AMES, Iowa -- High-performance parallel computing doesn't have to cost a fortune. And a team of researchers in the Scalable Computing Lab (SCL) at the U.S. Department of Energy's (DOE) Ames Laboratory is well on the way to proving that. They are determined to make traditionally expensive supercomputing capabilities more economical and attainable for the scientific and educational communities.
The researchers, Mark Gordon, David Halstead, John Gustafson, Steve Elbert, Don Heller, Dave Turner and Bruce Harmon, are networking personal computers (PCs) to create PC clusters that operate at speeds comparable to today's best parallel computers, and for a fraction of the cost.
Gordon, Ames Lab program director for Applied Mathematics and Computational Sciences, says one of the ways they plan to introduce the clustering technique to others is by devising a "cluster cookbook" for the world wide web that will provide guidelines on how to construct cluster networks. In addition, they will develop and host a hands-on PC cluster workshop in the fall of 1998.
"What precipitated the clustering effort is that in the last few years manufacturers of personal computers have been making them with speeds that are equivalent to workstations," says Gordon. "This means that you can use individual computers for serial high-performance computing. You can put on one $3,000 personal computer what you used to put on a high-performance workstation, and it will perform just as well for you.
"What's hard is going the next step -- networking clusters of these PCs together to make a true, parallel, high-performance computer that's competitive with boxed systems that cost millions of dollars," Gordon explains.
Parallel computers have multiple processors simultaneously working on different parts of one problem, making them far more efficient and able to handle more complex problems than sequential computers, which tackle a problem one step at a time.
The SCL researchers are achieving parallel power by networking today's high-performance and relatively low-cost personal computers -- harnessing their individual power in clusters of eight, 16, 32, and 64 to create parallel computers.
Six PC clusters already operate in various areas of the Ames Lab. Networks of eight PCs are doing materials simulations, while other clusters are performing such tasks as modeling new materials with desirable magnetic properties. And recently SCL staff constructed a cluster of 64 PCs, each with two central processing units, and are now testing its ability to perform parallel computations.
"At the end of the day, this thing should be four times as fast and have four times the storage capacity and memory as the largest supercomputer in the SCL, which cost just shy of a million dollars," says Halstead, assistant scientist. "So it's four times as fast for a third of the price."
Gordon continues, "If you use a computer benchmark to evaluate our clusters against one of the best parallel computers on the market today and then go the next step and divide that by the cost to get the price/performance ratio, the clusters blow everything else away. And that's our point -- not to show people how to do really great parallel computing because lots of people can do that, but to show them how they can do it in a very cost-effective way."
The members of the PC cluster team view themselves as the logical ones to demonstrate the feasibility of doing this type of low-cost parallel computing. They helped organize the first DOE cluster workshop, which was held in Des Moines last April and drew over 50 pioneers in this field from other national labs and universities.
"What we're trying to do is bring high-performance computing into university departments, individual research groups and the classroom," says Gordon.
Halstead adds, "We want to make this type of computing accessible to smaller groups so they can buy into it and scale it up as more money comes in."
Ames Laboratory is operated for the DOE by Iowa State University. The Lab conducts research into areas of national concern that include energy resources, high-speed computer design, environmental cleanup and restoration, and the synthesis and study of new materials.
Released: Feb. 16, 1998
Contacts:
David Halstead
Saren Johnston, (515)
294-3474
Last revision: 4/21/03 kbg