Skip to main content

U.S. intends to build record-setting exascale supercomputer by 2021

Oak Ridge Leadership Computing Facility/Oak Ridge National Laboratory/U.S. Department of Energy

The United States government has announced plans to build a second, faster exascale supercomputer, after it already announced plans to build a $500 million exascale supercomputer less than two months ago.

The new plans to build another, faster exascale supercomputer were announced by the U.S Department of Energy (DOE) on May 7. According to the Department of Energy’s press release, the new supercomputer has been named Frontier and is expected to cost $600 million “for the system and technology development.”

And unlike Aurora, another proposed exascale supercomputer the Department of Energy announced plans for in March, Intel will not be involved in the development of the Frontier supercomputer. In fact, according to the DOE, computer processor manufacturer AMD will be partnering with Cray Inc. to build Frontier together. Frontier’s completion date is set for 2021, and the supercomputer will be housed at the Oak Ridge National Laboratory.

As The Verge notes, while Aurora is still on track to be the first exascale supercomputer in the United States, Frontier is being built with the intention of it being the fastest. Frontier is already being touted by the DOE as being “record-setting” and “the world’s most powerful computer,” and it hasn’t even been built yet. Both supercomputers have completion dates set for 2021, but Aurora will only have a processing power of at least one exaflop. Frontier is expected to have at least 1.5 exaflops. Frontier’s overall anticipated processing power was described by the DOE as being able to solve calculations “up to 50 times faster than today’s top supercomputers” with a rate that exceeds “a quintillion, or, 1018 , calculations per second.”

As the DOE explains, Frontier’s boosted processing power is supposed to help researchers “deliver breakthroughs in scientific discovery, energy assurance, economic competitiveness, and national security.” Furthermore, since Frontier is considered to be a “second-generation A.I. system,” it will be expected to support deep learning, machine learning, and data analytics for real-world applications like manufacturing and health.

As mentioned earlier, AMD and Cray Inc. will be co-designing and developing the technologies integral to Frontier’s performance. When it’s built, Frontier’s system is expected to have the following components: over 100 Cray Shasta cabinets, “A.I.-optimized” AMD EPYC processors, and Radeon Instinct GPU accelerators.

Editors' Recommendations