Data Centre

HPC

Uncle Sam bungs rich tech giants quarter of a billion bucks for exascale super R&D

Carrot dangled for 2021 mega-machines

By Chris Williams, Editor in Chief

18 SHARE

The US government has dangled $258m in funding in front of six American tech giants to encourage the development of exascale supercomputer systems.

Uncle Sam wants at least one exascale machine powered up on American soil by 2021, and in production by 2022. A second system should also be arriving around that time, ideally.

By exascale, we mean a computer that can hit at least one exa-FLOPS, or a billion billion floating-point math calculations per second, which would smoke today's publicly known world's-fastest machine (China's 93PFLOPS Sunway TaihuLight). The US Department of Energy wants number-crunching beasts that can sustain exascale levels of performance for applications, rather than a peak 1EFLOPS figure that isn't achievable outside of a benchmark.

To help make this all happen, the Rick Perry-led energy department is offering about quarter of a billion dollars over a three-year period to half a dozen technology suppliers to carry out exascale R&D work. The companies will spend some of their own dosh on the effort – at least 40 per cent of the total research bill – ensuring taxpayers aren't footing the entire expense.

The plan, we're told by officials, is this: the tech giants will be paid to investigate specific areas of research, such as an application or system design, with input from supercomputing software developers and engineers. They'll be expected to produce final reports on each of the studies. The overall goal is to work out a way to build and supply affordable exascale machines that meet US boffins' requirements. Universities and government research labs will ultimately buy and operate the systems when they are ready using their own budgets.

"It isn't that we send the money, wait three years, and get an answer," said Paul Messina, director of the energy department's Exascale Computing Project. "The funding is based on specific work packages."

The six companies are: AMD, Cray, HPE, IBM, Intel and Nvidia. They will, according to Messina, be expected to come up with "innovative memory architectures, higher-speed interconnects, improved reliability systems, and approaches for increasing computing power without prohibitive increases in energy demand."

"It is essential that private industry play a role in this work going forward," he added.

Uncle Sam previously injected about $40m into exascale research.

"Here's another way to think about this R&D investment," said Nicole Hemsoth, co-editor of our sister site The Next Platform, in an analysis of today's news. "$258 million in funding across six companies to develop something that might be able to stand in as the correct architecture eventually."

Meanwhile, the government is said to be renegotiating with Intel and Cray over the construction and installation of the $200m 180-petaFLOPS Aurora supercomputer. It was due to arrive in 2018, but it looks as though that deadline will be pushed back as the vendors and the Department of Energy hammer out a new contract. ®

Sign up to our NewsletterGet IT in your inbox daily

18 Comments

More from The Register

Cray's pre-exascale Shasta supercomputer gets energy research boffins hot under collar

US DoE sees multiple CPU, GPU, interconnect support and snaps one up for $146m

WekaIO almost, but not quite, summits Summit supercomputer on storage performance

Takes second place in IO-500 10 Node Challenge

HPE supercomputer is still crunching numbers in space after 340 days

No rad hardening so even HPE is 'pleasantly surprised'

Sorry, but NASA says Mars signal wasn't Opportunity knocking

Mislabelled signal raised rover fans' hopes, just for a while

EU plans for domestic exascale supercomputer chips: A RISC-y business

Analysis Consortium possibly looking to flex Arm muscle, too

Spectre/Meltdown fixes in HPC: Want the bad news or the bad news? It's slower, say boffins

MIT Lincoln metalheads broke big iron so you don't have to… oh, you still have to, don't you?

US regains supercomputer crown from Chinese, for now

America! FLOP yeah!

Lazy parent Intel dumps Lustre assets on HPC storage bods DDN

Chipzilla offloads devs, support teams and contracts

Heads up: Fujitsu tips its hand to reveal exascale Arm supercomputer processor – the A64FX

Hot Chips AKA how it learned to stop loving SPARC64

HPC botherer DDN breaks file system benchmark record

Runs SPEC SFS 2014 software builds 25% faster than E8 Optane system