The U.S. Department of Energy says it is working on a supercomputer that will break the target of exaFLOP computation – a quintillion (1018) floating-point computations per second – in order to handle high-performance computing and artificial intelligence.
Being built in conjunction with Intel and Cray Computing, the Aurora supercomputer will cost more than half a billion dollars and be turned over to Argonne National Laboratory in Chicago in 2021, according to a statement by the DoE.
“Aurora and the next-generation of Exascale supercomputers will apply HPC and AI technologies to areas such as cancer research, climate modeling, and veterans’ health treatments,” says Energy Secretary Rick Perry. “The innovative advancements that will be made with Exascale will have an incredibly significant impact on our society.”
Projects include “developing extreme-scale cosmological simulations, discovering new approaches for drug response prediction, and discovering materials for the creation of more efficient organic solar cells,” DoE said in a statement.
“Argonne’s Aurora system is built for next-generation Artificial Intelligence and will accelerate scientific discovery by combining high-performance computing and artificial intelligence to address real world problems,” says Argonne Director Paul Kearns, “such as improving extreme weather forecasting, accelerating medical treatments, mapping the human brain, developing new materials, and further understanding the universe – and that is just the beginning.”
The underpinnings of Aurora will be a future generation of Intel Xeon Scalable processor, Intel Optane DC Persistent Memory, Intel Xe compute architecture and Intel One API software. Cray will contribute its next-generation Shasta high-performance switch fabric codenamed “Slingshot.”
The current fastest supercomputer according to the latest semi-annual Top500 ranking is called Summit, whose computations top out at 143.5 petaFLOPS (143.5 x 1015). It also is a DoE supercomputer, in use at Oak Ridge National Laboratory in Tennessee. It runs on 2,282,544 IBM Power9 cores and 2,090,880 Nvidia Volta GV100 cores.
The DoE announcement of Aurora two years before it’s due to be delivered and relying on technology that has yet to materialize is seen as an attempt to set a stake in the ground for performance in the international competition for the fastest supercomputer.
The U.S. holds the number one and two rankings in the Top500 list as well as three others in the top 10, but China holds the the most places with 227 of the top 500 slots.