Tuesday, March 6, 2018

Exascale Computers: Competing With China Is About to Get Serious

From The Next Platform:

U.S. Exascale Efforts Benefit in FY 2019 Budget
There was concern in some scientific quarters last year that President Trump’s election could mean budget cuts to the Department of Energy (DoE) that could cascade down to the country’s exascale program at a time when China was ramping up investments in its own initiatives.

The worry was that any cuts that could slow down the work of the Exascale Computing Project would hand the advantage to China in this critical race that will have far-reaching implications in a wide range of scientific and commercial fields like oil and gas exploration, financial services, high-end healthcare, national security and the military.

In a talk about exascale computing a year ago at the Princeton Plasma Physics Lab, Paul Messina, Argonne National Lab Distinguished Fellow, senior strategic advisor and former head of the Exascale Computing Project, mentioned the concerns about the Trump administration and, pointing to China’s efforts, said that “for the United States, it’s important for us to be right there, if not beyond.”
As we noted last summer, the DoE and its hardware partners – Intel, IBM, Nvidia, Cray, Hewlett Packard Enterprise and AMD – got a financial boost of almost a half billion dollars in research, development and deployment investments. In addition, it was during Trump’s first year that the timetable for the first U.S. exascale system to launch accelerate from 2023 to 2021. Now those involved with U.S. exascale efforts are getting a look at how the Trump administration values the exascale initiatives in the coming year, and they will like what they see.

The Trump administration last month released its budget proposal for the government’s fiscal year 2019, a proposal that puts into writing the administrations spending priorities and goals for the year. Many the departments within the government under the president’s budget would see cuts in their budgets, some as large as 29 percent. However, the DoE would get about 2 percent bump, to $30.6 billion. Within that figure would be $636 million for exascale computing, a $376 million increase from the FY 2017 allocation. The money would be spread across two different offices within the DoE – $473 million in the Office of Science and $163 million in the National Nuclear Security Administration (NNSA), according to FY 2019 Budget Request Fact Sheet released by the DoE.
According to the agency’s Budget in Brief document, the money would be used to “support development of an exascale computing software ecosystem by preparing mission critical applications to address exascale challenges” and fund research, development and design at both the Argonne and Oak Ridge national labs to help the country on track to roll out its first exascale-capable system in 2021 and a second one built via a different architecture a year later. In addition, DoE noted that the partnership between the Office of Science and NNSA will boost the United States’ national security by “supporting the nuclear stockpile while supporting the next generation of science breakthroughs not possible with today’s fastest computing systems.”

Another interesting point in the budget proposal is $105 million to “address the emerging urgency of building U.S. competency and competitiveness in the developing area of quantum information science, including quantum computing and quantum sensor technology.”...MUCH MORE
Feb. 19
Oak Ridge National Laboratory's Report On China's Fastest Supercomputer in the World
Sunway TaihuLight is the computer ORNL will surpass later this year with the new Summit system, more than doubling the speed of the Sunway TaihuLight.
However....China is already planning an exascale machine, five times faster than Summit, capable of a billion billion calculations per second....
January 29
"China set to launch its new supercomputer"
NVIDIA watches, some commentary after the jump...

.... On the last Top500 list of the world's fastest supercomputers, 87 systems use NVIDIA GPUs as accelerators. Not the top two however. You have to go down to #3, Switzerland's Piz Daint which was upgraded with NVIDIA chips in 2016. China is using an entirely different architecture which we touched on in June 2016's "Milestone: China Builds The (NEW) World's Fastest Supercomuter Using Only Chinese Components (and other news) INTC; NVDA; IBM".

The different approaches have implications beyond simple bragging rights. As we said a couple weeks ago in  the outro from "'Can Chinese AI Chip Makers Compete with Nvidia?' (NVDA)":

One thing not mentioned in the Nanalyze report that is critical to understanding Chinese R&D is the importance of the People's Liberation Army (Navy) and the advantage the closed loop of academia, end-user and PLA contracted-and-government-owned companies conveys:...
You'll note the Asia Times story has the co-developer research lab as the PLA funded National University of Defence Technology.

The 200 petaflop/second (quadrillions of calculations per second) Summit supercomputer being built at Oak Ridge National Laboratory will blow past the current world's fastest, the 93 petaflop/s Sunway TaihuLight but this latest monster is planned to be five times faster still.

At that point, if you were to use it for, saaay, training artificial intelligence, you are going places that the human mind literally can't comprehend much less forecast or, dream on, guide.
January 11
With the Summit Supercomputer, U.S. Could Retake Computing’s Top Spot (NVDA)
September 2017
"The Astonishing Engineering Behind America's Latest, Greatest Supercomputer"

And many more, including:

June 2016