server rack, this model is designed for easy deployement.

In addition, Fujitsu plans to work with application vendors to be available commercial application software for PRIMEHPC, including the crash analysis application LS-DYNA, which is widely used worldwide.

Fujitsu and Cray

Cray, a Hewlett Packard Enterprise company (NYSE: HPE) and Fujitsu announced a partnership to offer high performance technologies for the exascale era. Under the alliance agreement, Cray is developing the first-ever commercial supercomputer powered by the Fujitsu A64FX Arm®-based processor with high-memory bandwidth (HBM) and supported on the proven Cray CS500 supercomputer architecture and programming environment. Initial customers include Los Alamos National Laboratory, Oak Ridge National Laboratory, RIKEN Center for Computational Science, Stony Brook University, and University of Bristol. As part of this new partnership, Cray and Fujitsu will explore engineering collaboration, co-development, and joint go-to-market.

"Our partnership with Fujitsu means customers now have a broader choice of processor technology to address their pressing computational needs," said Fred Kohout, senior vice president and CMO at Cray, a Hewlett Packard Enterprise company. "We are delivering the development-to-deployment experience customers have come to expect from Cray, including exploratory development to the Cray Programming Environment (CPE) for Arm processors to optimize performance and scalability with additional support for Scalable Vector Extensions and high bandwidth memory."

The new Fujitsu processor is unique in that it is the first processor to deliver HBM and Arm Scalable Vector Extensions (SVE). HBM2 provides transfer speeds that are significantly faster than DDR4 giving the A64FX a maximum theoretical memory bandwidth greater than 1 terabyte per second (TB/s), and support for Arm SVE provides improved performance for artificial intelligence and analytics. The Cray CS500 system can apply this compute power to a wide range of HPC and AI workloads while still delivering features of Arm-based systems with high parallelization, low power consumption and high reliability.

The Cray supercomputer powered by Fujitsu A64FX Arm will be available through Cray in mid 2020.