Skip to main content

The U.S. government plans to drop $500M on a ridiculously powerful supercomputer

The United States government has announced plans to build Aurora, a blisteringly fast “exascale” supercomputer by 2021 — an effort to boost research projects related to public health issues and environmental science.

The U.S. Department of Energy (DOE) formally announced its plans for the $500-million exascale supercomputer project on March 18. Once built, the Aurora supercomputer  is expected to offer a combination of “high-performance computing and artificial intelligence at exascale.”

In context, the word “exascale” refers to the fact that the Aurora supercomputer will be designed to perform at least one exaFLOP, which, according to the DOE, “is equal to one quintillion floating point computations per second.” This level of performance is expected to allow researchers “to address scientific problems at exascale” and promote the development of “new scientific innovations.”

While Aurora will be the first supercomputer in the U.S. to perform at one exaFLOP, it may not be the first exascale supercomputer in the world. As China’s state-controlled Xinhua notes, China’s version of an exascale supercomputer may come out even sooner, in 2020.

The Aurora supercomputer’s projected capabilities and impact were further addressed by U.S. Secretary of Energy Rick Perry in a statement on the upcoming project:

“Achieving exascale is imperative not only to better the scientific community, but also to better the lives of everyday Americans … Aurora and the next generation of exascale supercomputers will apply [high-performance computing] and A.I. technologies to areas such as cancer research, climate modeling, and veterans’ health treatments. The innovative advancements that will be made with exascale will have an incredibly significant impact on our society.”

The Aurora supercomputer project will be built by Intel and subcontractor Cray, Inc. and will be located in Chicago at the Argonne National Laboratory.

Very few details about the actual technologies involved in building  Aurora have been released. That said, Intel is expected to build Aurora using a future generation of its Xeon Scalable processor, a future generation of its Optane DC Persistent Memory, and its Xe architecture and One API software. All these technologies will have to be specifically designed for Aurora’s unique combination of AI and exascale computing.

Editors' Recommendations

Anita George
Anita has been a technology reporter since 2013 and currently writes for the Computing section at Digital Trends. She began…
The Meta Quest 3 will get an exciting new type of app
A Windows app extends into 3D space via a Meta Quest 3 VR headset.

A Windows app extends into 3D space via a Meta Quest 3 VR headset. Microsoft

At Build 2024, Microsoft announced it partnered with Meta to extend Windows apps into 3D space with the help of a Quest VR headset. When working on physical objects, it’s important to have spatial awareness of components.

Read more
How to build a PC from scratch: A beginner’s guide
Installing RAM in a desktop PC.

Building a PC for the first time, or even the second or third time, can feel a little intimidating. But one of the best parts about building a computer is that, for the most part, the parts fit where they should, and don't fit where they shouldn't. A graphics card will fit in the graphics card slot, and good luck putting the CPU in the wrong socket.

With a little care, time, and this handy guide, you can build a PC without hassle. We're here to walk you through it.

Read more
Microsoft just made Paint relevant again
Person using Windows 11 laptop on their lap by the window.

The controversial Recall feature has grabbed all the headlines from Microsoft’s Copilot+ announcements yesterday, but this new AI feature is also making Paint relevant again.

It’s called Cocreator, and it’s a new AI feature that can turn your quick sketch, augmented by text, into a much more realistic and impressive image. The exciting thing is that it does all this in real time. It might not get it right the first time, so you'll need patience, and the more details you give about what you want in the image, the better.

Read more