Skip to content

1.1 trillion operations per second: US has the fastest supercomputer in the world

    A systems engineer stands in a large room with the Frontier supercomputer.
    enlarge Systems engineer Matt Ezell, the systems leader of the Frontier supercomputer at Oak Ridge National Laboratory.

    The US has recaptured the top spot in the world supercomputer rankings with the exascale Frontier system at Oak Ridge National Laboratory (ORNL) in Tennessee.

    The Frontier system’s score of 1,102 exaflop/s makes it “the most powerful supercomputer that ever existed” and “the first true exascale machine,” the Top 500 project said Monday in announcing its latest ranking. Exaflop/s (or exaflops) represents 1 trillion floating point operations per second.

    Frontier was more than twice as fast as a Japanese system that came in second in the ranking, which is based on the LINPACK benchmark that measures the “performance of a special system for solving a dense system of linear equations” .

    “Based on the latest HPE Cray EX235a architecture and equipped with AMD EPYC 64C 2GHz processors, the system has 8,730,112 total cores, power efficiency of 52.23 gigaflops/watt, and relies on gigabit Ethernet for data transfer,” according to the Top 500 group. † Frontier “is currently being integrated and tested at ORNL in Tennessee, US, where it will be managed by the Department of Energy,” the group said.

    While China may also have broken the exaflop barrier, the US system is the first to demonstrate its speed in official Top 500 tests.

    Frontier also tops the efficiency rankings

    Frontier was also the second most efficient supercomputer in the new Green 500 ranking. The top spot in Green 500 was taken by a smaller version of the Frontier system in Oak Ridge, one with 120,832 cores instead of the 8.7 million in the exascale system.

    Second in the Top 500 with a score of 442 petaflop/s was the previous top supercomputer, the Fugaku system at the RIKEN Center for Computational Science in Kobe, Japan. While the Fugaku system has a theoretical peak of more than 1 exaflop/s, Frontier is the only system to have demonstrated this in Top 500 testing, according to the Top 500 announcement. ORNL said Frontier’s theoretical peak is 2 exaflop/s.

    Fugaku was in first place as of the June 2020 ranking and remained at the top in November 2020, June 2021 and November 2021. The US was last in first place in the biennial ranking in November 2019, when it was defeated by Oak Ridge IBM-built Summit system posted a score of 148.6 petaflop/s.

    DOE’s exascale project

    The DOE announced in 2018 a $1.8 billion project to develop at least two exascale supercomputers to support a range of scientific research.

    “Work to deliver, install and test Frontier began during the COVID-19 pandemic, when shutdowns around the world put pressure on international supply chains,” ORNL said this week. “More than 100 members of a public-private team worked around the clock, from sourcing millions of components to delivering system components on time to meticulously installing and testing 74 HPE Cray EX supercomputer cases, including more than 9,400 AMD-powered nodes and 90 miles of network cables.”

    Researchers will “be able to access the fully operational Frontier system later this year,” ORNL associate lab director Jeff Nichols said, adding that it will be available to “scientists and engineers from around the world.”

    “Frontier users will model the entire lifespan of a nuclear reactor, uncovering disease genetics and building on recent advances in science and technology to further integrate artificial intelligence with data analytics and modeling and simulation,” ORNL also said.

    HPCwire’s coverage of the Top 500 list further details Frontier’s makeup:

    Frontier consists of 74 cabinets, each weighing 8,000 pounds. 9,408 HPE Cray EX nodes are scattered across these cabinets, with each node powered by an AMD “Trento” 7A53 Epyc CPU and quad AMD Instinct MI250X GPUs (a total of 37,632 GPUs across the system). The system has 9.2 petabytes of memory evenly distributed between HBM and DDR4, and uses HPE Slingshot-11 networks. It is supported by 37 petabytes of node-local storage on top of 716 petabytes of central storage. The system is 100 percent liquid-cooled using hot (85 °F) water, and space wise, the final Frontier system covers 372 square feet.

    HP Enterprise bought the supercomputer company Cray in 2019.

    China reportedly has exascale supercomputers

    There is some evidence that China has two exascale supercomputers. But operators “of those systems have not submitted test results for evaluation” in the Top 500 rankings, possibly due to “tensions between the United States and China,” The New York Times wrote.

    Officially, China’s top-ranked system is the sixth fastest in the Top 500. That system, known as Sunway TaihuLight, ranked #1 four times in a row in 2016 and 2017, but hasn’t updated its score since. Another Chinese system, called Tianhe-2A, was in first place from 2013 to 2015 and now ranks ninth.

    “There are rumors that China has something” with exascale power, but “there is nothing official,” Jack Dongarra, a computer scientist who designed Linpack, told The New York Times. Dongarra is a professor at the University of Tennessee and a position at ORNL.