The system will have a total theoretical processing performance of 4 petaflops.
Japanese IT giant Fujitsu is building a new deep learning supercomputer based on Nvidia's DGX-1 AI systems for the Riken Centre for Advanced Intelligence Project.
The system is comprised of two server architectures, with 24 Nvidia DGX-1 servers—each including eight of the latest Nvidia Tesla P100 accelerators and integrated deep learning software—and 32 Fujitsu Server Primergy RX2530 M2 servers, along with a high-performance storage system, Fujitsu said in a statement.
The system, which is scheduled to go online in April, will have a total theoretical processing performance of 4 petaflops. Riken will use the new supercomputer, scheduled to go online next month, to accelerate AI research in several areas, including medicine, manufacturing, healthcare and disaster preparedness.
Fujitsu said the file system will run Fujitsu Software FEFS on six Fujitsu Server Primergy RX2540 M2 PC servers, eight Fujitsu Storage Eternus DX200 S3 storage systems and one Fujitsu Storage Eternus DX100 S3 storage system to provide the IO processing demanded by deep learning analysis.
Along with the standard DGX-1 deep learning software environment provided by Nvidia in a public cloud, Fujitsu said it will also integrate a customised software environment for use in a secure on-site network.
The new supercomputer will be installed in Fujitsu's Yokohama data centre. Fujitsu will also provide Riken with R&D support when using the system.