The United States has retaken first place in a major metric of technological heft – the world’s fastest supercomputer is once again American.
The Oak Ridge National Laboratory, a branch of the U.S. Department of Energy, unveiled the new computer, called Summit, on Friday. The system, according to the department’s announcement, is rated to perform 200,000 trillion calculations per second, or 200 petaflops. That makes it eight times more powerful than America’s former fastest supercomputer, the Titan, and gives it a substantial edge over China’s 93-petaflop TaihuLight, which had been the world’s fastest supercomputer since 2016.
Summit will be used for research in fields including energy, advanced materials, and artificial intelligence. According to supercomputing news site Top500, it was built by IBM using IBM’s Power9 CPUs and Nvidia Tesla V100 GPUs.
Get Data Sheet, Coins2Day’s technology newsletter.
Advances in computing power have been described as a 21st century equivalent of the Space Race between America and the U.S.S.R., though the primary contenders this time around are the U.S. And China. The debut of Summit is a step towardswhat might be called that competition’s moon landing – an exascale computer, capable of a billion billion calculations per second. The U.S. Has set a goal of developing an exascale computer by 2021, but according to DoE, Summit can already reach exascale speeds for “certain scientific applications.” The Chinese Academy of Sciences has said they’ll deploy an exascale supercomputer by 2020.
Like the space race, the development of such incredibly powerful computers will spread innovation across many sectors, and could confer long-term competitive advantages to the nation in the lead. The most dramatic competition may be in the development of artificial intelligence, but much attention is also focused on Summit’s application in health sciences. The ability to search and collate vast amounts of data is expected to speed the discovery of new medicines and advanced treatments, such as by correlating diseases to genetic markers, or developing individually customized treatments for each patient.
