Printed from the Urban Tulsa Weekly website: http://www.urbantulsa.com

POSTED ON SEPTEMBER 21, 2011:

We Don't Have To Be Muggles*

Supercomputing -- A wizard gambit for T-Town's economy

By RAY PEARCEY

Transcendent Man is a magnetic, 2009 documentary film directed by Barry Ptolemy. The project is a much-anticipated look at the life and work of serial inventor, futurist and scientist Ray Kurzweil.

Kurzweil invented one of the first flat-bed scanners, created an early text to speech reading device for the blind and fashioned a pioneering music synthesizing machine. Called the "ultimate thinking machine" by Forbes magazine and the "restless genius" by the Wall Street Journal, Kurzweil is a phenom. "Transcendent" follows the ever-fascinating Kurzweil as he travels around the globe pushing kernels of the wild, but heavily thought out notions at the heart of his New York Time's best-selling book: "The Singularity is Near: When Humans Transcend Biology".

Kurzweil sees a time when already in play, and accelerating advances in artificial intelligence, synthetic biology, ultra high-speed computing, sensor systems and robotics and material science will hugely augment human creative, intellectual, engineering and scientific capacity.

This epic transformation is what Kurzweil calls "the Singularity". We are still a bit away from the Singularity -- but not all that far -- according to Kurzweil who has many superbly accurate tech forecasts to his credit. A big part of the road to the Singularity is tethered to extremely high-speed computing, and affordable, wide scale access to this panoramic toolset.

OII & The "Super"

There is a game changing project in the works in T-town: it's not a major league baseball franchise or the launch of a high-speed rail project or an announcement about Twitter Corp's relocation to Green Country, but it has the heft, the dimension and the impact of any one of these things. And it might give us peerless company retention; venture attraction and a Tulsa bright kid "boomerang" cred unlike any seen in these parts.

The Oklahoma Institute for Innovation (OII) and its university and private partners are a rare thing, a cross cutting, focused catalyst that might help Tulsans get where we need to go much faster. OII has several projects: including one of signal import for Tulsa's future, one they are still hustling to pull off -- they call it the Tulsa Community Supercomputer, it is over four years in the making, is a boundary buster, and of great strategic significance -- and it comes with a surprising affordable $6.7 million hardware price.

A supercomputer is an extremely high performance computation engine, or an array of lots of conventional processing devices ganged together as a seamless, ultra capable computational gizmo. Another way of thinking about a Super: they are simply the fastest, most capable computing machines available at any given time. Supers have been used traditionally for nuclear weapons simulation, seismic exploration, particle physics, aerospace prototyping and climate modeling.

Having a publicly available "Super" would give Tulsa researchers/scientists, designers, engineers, entrepreneurs and artists a wild card advantage in our region.

Barry Davis, a veteran Tulsa venture capitalist and David Greer, the chief of TU's nationally renowned information security technology program and the acting director for the new project are ramrodding this giant effort: they are being assisted by Alex Barclay -- a TU grad student and compsci guru. The Supercomputing Center concept, sketched out in a recent Robert Evatt piece in the Tulsa World, has received a good response, Greer and Davis report, from Tulsa's university and private community.

There is a real challenge going forward, however: Tulsa' Super Center has to be economically viable, deftly designed for a wide spectrum of potential users and sufficiently advanced to be a leapfrog effort for our region. Davis and Greer are, rightfully, not interested in a "me too effort" or a pale, cookie cutter project.

There is another supercomputing facility Oklahoma at Norman/OU: this facility is designed for and dedicated to weather/climate research in accord with use constraints that brought it to OU. OII and crew want to create a general purpose, widely available operation in Tulsa. By having the system, which would actually be a linked array of from 80 to 100 high performance processors in a single-space, all of the overhead and facility management costs, which are considerable, can be made manageable for potential users. OII is looking at using space in the new City Hall/One Tech building in downtown Tulsa.

