IBM’s AC922 Summit, at Oak Ridge National Laboratory in Tennessee, is the reigning champion as the world’s fastest supercomputer.
With a peak speed of 200 petaFLOPs, Summit can perform 200 million billion calculations per second.
Already, at these mind-boggling speeds, supercomputers unlock new boundaries for science in cosmology, medicine, and chemistry to name just a few.
Where do we go from there? There’s no theoretical limit to supercomputing power and the Summit won’t keep its title for long as it’s threatened by at least one contender.
The U.S. government enlisted the help of leading tech companies to design a new supercomputer architecture that will hit the exascale range.
The U.S. Department of Energy (DoE) has awarded Intel Corporation and Cray a $500 million contract to build the world’s first exascale computer.
Intel makes semiconductors and data center chips, and it’s subcontractor Cray Inc specializes in supercomputers. Together they’ll design a new supercomputing system that would be able to hit speeds in the exaFLOP range.
One exaFLOP is one quintillion, or a billion billions calculations per second. For comparison, IBM’s Summit is in the quadrillion range with 0.2 exaFLOPs speed.
To be delivered by 2021, this exascale computer, called Aurora, is described on paper as the fastest and most advanced supercomputer in the world.
To read more, click here.