Japan's Institute for Solid State Physics (ISSP) at the University of Tokyo is the latest win for SGI, which is to pull the big red switch on a 2.65 petaflop system on July 1.
The system will be a hybrid of Intel Xeon processors and NVIDIA Tesla K40 GPUs, the company says in its canned statement.
The SGI ICE XA will have 1,872 nodes, 44,928 cores, and 240 TB of memory. The GPUs – 576 of them – will be installed in 288 of the nodes.
There's also to be a shared memory system comprising 19 Xeon-based SGI UV systems totalling 760 cores and 19 TB of memory, which would max out at 31.6 teraflops.
The company's E-Cell cabinet will provide the water cooling and heat exchanger for the whole system, with additional water-cooled heatsinks for CPUs and GPUs.
The ISSP wants large-scale parallel software workloads in the “computational physics and chemistry in material science.”
“Now, our Institute will have two sets of supercomputer systems, the SGI ICE XA and SGI UV, to enhance the joint-use service for material science researchers in Japan,” says ISSP associate professor Osamu Sugino.
“A higher calculation capability is required for research deployments with more complicated substances and high-affinity environments in order to conduct leading-edge studies using new algorithms.” ®
Sponsored: Webcast: Simplify data protection on AWS