Theoretical system performance is 15.88 petaflops.
Fujitsu has received an order for a next-generation supercomputer system from the Nagoya University Information Technology Center. The launch of the system is scheduled for July 2020.
This will be the world’s first system of 2,304 Fujitsu PRIMEHPC FX1000 supercomputer nodes that use Fugaku’s supercomputer technology, developed in conjunction with RIKEN. It is based on A64FX processors with scalable vector extension (SVE), which is an extension of the Arm v8-A architecture for supercomputers, and HBM2 memory.
The system will also include a cluster of 221 nodes on x86-compatible Fujitsu Server PRIMERGY CX2570 M5 servers, which are based on second-generation Intel Xeon processors and NVIDIA V100 GPUs, and storage systems.
The theoretical system performance is 15.88 petaflops.