High-end computers meet everyday products

Microchips for potato chips
High-end computers meet everyday products

JADE BOYD
Rice News staff

Laundry detergent, diapers and potato chips don’t immediately evoke images of gigantic supercomputers, but without them, many of the products consumers routinely buy would never make it to store shelves.

Proctor & Gamble (P&G) Corporate R&D Director Tom Lange blended eye-popping visuals, marketing 101 and gee-whiz science to wow more than 100 listeners in his Oct. 30 lecture, ”Will Pringles Fly?” which explored the way supercomputers are used in bringing day-to-day products to market. Sponsored by the Dean’s Office in the George R. Brown School of Engineering and the Computer and Information Technology Institute, Lange’s talk drove home the point that engineers of the future must be prepared to use supercomputers to model designs, streamline production and guide corporate decision-making.

Lange shattered the notion that everyday products involve everyday manufacturing when he revealed that P&G produces 1 billion disposable Pampers diapers every two days. The machines required for this Herculean task are more complex than a jet aircraft, he said, and even incorporate software from the nation’s top defense labs.

As for Pringles, the billion-dollar brand of potato chips referenced in Lange’s title, P&G produces 1 billion of them every two hours.

”Will Pringles fly?” Lange asked. ”They come down the line so fast they fly like shingles off of a roof in a hurricane.”

Modeling the precise aerodynamic properties of the chips allows engineers to control the fast-moving Pringles and even to make sure that cheese and other powdered flavors are spread evenly across each chip.

Lange told the audience that P&G owns several high-performance computer platforms, including a one-teraflop system, which can perform 1 trillion floating-point operations per second (FLOPs), the standard measure of supercomputer performance. All of these machines run around-the-clock in support of P&G’s $2 billion annual research and development effort.

P&G’s modeling and simulation group is tasked with streamlining production of dozens of products, ranging from laundry soap to perfumes and batteries to coffee, he said.

Lange told students and instructors in the audience that the time is rapidly approaching when the average engineer will need to access high-performance computers on a daily basis, and he said the greatest thing holding the industry back is the lack of applications

that run on systems using parallel processors.

”A good analogy would be air travel,” Lange said. ”Flying is routine in business today, and we need to develop the applications that engineers will use to make high-performance computing routine as well. If this was aviation, then modeling and simulation would be at about 1935.”