Just a few months ago, it was confirmed (for the fifth consecutive year) that China — not the United States — possesses the fastest supercomputer in the world.
The Tianhe-2 (Milky Way-2) ranked first yet again on the TOP500 list of the world’s most powerful supercomputers.
The United States’ Titan, located at one of the Department of Energy’s laboratories, ranked second, but it wasn’t a close second.
In fact, the Tianhe-2 doubly outperforms the Titan. It is capable of performing twice the number of operations in a single second.
According to this infographic from the TOP500, the United States might still have the most supercomputers, but we definitely don’t have the best — by a long shot.
To pour more salt on the wound, the U.S. is actually nearing historic national lows in regards to the number of overall supercomputer systems we operate
“Although the United States remains the top country in terms of overall systems with 233 (up from 231 in November 2014), this number is down from 265 on the November 2013 list.”
Sometimes the greatest technological advances come from divine inspiration and altruistic desire to make the world a better place, and sometimes they come by sheer accident.
This time, it will come from the competitive desire to be better than the Chinese…
And maybe a little realization that the United States has been fairly neglectful to computing R&D over the past few years.
You Get What You Pay For
A recent report out of MIT warns that cutbacks in technology R&D spending leave the United States at risk to lose its position as the world’s leader in tech innovation.
Others argue that the hold has already been lost.
As Europe and Asia both increase government investments in technology, the United States continues to pull back.
Today, only 3.6% of the national budget is designated for R&D, compared to more than 9% in 1968.
Economists predict that by 2020, that investment will be reduced even further — to little more than 2%.
These cutbacks don’t just hurt the United States’ reputation and economic opportunities; they are also stunting the possibility of incredible breakthroughs and slowing the progress of development.
The most obvious example is in the biotech sector.
One of the report’s authors points out, “In the last decade we’ve had zero Alzheimer’s drugs approved by the FDA, and the reason is that we don’t understand basic science of degenerative diseases.”
The executive order signed by President Obama in July establishes the National Strategic Computing Initiative (NSCI), an organization that now has the responsibility to reclaim first place in the TOP500 rankings.
But it seems like the order is a little more than that.
It symbolizes recognition that the United States needs to regain its overall position as a tech powerhouse and raise entrepreneurial spirit…
Or miss out on the innovation that keeps us on top.
Taking Back the Trophy
NSCI will involve collaboration between the National Science Foundation, Departments of Energy and Defense, NASA, the National Oceanic and Atmospheric Administration (NOAA), the FBI, the National Institutes of Health, and other private sector contributors.
“Once complete, the new supercomputer will be operating at one quintillion or 1 exaflop calculations for every second (which is basically 1,000,000,000,000,000,000 for each second). The fastest supercomputer, Tianhe-2, currently runs at 38.86 quadrillion, 33.86 petaflops or 38,860,000,000,000,000 per second calculations.”
If you’re like me and need a translation for that lingo, here it is:
Supercomputers are rated and ranked based not on capacity but on capability — that is, the speed at which they can perform computations and operations.
This speed is quantified in “petaflops.”
A single petaflop means the computer is performing one quadrillion operations per second.
China’s top-ranked supercomputer, Tianhe-2, currently runs at 38.86 petaflops.
That means 38.86 quadrillion calculations per second.
If you’re also like me and are struggling to grasp the vastness of these quantities… well, buckle up.
The new supercomputer to be developed by the United States will operate at one “exaflop,” which is equal to one thousand petaflops.
That’s one quintillion calculations every second.
This computer will not just exceed the capabilities of Tianhe-2…
For lack of a better term, we’re about to blow Tianhe-2 out of the water — operating almost 20 times faster.
The order came just a few weeks after the TOP500 rankings, as the U.S. almost simultaneously put a stop to technology exports to Chinese tech centers associated with Tianhe-2. (Why help the competition, right?)
In response to the President’s order, China has stated that it will block exports of advanced drones and supercomputers — for the sake of their national security.
According to President Obama, this supercomputer will give us the solutions to large, unanswered questions about the economy, cyber security, and scientific mysteries.
Defying the Prophecy
There are a few objectives outlined in Obama’s executive order. Without going too in-depth, let’s look at just a few:
“Establishing, over the next 15 years, a viable path forward for future HPC (high-performance computing) systems even after the limits of current semiconductor technology are reached…”
Those “limits” refer to Moore’s Law.
Moore’s Law, as one researcher puts it, is the reason “your two-year old smartphone feels like a 1975 Ford Pinto.”
Rapid innovation makes digital devices cheaper, more powerful, and more energy-efficient than devices from the previous generation.
It means devices are replaced quickly, and the old ones become almost immediately obsolete.
The concern arises when we reach the end — when we can’t shrink semiconductors any further or make processors handle data any faster.
Does innovation just die?
No.
OEMs and component-makers are already planning for a post-Moore’s Law world, and apparently the United States’ government finally is, too.
The post-Moore’s Law era will mean alternative designs and solutions that have (until now) remained relatively unexplored.
The stated objective from the executive order means that the NSCI will be exploring beyond silicon semiconductors, searching for that new alternative design, and taking measures to ensure that their technology won’t be obsolete as soon as it’s born.
“Developing an enduring public-private collaboration to ensure that the benefits… are shared between the United States Government and industrial and academic sectors.”
This is far from the first time that the U.S. government has collaborated with private tech companies. In fact, supercomputing got its start in government labs with the help of private tech firms.
However, this collaboration might hit some road bumps, as the needs of the government often don’t completely align with the needs of the private sector.
One analyst admits, “It’s going to take a lot of thought to make sure they don’t mandate something that has no applicability in the marketplace.”
HPC Vice President at IBM, Dave Turek, also admits, “There’s always a balancing point: how do you begin to shape your design and architecture for the extreme needs of the government versus the benefits that would accrue into the industrial space?”
The bottom line for the private sector (and their investors as well) is whether or not initiatives like this one will be profitable.
So far, it seems that the usual suspects — IBM, Cray Inc., and Intel Corp. — are already in line for partnership.
Although the U.S. blocked Intel’s export of chips to the Tianhe-2, it did sign a $200 million agreement with the company to work on a new supercomputer, called Aurora, at a government laboratory.
Cray Inc., the other supercomputer giant, is also grouped into the deal, which was solidified before President Obama signed the executive order.
We have yet to hear of a completion date for the NSCI project.
However, 90 days after the executive order, the NSCI Executive Council is required to announce the details of its implementation plan, and that’s coming up in just a few days
Based on a history of relationships, we expect to see IBM, Cray Inc., and Intel involved with this as well.
Stay tuned.
{$custom_ptt_jessica_signoff}