Mouse over for marker text
Name:
ENIAC
Region:
Philadelphia and its Countryside/Lehigh Valley
County:
Philadelphia
Marker Location:
Chancellor St. between 33rd and 34th St.
Dedication Date:
June 15, 2000
Behind the Marker
During World War II, the American Army first experienced the offensive sting of the German Afrika Korps in the deserts of North Africa in the winter of 1943. There, the rout at Kasserine Pass uncovered dangerous inadequacies in American training, materiel, and weaponry. The rivets in American tanks turned into shrapnel when German shells penetrated their armor. And nothing in the United States' arsenal could contest the famed and feared 88-millimeter German gun.
American artillery commanders also noticed another flaw in their guns. The intricate gunnery tables that were supposed to show where shells would fall were often wrong. Usually fire controllers knew where shells would land based on calculating the angle of the barrel, atmospheric conditions, the speed of the shell leaving the barrel of the gun, the weight or type of shell, and the amount of powder. Gunners finally realized that it was the sandy desert soil that caused their guns to recoil differently than the test guns. Most of the data had been generated at the Army's ballistics laboratory in Aberdeen, Maryland, where the soil was much denser. Firing tables that represented thousands of hours of laborious calculations were worthless.
Gunnery table errors posed grave problems for gunners in the field. They were also disconcerting to the men and women "computers" at the University of Pennsylvania's Moore School of Electrical Engineering, who performed the tedious calculations contained in the tables. It took 750 multiplications to calculate a single trajectory, and most tables contained 2,000 to 4,000 of these. Even using a mechanical calculator device, called a differential analyzer, it took a month to complete the thousands of trajectories. With already completed tables no longer accurate, human and mechanical calculators struggled to keep up with the growing backlog.
For U.S. Army ordinance officer Captain Herman H. Goldstine, the situation was grave. Working diligently to improve the computing division at Philadelphia, he had molded a motley assortment of engineers, math majors, and a smattering of Ph.D.s into an efficient operation capable of producing high-quality tables. Now that the old tables needed correction, the growing backlog of gunnery tables made him nervous. One day in March 1943, he muttered anxiously about the stacks of gunnery tables to a young assistant laying cables for the computing machines. The assistant joked that Goldstine should go upstairs and talk to a young physicist from Johns Hopkins University named John Mauchly, who had an idea for a machine much better than any mechanical analog calculator.
Though it was news to Goldstine, Mauchly's idea for an electronic computer had first appeared in an August 1942 memorandum circulated throughout the Moore School. A young University of Pennsylvania electrical engineering graduate student twelve years his junior, J. Presper Eckert, Jr., had convinced Mauchly that using vacuum tubes instead of mechanical relays for the necessary addition, multiplication, division, and subtraction circuits would improve the speed of the hypothetical device. Based on Eckert's advice, Mauchly figured that the machine could finish a complete table in under two minutes.
For Goldstine, the hiring of more and more human "computers" was no longer an option. Goldstine told Mauchly if the ordnance department was willing to throw a million dollars at General Motors for a prototype it didn't use, "why not spend equal or similar amounts of money on trying out an electronic computer?"
Not everyone within the U.S. research and development community believed in the feasibility of an electronic computer. Some wanted to develop more sophisticated mechanical differential analyzers. Others thought that a machine of such complexity, requiring thousands of sensitive vacuum tubes, was inherently unreliable. Despite the concerns, in May 1943, the Moore School team received its first contract of $61,700.
Eckert, who had spent his summers designing measuring devices for industrial applications, provided the engineering expertise needed to turn Mauchly's theoretical ideas into a functioning machine. On paper, the team developed the computer's schematic. It would have several–eventually thirty–"modules," which would perform separate computational roles and "speak" to each other through electrical pulses. At the core of the machine were the accumulators, which added and multiplied large numbers.
Unlike mechanical computers, which performed their computations by mechanically rotated metal rings, the accumulator would use neon vacuum tubes to send fast moving electrical signals between contiguous "rings." The new machine's read-only memory allowed it to simply "look up" a value in a table instead of performing the computation. The computer, however, was not programmable. Instead, technicians had to unplug connecting wires and reconfigure connections for each application, and input and output Information via IBM card-punch machines. When completed, the new, $500,000 Electronic Numerical Integrator and Computer (ENIAC) contained 17,000 vacuum tubes and took up 1,800 square feet of floor space, an entire room on the first floor of the Moore School at the corner of 33rd and Walnut Streets in West Philadelphia.
