IBM opens Deep Computing Institute
May 24 — Big Blue, the nickname for International Business Machines Corp., the world’s largest computer maker, is setting out to prove not only how wide, but how deep, its computing prowess has become.
IN THE PAST YEAR, IBM has begun to lay out its vision of the future based on “pervasive computing” — making computers more widely available by embedding them in everyday objects —and “deep computing” — using supercomputer power to untangle some of the world’s thorniest computing challenges.
On Monday, the Armonk, N.Y.-based company said it will unveil its Deep Computing Institute, a $29 million research effort to join academic and industry researchers in a bid to use massive computational power to attack complex problems.
Deep computing techniques involve using supercomputer-scale machines, advanced software and sophisticated mathematical formulas to enable researchers to take on daunting scientific and commercial tasks that can involve trillions of variables.
The term “deep computing” was inspired by IBM’s Deep Blue chess-playing computer, which defeated world chess champion Garry Kasparov in 1996.
William Pulleyblank, director of mathematical sciences at IBM
Research, will serve as director of the institute. The institute will be
guided by an advisory board of leaders from universities, government
laboratories and corporations.
IBM is already a top maker of supercomputers — supplying more than 100 of the world’s 500 largest supercomputers — but the company is pushing to widen the use of the technology beyond academia to a range of new business problems
“Deep computing combines the best of business and scientific computing techniques to find the value buried in all this data and to apply that information to solve real-world problems,” Pulleyblank said in a statement detailing the plan.
“Deep computing...can help find the proverbial needle in an entire field of haystacks, analyzing vast reservoirs of data to uncover key relationships between gene sequences needed to understand disease,” he said, using a biotechnology example.
Supercomputers are the world’s fastest computers and allow researchers to throw maximum data-crunching power at a single research problem, or set of problems. Mainframes, their less powerful commercial siblings, handle general purpose calculations but are less suited to such intensive research.
IBM is already a top maker of supercomputers — supplying more than 100 of the world’s 500 largest supercomputers — but the company is pushing to widen the use of the technology beyond academia to a range of new business problems.
IBM is committing more than 120 scientists and technologists in eight research labs in New York; San Jose, Calif.; Austin, Texas; Tokyo; Zurich, Haifa, Israel; Beijing and New Delhi to the Deep Computing Institute project.
They will collaborate initially on deep computing projects in areas ranging from how to schedule personnel in complex environments, such as airline flights, and the modeling of precise weather patterns.
Deep computing techniques are spurring advances in computer optimization, simulation, visualization, and advanced pattern matching and discovery in fields ranging from financial investment risk analysis to drug research to oil discovery.
“Thanks to the tremendous advances in computing power and mathematical algorithms, it’s now possible to tackle problems of unbelievable complexity — things we couldn’t dream of doing even a few years ago,” he said.
As part of the project, the Deep Computing Institute will publish over the Internet the IBM Visualization Data Explorer, a powerful software package that can be used to analyze and create three-dimensional representations of data.
Data Explorer uses computational and 3-D graphics rendering tools in a programmable framework that allows computer users to rapidly create visual images, uncovering patterns, trends, and ’what-if’ scenarios hidden within highly complex data.
The product is used by a multinational bank to identify and manage financial risk in portfolios worldwide. In energy exploration, it can depict oil-flow simulations to improve drilling success rates and boost underground reservoir yields.
The underlying programming code for IBM Data Explorer will be given away as “open source” software to researchers at the institute’s Web site, beginning May 26. The site will be located at http://www.research.ibm.com/dci/software.html.
Gina "Nanogirl" Miller
"Nanotechnology: solutions for the future."