Inventions of George Stiebitz
George Robert Stibitz (George Robert Stibitz) - an outstanding American scientist and physicist, who is considered one of the creators of modern digital computers. He worked as a researcher at Bell Labs, a major research center in telecommunications, electronic and computer systems. In the 30s-40s, Shtibitz was engaged in the successful implementation of the provisions of Boolean logic, using electromechanical relays as switches. In 1937, the scientist assembled the first in the US electromechanical circuit performing the operation of binary addition.
George Robert Stibitz (April 30, 1904 - January 31, 1995)
')
The idea of ​​using electromagnetic relays to create a counting machine of the new model emerged from Stibitsa spontaneously when he was at home. Without losing time, the researcher began assembling a new calculator right in his kitchen. At the same time, Schthitz used all the handy materials: boards, tin cans, a tobacco box, a lantern bulb, a pair of relays and wires. From all this set, he managed to collect a primitive circuit, which could even add two binary numbers and demonstrate the result of addition. George called the homemade car Model K, where k came from the "kitchen" - the kitchen (in honor of the place of "birth" of the machine).
Model K workflow
Shtibitz was sure that on the basis of the relay it is possible to create a device capable of performing sequential calculations and storing their intermediate and final results. In particular, such a machine could multiply and divide complex numbers, since these operations took a lot of time from the staff of his department who were involved in the development of amplifiers and filters.
Model K implementation
The bosses of Bell Labs approved the project and began the development of a computing device. During the period from 1939 to 1940, Shtibitz, together with colleague engineer Samuel B. Williams (Samuel B. Williams), created a device that skillfully added complex numbers and performs subtraction, multiplication, division operations. Shtibitz was the architect of the car, and Williams was the chief engineer. The invention was called the CNC (The Complex Number Calculator) complex number calculator, also known as the Model I. The device was demonstrated at Durmouth College (although the calculator itself was in New York). During the presentation, CNC was the first to use remote access to computing resources. The communication was carried out using teletype on special telephone lines.
From the memories of Stibitsa:
In Model I, there were only 450 bipolar and ten multipolar relays, which served to store the input data and intermediate results. Arithmetic was used with a comma fixed before the first significant digit of the number. The coding (“Stiitz-code”) of the decimal digit was performed using four relays so that each digit n was represented by the binary code n + 3. It simplified the execution of transfer and subtraction operations.
Table with Stibits code values ​​for decimal digits (their direct code is also indicated for comparison):
CNC used hardware control if more than one relay was triggered in the top five or in the high order. The control circuit gave an error signal.
Model I
Model I worked with 10-bit numbers, but only eight digits were printed (the rest served to round the result). The device was a non-programmable machine with a clearly defined sequence of actions. Each subsequent operation began after the completion of the previous one, so it was possible to interrupt the operation of the machine. One of the three standard teletypes with a modified keyboard was used as an input / output device. The CNC was located in a separate room, and the operator communicated with the machine remotely using multi-core cables. Through them, teletypes installed in a special room were connected to the device. The performance of the Model I was approximately one multiplication per minute.
As already mentioned, Stibitz was the first to successfully demonstrate remote access to a computing device. In the fall of 1940, a meeting of the American Mathematical Society was held in Hanover (New Hampshire) at which the Model I was presented. Shtibitz spoke to a group about the CNC, demonstrating its work. With the help of a teletype and a telephone cable, we managed to connect three terminals to Dartmouth College, which were located in New York. Data and computation results were transmitted via cable. Williams stayed with the car to keep track of her work.
Operator at the Model I control panel
The presentation impressed those present at the meeting, such as: John von Neumann, Norbert Wiener, Richard Courant. Participants were able to independently test the machine and work at the teletype console. Model I was the beginning of the telecommunications era, when coded machine data was transmitted over telephone channels.
Model I worked from 1940 to 1949. It was actively used for the internal needs of Bell Labs. It took about $ 20,000 to manufacture a computing device.
