AREA NEWS
Story last updated at 1:05 p.m. on Tuesday, July 27, 1999

photo: news

  Richard Alexander, seated, a systems programmer at Oak Ridge National Laboratory, and Thomas Zacharia, head of ORNLs Computer Science and Mathematics Division, check out the lab's new IBM supercomputer.
-- Staff photo by Kelley Scott Walli

Sci/Tech: Growing by leaps and bounds
ORNL computing: Moving at the speed of byte

by Larisa Brass
Oak Ridger staff

There's nothing that provides us with a greater reminder of how far technology has come in the past 50 years than computers.

They started out as room-sized behemoths, barely able to store a few numbers at a time. They grew to be compact, lightning-fast and available to nearly every household.

Closely tied to the history of computers is the history of Oak Ridge, where early on they became an integral part of research here.

In 1953, Oak Ridge National Laboratory received its first computer, the ORACLE (for Oak Ridge Automated Computer and Logical Engine.)

One of two "homemade" computers, built at Argonne National Laboratory in Chicago, the ORACLE was so big that portions of the wall had to be removed to install it in ORNL's 4500 building, which has remained the stronghold for the lab's main computers ever since.

"At that time, (the ORACLE) was the fastest computer in the world," says Betsy Riley, assistant to the director of the Computer Science and Mathematics Division.

Until the ORACLE arrived, computations had been worked by a math panel headed up by famed mathematician Alston S. Householder or the researchers themselves.

A computer held the potential of figuring complex problems in a few minutes that could take researchers months to solve on their own.

But compared with today's user-friendly machines, the ORACLE was no quick study.

"There was no compiler at that time," remembers Henry Levy, who came to the laboratory as a crystallographer in 1943 and became one of the first to learn how to use the ORACLE. "One had to write in machine language. You had to learn the language."

"I still remember some of the commands," says Bill Busing, also a retired crystallographer who arrived at ORNL in 1954 and learned the ropes of the ORACLE from Levy. "I believe two, four was clear and add ... Two, two was subtract ... ."

One step up from having to write directly in binary code, researchers would punch out the programs -- combinations of numbers and the letters A through F -- onto paper tape and submit them to the ORACLE operator at their assigned time slots -- which could land in the middle of the night.

The ORACLE contained a five kilobyte memory -- or 1/200th of a megabyte, says Busing.

"And you may have 100 megabytes on that machine over there," he says, gesturing to a modern PC sitting on the other side of the room.

"One of the early programs I wrote had about 100 words. That's about 500 bytes in the program itself," says Busing. "And you loaded that in the memory and hoped the space would be enough to store temporary numbers and things like that. Pretty simplistic, huh?"

"Times have changed, haven't they," says Levy.

Indeed they have.

In April, a new supercomputer about the size of a tall stereo cabinet was delivered to the lab to replace nearly 10 generations of computers that have come and gone since the ORACLE.

When upgrades on the IBM RS/6000 SP are complete, it will operate at one teraflop of speed -- that's a trillion calculations per second -- and have over 300 gigabytes of memory. A gigabyte is about a thousand megabytes.

All this is made possible by the gigantic leaps in technology that have characterized computing since it first began.

The development of the network, the microprocessor, the Internet and parallel processing have brought about more ways to use the computer than anyone could have ever imagined in the beginning.

"The chairman and CEO of IBM in the (19)50s, Tom Watson, is famous for a statement he made to the world after IBM produced one of its first big machines," says Ron Leinius, a retired computer scientist who oversaw the computer program at all three local Department of Energy sites.

"He said that they had carefully calculated the need for computation in the United States, and they figured six or seven computers would be required before saturation."

"This is really a very interesting point," says Busing, "because what it really illustrates is how poor human beings' imagination is."

We've come a long way, baby.


All Contents ©Copyright The Oak Ridger