Harvard Mark I

Harvard-IBM Mark 1, right side

The IBM Automatic Sequence Controlled Calculator (ASCC), better known as the Harvard Mark I or Mark I, was the first electromechanical computer, built at IBM and shipped to Harvard in 1944. It had 760,000 wheels and 800 kilometers of cable and was based on Charles Babbage's Analytical Engine.

The computer used electromagnetic signals to move the mechanical parts. This machine was slow (it took 3-5 seconds per calculation) and inflexible (the sequence of calculations could not be changed); but it executed basic mathematical operations and complex calculations of equations on the parabolic movement.

It worked with relays, programmed with switches, and read data from punched paper tapes.

Origin

The original concept was presented to IBM by Howard Aiken in November 1937. After a feasibility study by IBM engineers, company president Thomas Watson personally approved the project and its funding in February from 1939.

Howard Aiken had begun looking for a company to design and build his calculator in early 1937. After two failed attempts, he was shown a device that the son of Charles Babbage had donated to Harvard University 70 years earlier. This led him to study Babbage and add references to the Analytical Engine to his proposal; the resulting machine "carried principles of Babbage's Analytical Engine almost to full effectiveness, while adding important new features."

The ASCC was developed and built by IBM at its Endicott plant and shipped to Harvard University in February 1944. It began making calculations for the US Navy Office in May and was officially presented to the university on August 7, 1944, where he achieved great success.

Features

Mark I was a large machine, measuring about 15.5 meters long, about 2.40 meters high and about 60 centimeters wide and weighing approximately five tons. It had glass covers that allowed all the machinery inside to be seen.

The Mark I received its instruction sequences (programs) and data through perforated paper tape readers, and the numbers were transferred from one register to another by means of electrical signals. The internal wiring of the Mark I had a length of more than 800 kilometers, with more than three million connections. The results produced were printed using electric typewriters or card punches.

Although it had electromechanical components, it was an electric automatic machine. He was able to perform 5 arithmetic operations (addition, subtraction, multiplication, division and reference to previous results). Its interior was made up of 750,000 pieces of different varieties (rotary wheels for registers, relays...).

It was made up of more than 1,400 ten-position rotary switches on the front of the machine to display the values of the constant registers that were fed into it. In addition to the constant registers the machine contained 72 mechanical registers. Each of the mechanical registers was capable of storing 23 digits, the digits used for the sign were 0 for a positive sign and 9 for a negative sign.

The position of the decimal point was fixed during the solution of a problem, but could be previously set to be between any two digits. The machine also had mechanisms that allowed double-precision calculations (46 decimal places) to be carried out, through the union of two registers, in a manner analogous to Babbage's analytical engine.

Operation

The Mark I was programmed by receiving its sequences of instructions through a paper tape, in which the instructions and numbers were perforated and transferred from one register to another by means of electrical signals..

When the machine was running, the noise it produced was similar to a room full of people typing synchronously. The minimum time to transfer a number from one register to another and to perform each of its basic operations (subtraction, addition, multiplication and division) was 0.3 seconds. Though division and multiplication were slower.

The ability to modify the instruction sequence based on the results produced during the computation process was small. The machine could choose from several algorithms for the execution of a certain calculation. However, to change from one instruction sequence to another was costly, since the machine had to be stopped and the operators had to change the control tape. Therefore, it is considered that the Mark I does not really have hardcore jumps. Although, later what was called Subsidiary Sequence Mechanism was added (it was capable of defining up to 10 subroutines, each of which could have a maximum of 22 instructions), which was made up of three connection boards that were accompanied by three paper tape readers. And it was possible to affirm that the Mark I could transfer control between any of the readers, depending on the content of the records.