A Birthday Party for Eniac

By Philip Elmer-DeWitt.

A few weeks after Kay McNulty graduated from Philadelphia's Chestnut Hill College in 1942 with a degree in mathematics, she got a job at the Army's Ballistic Research Laboratory as a human "computer," calculating artillery trajectories. For three years she did the kind of mind-numbing mathematical drudgery--punching numbers into a mechanical calculator and copying down the results--that in those days was measured in "girl hours." Then she was invited by the University of Pennsylvania's Moore School of Electrical Engineering to help J. Presper Eckert and John Mauchly put the finishing touches on a new kind of computing device called ENIAC (for Electronic Numerical Integrator and Computer). That machine and its descendants were destined not only to make her old job obsolete but to change the world profoundly.

One day last week McNulty (now the widowed Mrs. John Mauchly), Eckert and 500 computer enthusiasts gathered at a "black-tie optional and hackerwear essential" party at Boston's Computer Museum to celebrate the 40th anniversary of the dedication of the first all-electronic digital computer. On that day in 1946, ENIAC in 20 seconds performed a mathematical calculation that would otherwise have required 40 girl hours to complete. Newspapers headlined the performance. It "solves the unsolvable," reported the Philadelphia Inquirer. Indeed, many experts mark ENIAC's feat as the beginning of the modern computer age.

By every measure, ENIAC was an imposing machine. It weighed 30 tons and occupied a space as large as a boxcar. Its 40 modular memory and processing units, each housed in a 9-ft.-high black metal cabinet and bristling with dials, wires and indicator lights, filled a room the size of a small gymnasium. Its 18,000 vacuum tubes radiated so much heat that industrial cooling fans were needed to keep its circuitry from melting down.

ENIAC was the technological wonder of its day. Programming the machine could take as long as two days as "coders" armed with detailed instructions fanned out among the panels, setting dials and plugging in patch cords in an arrangement that resembled an old-fashioned telephone switchboard. Data were fed into ENIAC in the form of IBM punch cards; a million cards were required for the monster's first assignment, a top-secret numerical simulation for the still untested hydrogen bomb. Every time a tube burned out, which happened twice a day at the start, a technician had to rummage among the tangle of wires to locate and replace the dead component. To prevent rodents from nibbling at ENIAC and destroying vital parts, Eckert recalled at the anniversary party, the scientists captured some mice, starved them for several days and then fed them bits of the insulating materials used in the machine. Any pieces the mice seemed to favor were removed from ENIAC and replaced with less tasty parts.

Other computing machines of that era solved mathematical problems in one of two ways. Either they estimated the solutions, using mechanical analog devices like slide rules and differential gears, or they computed them digitally, using the on-off action of slow-moving electromagnetic telephone switches. ENIAC, by contrast, was the first digital computer both to store and to process information with vacuum tubes; as a result, it was able to perform calculations 1,000 times as fast as its electromechanical predecessors. "I was convinced that you could produce great speeds electronically if you put your mind to it," says Eckert. "ENIAC proved that this was the way to go."

Anticipating the gung-ho spirit of their spiritual successors in Silicon Valley, the ENIAC team members worked with demonic intensity. "Eckert was completely devoted to the machine," recalls John Grist Brainerd, the project director. "He would work on it day and night, and worry, worry." Two cots were installed on the ground floor of the Moore School so that the exhausted computer scientists could rest near their cherished machine. "When it finally turned on, everyone was elated," recalls Kay Mauchly. "It seemed like every day was a happy day."

Those happy days soon came to an end. A month after the ENIAC's public unveiling, Eckert and Mauchly resigned rather than turn their patent rights over to the university. Five years later they developed the first commercial computer, UNIVAC 1, but business reversals forced them to sell their fledgling computer company to Remington Rand. The final insult came in 1973. Seeking to invalidate Mauchly and Eckert's patent for "the" electronic computer, Honeywell convinced a federal judge that Mauchly had based his ideas for ENIAC on the work of a computer pioneer named John Atanasoff. The patent was dismissed, and Mauchly and Eckert lost legal claim to one of the great inventions of the 20th century.

ENIAC was decommissioned in 1955, having churned out military and scientific calculations for nearly a decade. Today its cabinet-size modules are scattered among several museums and institu tions. Four remain at the Moore School, gathering dust and cobwebs in a foyer off the old building's main hallway. Nearby, someone has hung a contemporary computer chip and a sign that says it all: "In less than 40 years, advances in microelectronics technology have enabled the digital computer with performance far superior to the ENIAC to be placed on a onequarter-inch piece of silicon."

With reporting by Timothy Loughran/ Boston and Lawrence Mondi/ Philadelphia