After the successful launch of the first model, Stibitz moved to the National Defense Research Committee (NDRC). He was going to take on the creation of a more versatile machine. Since the beginning of World War II, Bell Labs began developing the M-9 artillery gun control device. It was a rather complex electromechanical device that skillfully guided an artillery gun at a target moving in the air. Before the serial release of weapons, they tested it, checked the accuracy of shooting, etc. This whole process was accompanied by constant computations and in order to reduce their volumes, simplify the calculations, Stibits proposed the creation of a specialized computing machine - the Relay Interpolator or Model II. In addition, Model I was a specialized computing machine and did not have a device for automatic control of calculations. Such a device appeared in the Model II, controlled by a program “applied” to a punched tape. In 1943, Model II was put into action. EJ Andrews became technical manager.
Model II was a software-controlled machine with a standard five-channel tape that was used as a program carrier. It contained about 440 relays, performed only addition and subtraction operations. There were several software tapes in the device, thanks to which various interpolation methods could be applied.
Stibits with his first invention
The interpolator worked around the clock, was very reliable due to the binary-bi-quinary decimal digit encoding system. Each decimal place was represented by two digits. One of them was the digit of the five-fold system and took values ​​from 0 to 4. The other was the digit of the binary system. As a result, seven relays were required to represent any decimal digit, although only two were switched on at a time. This coding system made it possible to carry out a simple hardware control of the correctness of the interpolator at each step of the calculations. In subsequent years, it was used in all Bell Labs relay machines and in a number of computers from other companies.
Relay computers were less popular than electrical and electromechanical analog devices, which outperformed the former in speed. This point was taken into account in the Model III machines (also known as the “Ballistic computer”) and Model IV. They were also relay, but with an increased number of relays (up to 1,400). In addition, the machines became more efficient and reliable, including ten memory registers. Up to seven teletypes could be connected to them. Both machines performed the work of one hundred calculators with desktop calculators. The devices were able to read tables of several variables from punched tapes and interpolate. Model III still solved ballistic equations describing the path of an aerial target.
Machines Model III and Model IV operated for almost 15 years.
In 1946, the Model V universal computer was developed, which has six processors with 9000 relays each. This was the most significant development of the company Bell Labs.
Relay Model V machine
The Model V was an extremely reliable and precise machine. The storage device consisted of thirty 8-bit registers. Data input and output was made through punched tapes, numbers were represented in the form of a floating point. One could even extract the square root and calculate functions like sin (x), log (x), 10x. For this, there were special blocks in the car. Execution time of arithmetic operations: division - 2.7 seconds; square root - 4.5 seconds; calculation of the logarithm - 15 seconds. In the car there were two identical arithmetic devices (AU), each of which was associated with 15 memory registers. Thanks to this, it was possible to solve two problems simultaneously. Or combine both AUs to perform more complex calculations. While working, a new program could be loaded into the machine, the execution of which was performed by the free AU. In addition, it was possible to simultaneously use multiple software punched tapes. Depending on the results of intermediate calculations, the control device connected one of them. Thus, a semblance of the program branching was created.
The car weighed about 10 tons and cost customers $ 500,000.
Model V worked until 1956, after which it passed into the possession of the Polytechnic Institute of the city of Brooklyn.
George Robert Stibitz (April 30, 1904 - January 31, 1995)
')
Brief autobiographical reference
George Stibits was born in the American city of York (Pennsylvania). He earned a bachelor’s degree from the University of Denison (Granville, Ohio); in 1927, he received a master's degree from Union College (Schenectady, New York); in 1930 - received the degree of doctor of philosophy of physical and mathematical sciences at Cornell University (Ithaca, New York). After that, Stiebitz got a job at Bell Labs, from which his inventor began.
The idea of ​​using electromagnetic relays to create a counting machine of the new model emerged from Stibitsa spontaneously when he was at home. Without losing time, the researcher began assembling a new calculator right in his kitchen. At the same time, Schthitz used all the handy materials: boards, tin cans, a tobacco box, a lantern bulb, a pair of relays and wires. From all this set, he managed to collect a primitive circuit, which could even add two binary numbers and demonstrate the result of addition. George called the homemade car Model K, where k came from the "kitchen" - the kitchen (in honor of the place of "birth" of the machine).
Model K workflow
Shtibitz was sure that on the basis of the relay it is possible to create a device capable of performing sequential calculations and storing their intermediate and final results. In particular, such a machine could multiply and divide complex numbers, since these operations took a lot of time from the staff of his department who were involved in the development of amplifiers and filters.
