UO grid network will help research.

Byline: Greg Bolt The Register-Guard

Imagine an ultra-fast computing system that works something like the electrical grid, where people just plug into a wall socket and get all the processing power, software and storage space they need from a central computing "plant."

That basic idea, but in a more specialized way, is what's behind a new, high-performance computing network at the University of Oregon.

The concept is known as grid computing, and the UO won a $1 million grant to install a computer grid that will speed up research into how the human brain and computer brains work.

Ultimately, the grid may help scientists speed the diagnosis of a variety of brain disorders, including epilepsy, attention deficit disorders and depression.

The system is known as the ICONIC grid installation, for Integrated Cognitive Neuroscience, Informatics and Computation, and it gives the UO a neuroscience computing resource equalled by perhaps no more than a half dozen institutions in the country.

The system also has the potential to make the university part of what could become a nationwide computing network built on the grid principle.

"The ICONIC grid is a major resource for the University of Oregon's scientific research," said Allen Malony, director of the UO's Neuroinformatics Center and a professor of computer and information sciences.

"These machines are very powerful, and we are trying to develop the grid in a such way that it can be extended not only within the university but also within the state and within a national computational grid."

For now, the computing grid will be used in cognitive neuroscience and neuroimaging, in neuroinformatics and in computer science research.

Neuroinformatics is a relatively new field that looks for new ways to store, process and analyze the vast amounts of data generated by neuroscience.

Also, the grid offers research opportunities in an area of computer science known as high-performance parallel processing, which is Malony's area of expertise.

Malony hopes to use the grid to develop new tools and programs to speed scientific analyses.

The UO grid consists of three powerful IBM servers running Linux operating systems, each tailored for one of the three main research areas.

But the servers are connected via a high-speed network and can share processing power, so they can essentially act as a single, high-performance computer to tackle large amounts of data in a fraction of the time it takes now.

The system also includes a five terabyte storage array to hold the massive amounts of data generated by neuroimaging and neuroscience.

Funding for the computer grid comes from the National Science Foundation.

The university is in the process of installing and connecting the computers, and expects to have them available to researchers by late spring.

Landing the NSF grant is another step forward for the university's broader Brain, Biology and Machine initiative, which aims to make the UO a leader in brain function research and the analysis of the associated data.

The university has so far attracted more than $40 million in funding for the initiative.

The grid will be of immediate help to researchers at the Lewis Center for Neuroimaging, where the university's powerful MRI scanner allows them to map areas of the brain associated with functions such as speech, language and attention.

Ray Nunnally, the center's director, said it will allow them to do in minutes things that now can take days.

Using the MRI, researchers can build a three-dimensional picture of the brain and identify areas that are activated when it performs different tasks.

That can involve imaging about 180 "slices" of the brain twice a second, and processing that much data now takes almost two days.

"We want to get that down to 10 minutes as opposed to 40 hours," Nunnally said.

With that kind of speed, research will take place at a faster pace, allowing scientists to get results and move on to the next step more quickly.

"It will allow for integrated neuroimaging solutions to be applied to the study of human brain function," Malony said. "These are all exciting research areas that will benefit from having higher resolution, integrated neuroimaging capabilities."