With its ability to perform 5,000 additions and 300 multiplications per second, ENIAC was much faster than any mechanical calculator. But design and construction took so much time that ENIAC was not completed until December 1945, four months after the official end of World War II. In 1947 the Army moved it to Aberdeen, Maryland, where it continued to provide calculations for ballistics tests, the atomic weapons program, and the first numerical weather predictions by computer. There, ENIAC's deficiencies soon prompted Eckert and Mauchly to find a way to improve ENIAC's memory.
In 1946, Mauchly and Eckert left Penn to start their own commercial computer company. Their Eckert-Mauchly Computer Company would build both of ENIAC's successors:
BINAC and UNIVAC I.
J. Presper Eckert and J.W. Mauchly working on the Electronic Numerical Integrator...
American artillery commanders also noticed another flaw in their guns. The intricate gunnery tables that were supposed to show where shells would fall were often wrong. Usually fire controllers knew where shells would land based on calculating the angle of the barrel, atmospheric conditions, the speed of the shell leaving the barrel of the gun, the weight or type of shell, and the amount of powder. Gunners finally realized that it was the sandy desert soil that caused their guns to recoil differently than the test guns. Most of the data had been generated at the Army's ballistics laboratory in Aberdeen, Maryland, where the soil was much denser. Firing tables that represented thousands of hours of laborious calculations were worthless.
Programmers Betty Jennings and Frances Bilas (right) arranging the program settings...
For U.S. Army ordinance officer Captain Herman H. Goldstine, the situation was grave. Working diligently to improve the computing division at Philadelphia, he had molded a motley assortment of engineers, math majors, and a smattering of Ph.D.s into an efficient operation capable of producing high-quality tables. Now that the old tables needed correction, the growing backlog of gunnery tables made him nervous. One day in March 1943, he muttered anxiously about the stacks of gunnery tables to a young assistant laying cables for the computing machines. The assistant joked that Goldstine should go upstairs and talk to a young physicist from Johns Hopkins University named John Mauchly, who had an idea for a machine much better than any mechanical analog calculator.
T. Kite Sharpless repositioning ENIAC cables, Philadelphia, PA, February 2,...
For Goldstine, the hiring of more and more human "computers" was no longer an option. Goldstine told Mauchly if the ordnance department was willing to throw a million dollars at General Motors for a prototype it didn't use, "why not spend equal or similar amounts of money on trying out an electronic computer?"
Not everyone within the U.S. research and development community believed in the feasibility of an electronic computer. Some wanted to develop more sophisticated mechanical differential analyzers. Others thought that a machine of such complexity, requiring thousands of sensitive vacuum tubes, was inherently unreliable. Despite the concerns, in May 1943, the Moore School team received its first contract of $61,700.
Vacuum Tubes of ENIAC
Unlike mechanical computers, which performed their computations by mechanically rotated metal rings, the accumulator would use neon vacuum tubes to send fast moving electrical signals between contiguous "rings." The new machine's read-only memory allowed it to simply "look up" a value in a table instead of performing the computation. The computer, however, was not programmable. Instead, technicians had to unplug connecting wires and reconfigure connections for each application, and input and output Information via IBM card-punch machines. When completed, the new, $500,000 Electronic Numerical Integrator and Computer (ENIAC) contained 17,000 vacuum tubes and took up 1,800 square feet of floor space, an entire room on the first floor of the Moore School at the corner of 33rd and Walnut Streets in West Philadelphia.
With its ability to perform 5,000 additions and 300 multiplications per second, ENIAC was much faster than any mechanical calculator. But design and construction took so much time that ENIAC was not completed until December 1945, four months after the official end of World War II. In 1947 the Army moved it to Aberdeen, Maryland, where it continued to provide calculations for ballistics tests, the atomic weapons program, and the first numerical weather predictions by computer. There, ENIAC's deficiencies soon prompted Eckert and Mauchly to find a way to improve ENIAC's memory.
In 1946, Mauchly and Eckert left Penn to start their own commercial computer company. Their Eckert-Mauchly Computer Company would build both of ENIAC's successors:
BINAC and UNIVAC I. 