Going somewhere | National Center for Supercomputing Applications at the University of Illinois
04.28.11 - Permalink
NCSA turned 25 in January. Bob Wilhelmson was at NCSA from the very beginning. Although he formally retired last year, he still is involved with the center. When the unsolicited proposal that gave birth to NCSA was written and sent to the National Science Foundation (NSF) in 1983, high-performance computing was being discussed by folks at NSF and in the research community, says Wilhelmson, but he and the other co-principal investigators on the proposal led by Larry Smarr "thought we'd move things along a bit."
That decision to move the discussion to its logical conclusion greatly impacted Wilhelmson's work as an atmospheric scientist at the University of Illinois. Like many campus researchers at the time, he needed to travel to the National Center for Atmospheric Research (NCAR) in Colorado to use its large computing systems. Or he'd submit jobs to NCAR via telephone modems and punch cards and then wait for output in the mail. Some researchers on campus were even traveling to Germany to do their computing. To them, that just didn't make sense.
Wilhelmson studies severe storms, looking, for example, at supercells (strong persistent storms), which produce the strongest tornadoes. When Blue Waters comes online, he believes he will finally be able to simulate these storms and any embedded tornadoes with enough fidelity to capture what is happening at the smallest scales within the storm and around the tornado that influence tornadogenesis, evolution, and decay. Atmospheric scientists currently cannot do this.
"And when we do that, we're not only simulating the tornado but we're simulating the whole storm with high fidelity. How the storm behaves as a whole also influences what is happening on the ground in terms of high winds, tornado structure and strength, and damaging hail," he explains.
In the beginning . . .
Simulations, even in the early days, produced lots of data that needed to be analyzed and visualized. Wilhelmson says that he would create two-dimensional contour plots of his NCAR data that were written to microfilm and mailed to him if he was at Illinois.
"Nobody even thinks of that today," he laughs. "Decisions were made early on that we were going to do things a little differently at NCSA. We needed new and faster ways."
This included animating the data using software developed at NCSA. In addition, new ways were developed to communicate scientific results to other researchers and the public. NCSA brought in people from the West Coast who were developing software and animations for commercials and movies. This has led to productions shown in planetariums, in museums, and visualizations in 2D and 3D IMAX films and television documentaries.
"Our thought was, why can't we use what they are doing and apply it to the data that we're producing on our supercomputers. And that's how we proceeded. Software was developed here and when coupled with high-end visualization software in use on the West Coast helped to lead the revolution in scientific visualization at the time. Further, there was a marrying of computing technology on the very large high-performance scale and the computing technology on one's desktop so that you could actually see and animate these visualizations on a computer screen in three spatial dimensions and time," says Wilhelmson.
Digital visualization of data produced during a simulation impacted his research. His team already had millions of pieces of data, "and then all of sudden computers get faster and there's billions and trillions of pieces of data we're trying to make sense of, both in terms of structure of storms and storm evolution." What was exciting about NCSA's approach, he says, was that he could see renderings of the data no matter what its size, and ask questions about it that could be answered quickly with new animations. Today, researchers carry out simulations and produce visual animations directly from the simulation data, which is made available through web browsers.
"When we started, there was this 'I wonder where we're goin', but I know we're not stayin' where we're at. We're goin' somewhere.' That was before Facebook, before the Internet as we know it today," says Wilhelmson. "There was an excitement when we started because we were all packed together in a building, we were young, we had dreams of what we wanted to do, and what NCSA as an organization should be."
The university and NSF didn't know NCSA was going to be a highly recognized organization 25 years later, he says, but they provided the backing and Illinois had the leadership and the scientists on campus who could carry the idea forward.
"The capability of NCSA is rooted not so much in the hardwarethat changes every few yearsbut in its people," says Wilhelmson. "It is taking innovative ideas and developing them so that the end user can carry out their research more effectively and answer questions that they only dreamed of answering a few years before."