The One Tech/City Hall building is ideally suited -- it has superb networking, energy management, cooling and electric power setups -- all optimal for the Super project and they have Mayor Dewey Bartlett's enthusiastic endorsement.

A Whole New Way of Doing Science/Tech

Development

Another part of the rapidly evolving picture is the explosive character of a lot of contemporary science and technology development. The stupendous speed of state of the art supercomputers allow scientist and technologists to examine huge data sets, vast numbers of hypotheticals, millions of combinations and hybrid theories and to do "pattern matches" that couldn't be done in the absence of vast computational power.

Let's make things a little more tangible: imagine that you're CEO at a small engineering company. And you have struggled like crazy and successfully bagged a multimillion-dollar NASA contract to design a space born antenna -- an antenna that will travel hundreds of millions of miles atop a planetary probe. You antenna/communication array will provide the only command/control and data return link to Earth for a $400 million, 2.5 year automated mission to Mars. Imagine that your objective is to build the antennae using the best materials available and fashion an antenna geometry both keeps NASA's robotic spacecraft in nearly continuous communications with the Earth and allows it to avoid mission ending overexposure to the Sun and cosmic radiation bursts.

Imagine further that there are, say 10,000 combinations of materials, component systems and geometries, available for crafting this antenna. You and your team are only 30 strong and don't have the eyes, ears and brains, the time or the money to explore all 10,000 possibilities. You elect to solve this problem by building a rich computer model, and commit to run 10,000 simulations -- sorting through all the design possibilities using everything known about the environment that the spacecraft occupy, its position along the trip path, material dynamics, thermal kinetics and an ensemble of tactics for maintaining contact with Earth.

Your super-computer first identifies hundreds of prospects, and then though a rigorous Darwinian like pruning process, green lights a few dozen especially promising antenna designs. This allows your team to spend all it's precious time building a "Goldilocks" batch of optimal prototypes. Agile computer modeling also empowers you to spot designs that might've been overlooked because your team might allow blind spots, dead end experiences, fatigue or boredom to get in the way of producing a superior solution.

Evolutionary Engineering

Supercomputing is radically re-fashioning the scientific, engineering and bio-systems landscape: one fevered line -- the new design/production path called evolutionary engineering. To do evolutionary engineering you need a machine that can do the trillions of computations required to check out our imaginary antenna designs and help you pick a final: at the moment, if you and your firm are in Tulsa today, you have only have one choices for getting this simulation done -- contract it out at huge expense -- this needs to end.

There are lots of other examples: finishing an animated film in a few weeks as opposed to months giving local filmmakers/animators a competitive advantage that few small players currently enjoy: crafting a novel aircraft wing/airfoil design with minimal engineering staff, or reworking Tulsa's flood plain in the wake of climate change dynamics -- the project list is large.

Training Super-computer Wizards: Beyond Muggle Work

Teaching super computer management, programming and applications is a unique facet of what OII and crew are proposing -- a curriculum offering that would be tightly linked to Tulsa Community College. This is an educational offering that would train a cadre of folks to manage the world's most powerful and important computing machines: a program that won't require undergraduate or graduate degrees. The super computing training gambit is close to the German industrial/manufacturing apprenticeship concept -- one that is a humongous boon to the German economy and to high-end German companies like BMW and Mercedes Benz.

The Other OII Efforts

The Oklahoma Innovation Institute is more than a determined, four-year plus effort to secure a supercomputer platform for T-town -- it is an interesting mix of critical efforts that we need if we want to be players in Kurzweil's singularity. In a follow up piece I'll do in a couple of weeks, I'll highlight additional, ultra cool initiatives they have put together and talk to some Tulsa locals who lust for super computing access.

*In the Harry Potter universe, a muggle is a person or clan without magical powers or any special élan.

Send all comments and feedback regarding Cityscape to rpearcey@urbantulsa.com

URL for this story: http://www.urbantulsa.comhttp://www.urbantulsa.com/gyrobase/Content?oid=oid%3A42740