On March 18, 2019, Argonne National Laboratory released more information about Aurora, “America’s next-generation supercomputer.” Intel has teamed up with the Department of Energy (“DOE”) to create the computer at Argonne’s lab facility which is estimated to cost upwards of $500 million. Cray Inc.—known for its 45 years of building the “world’s most advanced supercomputers” will be a sub-contractor on the deal. Together, Cray Inc. and Intel will work to construct “the fastest supercomputer in U.S. history.”
Aurora will be an “exascale” system, and complete “a quintillion calculations per second.” This speed of data processing will have enormous effects across various sectors of our society including science, health, and technology. The Director of Argonne, Paul Kerns, explained that the system can be used for things like “extreme weather forecasting, medical treatments, and brain mapping.”
The US has been attempting to develop an exascale supercomputer for the past several years. Other countries, such as China, have been racing to create an exascale computing system, but will not release it until 2020. Aurora is scheduled to be completed by the end of 2021, and it will be the United State’s “first exascale computing system.”
Exascale computing will allow for a processing speed so high that a computer operating with this level of functionality will be able to “simulate the processes involved in precision medicine, regional climate, additive manufacturing, the conversion of plants to biofuels, etc.”
What really sets Aurora apart from other supercomputers will be its “ability to use artificial intelligence (AI) to guide its models and simulations.” The computer will use AI to narrow the number of possible computations the system could run in order to produce the most effective problem-solving. This is the first supercomputer to involve a “convergence of artificial intelligence and high-performance computing.”
Currently, the fastest supercomputer is Summit, located at Oak Ridge National Laboratory. Aurora will overpower Summit not only in terms of speed, but also in terms of size. It is estimated to be as big as “about 10 tennis courts,” while Summit is only around “two tennis courts.”
Aurora is estimated to have a “multiplier effect for scientific discovery as machine learning methods and AI become more common and crucial.” The integration between AI and amplified processing speed make Aurora one of the most unique supercomputers built by the DOE to date.
Samantha Moench is a second-year law student at Wake Forest University School of law. She holds a Bachelor of Arts in Psychology and a minor in Mass Communications from Washington and Lee University. Upon graduation, she intends to practice real estate corporate law.