Skip to Content
Computing

The U.S. No Longer Has One of the Top Three Fastest Supercomputers

But with the devices still crucial for research, can $258 million of federal funding change that?

As America’s grasp on the world of supercomputing continues to slip, the Department of Energy is seeking to reinvigorate the nation’s capabilities.

A new ranking shows that for the second year running, the world’s fastest supercomputer is TaihuLight, housed at the National Supercomputing Center in Wuxi, China. Capable of performing 93 quadrillion calculations per second, it’s almost three times faster than the second-place Tianhe-2. And in third spot this year is a newly upgraded device, called Piz Dain, at the Swiss National Supercomputing Centre, which recently had its performance boosted by the addition of Nvidia GPUs.

Sadly for America, the upgraded Piz Dain pushes the Department of Energy’s Titan supercomputer, which is housed at Oak Ridge National Laboratory, into fourth spot. Able to make 17.6 quadrillion number crunches per second, Titan is just a fifth as fast as TaihuLight. In its defense, the U.S. still claims five of the top 10 spots, and it is home to 169 of the supercomputers that make up the fastest 500. China, meanwhile, can only claim 160.

But the news nonetheless serves to highlight America's decline as a supercomputing heavyweight. This is the first time since 1996 that America hasn’t held one of the top three spots. It indicates that, while it certainly has significant supercomputing resources, it can’t tackle its biggest problems at anywhere near the speeds enjoyed by researchers in China.

It may be tempting to suggest that the dawn of practical quantum computing, one of our 10 breakthrough technologies of 2017, would remove the need for supercomputers. But it will be several years before quantum computers can perform such large-scale calculations. Even then, while quantum devices promise to be able to solve some specific problems with incredible speed, it’s not clear whether they will be able to reliably solve all problems faster than regular supercomputers.

That leaves many areas of research currently reliant on the results of huge computational effort—among them drug discovery, materials science, and climate modeling—still in need of fast supercomputers.

The Department of Energy, whose Titan device was once the fastest in the world, is painfully aware of that fact. That’s exactly why it last week announced a $258 million funding injection into its exascale computing project. The money, which will be shared between AMD, IBM, Intel, Nvidia, Hewlett Packard Enterprise, and Cray, is intended to fuel development of machines that are 50 times faster than Titan.

The U.S. government reckons it will have a system capable of performing one quintillion operations per second—that’s 1,000 quadrillion, and 10 times the capacity of TaihuLight—by 2021. But China’s fierce investment into supercomputing has led it to claim that it may achieve the same feat by as soon as 2020. Speed, it seems, is of the essence for America.

(Read more: Top 500, Wall Street Journal, “The New Fastest Supercomputer Is Chinese Through and Through,” “10 Breakthrough Technologies: Practical Quantum Computers”)

 

Deep Dive

Computing

Inside the hunt for new physics at the world’s largest particle collider

The Large Hadron Collider hasn’t seen any new particles since the discovery of the Higgs boson in 2012. Here’s what researchers are trying to do about it.

Why China is betting big on chiplets

By connecting several less-advanced chips into one, Chinese companies could circumvent the sanctions set by the US government.

How Wi-Fi sensing became usable tech

After a decade of obscurity, the technology is being used to track people’s movements.

VR headsets can be hacked with an Inception-style attack

Stay connected

Illustration by Rose Wong

Get the latest updates from
MIT Technology Review

Discover special offers, top stories, upcoming events, and more.

Thank you for submitting your email!

Explore more newsletters

It looks like something went wrong.

We’re having trouble saving your preferences. Try refreshing this page and updating them one more time. If you continue to get this message, reach out to us at customer-service@technologyreview.com with a list of newsletters you’d like to receive.