The world’s most powerful supercomputer

On Tuesday, May 7, it was announced that the U.S. Department of Energy had contracted with Cray to build Frontier at Oak Ridge National Laboratory in Tennessee. Frontier will be the world’s most powerful supercomputer by 2021, exceeding a performance of 1.5 exaflops, and the contract is valued at 600 million USD for system and technology development. The laboratory’s Center for Accelerated Application Readiness is already taking proposals from scientists for projects to run on the system.

U.S. Energy Secretary Rick Perry at the announcement ceremony for the next-generation Frontier supercomputer at Oak Ridge National Laboratory. Photo: Brianna Paciorka/News Sentinel.

What exactly is a supercomputer?

A supercomputer can process a vast amount of data at a significantly higher speed compared to a general purpose computer. Instead of measuring the performance in MIPS, a supercomputer’s performance is measured in FLOPS (floating-point operations per second). Supercomputers are a powerful tool and can be used for a wide range of computationally intesive tasks, e.g. for quantum mechanics; weather forecasting; climate research; molecular modeling; and physical simulations. Exascale is the next level of computing performance that solves calculations five times faster than the current supercomputer Summit in the U.S., exceeding a quintillion (or a billion billion) calculations per second. This means that exascale supercomputers, such as Frontier, will be able to simulate processes more realistically.

A comparison of the system specifications between the two current supercomputers in the U.S., Titan and Summit, and Frontier. Image: Oak Ridge National Laboratory.

Frontier will help the U.S. to maintain leadership in high-performing computing and to further integrate artificial intelligence with data analytics and modeling and simulation. It is anticipated by Oak Ridge National Laboratory to have a big impact on research within the following areas, just to name a few:

  • Precision medicin
  • Regional climate
  • Additive manufacturing
  • The relationship between energy and water use
  • The unseen physics in materials discovery and design
  • The fundamental forces of the universe

By automatically recognizing patterns in data and guiding simulations, beyond the limits of traditional approaches, will drastically reduce the time to discovery.

The race is on

The fundamental reason for building an exascale computer for a value exceeding 600 million USD is to accelerate innovation in science and technology, which in turn will lead to economic benefits. However, it has also become part of the race in technology advancement between nations – especially between China and the U.S. The commitment to exascale computing came as a reaction by the U.S. to China’s quick rise to the top of high-performance rankings according to Moor Insights & Strategy analyst Patrick Moorhead. Eric Van Hensbergen, distinguished engineer and acting director of HPC at Arm Research, told DCD that “exascale supercomputers are a critical piece of national infrastructure“, and not having access to that resource could have drastic consequences on the country’s economy and threaten the national security.

The U.S. is not alone in building exascale computers. China is anticipated to launch its first exascale computer already by 2020, it is not anticipated to be as powerful as Frontier according to DCD. Apart from China, Japan and the EU are also in the race. In October 2018 the launch of EU’s 1 billion euro project to develop the world’s fastest supercomputer by 2023 was announced. The organization EuroHPC is set to manage the project and and will receive 486 million euro from the European Commission’s Horizon 2020 research budget, with a similar amount from 25 member states. It is also possible for companies to provide in-kind contributions to the project. But as each superpower gets its own supercomputers, the demand will come again to build something bigger, better and faster.