Model K implementation
The bosses of Bell Labs approved the project and began the development of a computing device. During the period from 1939 to 1940, Shtibitz, together with colleague engineer Samuel B. Williams (Samuel B. Williams), created a device that skillfully added complex numbers and performs subtraction, multiplication, division operations. Shtibitz was the architect of the car, and Williams was the chief engineer. The invention was called the CNC (The Complex Number Calculator) complex number calculator, also known as the Model I. The device was demonstrated at Durmouth College (although the calculator itself was in New York). During the presentation, CNC was the first to use remote access to computing resources. The communication was carried out using teletype on special telephone lines.
From the memories of Stibitsa:
When the work was finished, Sam and I washed our hands and returned to our daily activities, tearing ourselves away from them from time to time to look at our child and make sure that it “eats” and “sleeps” well.
In Model I, there were only 450 bipolar and ten multipolar relays, which served to store the input data and intermediate results. Arithmetic was used with a comma fixed before the first significant digit of the number. The coding (“Stiitz-code”) of the decimal digit was performed using four relays so that each digit n was represented by the binary code n + 3. It simplified the execution of transfer and subtraction operations.
Table with Stibits code values ​​for decimal digits (their direct code is also indicated for comparison):
| Base figure | Direct code | Shtibits code |
| 0 | 0000 | 0011 |
| one | 0001 | 0100 |
| 2 | 0010 | 0101 |
| 3 | 0011 | 0110 |
| four | 0100 | 0111 |
| five | 0101 | 1000 |
| 6 | 0110 | 1001 |
| 7 | 0111 | 1010 |
| eight | 1000 | 1011 |
| 9 | 1001 | 1100 |
CNC used hardware control if more than one relay was triggered in the top five or in the high order. The control circuit gave an error signal.
Model I
Model I worked with 10-bit numbers, but only eight digits were printed (the rest served to round the result). The device was a non-programmable machine with a clearly defined sequence of actions. Each subsequent operation began after the completion of the previous one, so it was possible to interrupt the operation of the machine. One of the three standard teletypes with a modified keyboard was used as an input / output device. The CNC was located in a separate room, and the operator communicated with the machine remotely using multi-core cables. Through them, teletypes installed in a special room were connected to the device. The performance of the Model I was approximately one multiplication per minute.
As already mentioned, Stibitz was the first to successfully demonstrate remote access to a computing device. In the fall of 1940, a meeting of the American Mathematical Society was held in Hanover (New Hampshire) at which the Model I was presented. Shtibitz spoke to a group about the CNC, demonstrating its work. With the help of a teletype and a telephone cable, we managed to connect three terminals to Dartmouth College, which were located in New York. Data and computation results were transmitted via cable. Williams stayed with the car to keep track of her work.
Operator at the Model I control panel
The presentation impressed those present at the meeting, such as: John von Neumann, Norbert Wiener, Richard Courant. Participants were able to independently test the machine and work at the teletype console. Model I was the beginning of the telecommunications era, when coded machine data was transmitted over telephone channels.
Model I worked from 1940 to 1949. It was actively used for the internal needs of Bell Labs. It took about $ 20,000 to manufacture a computing device.
After the successful launch of the first model, Stibitz moved to the National Defense Research Committee (NDRC). He was going to take on the creation of a more versatile machine. Since the beginning of World War II, Bell Labs began developing the M-9 artillery gun control device. It was a rather complex electromechanical device that skillfully guided an artillery gun at a target moving in the air. Before the serial release of weapons, they tested it, checked the accuracy of shooting, etc. This whole process was accompanied by constant computations and in order to reduce their volumes, simplify the calculations, Stibits proposed the creation of a specialized computing machine - the Relay Interpolator or Model II. In addition, Model I was a specialized computing machine and did not have a device for automatic control of calculations. Such a device appeared in the Model II, controlled by a program “applied” to a punched tape. In 1943, Model II was put into action. EJ Andrews became technical manager.
Model II was a software-controlled machine with a standard five-channel tape that was used as a program carrier. It contained about 440 relays, performed only addition and subtraction operations. There were several software tapes in the device, thanks to which various interpolation methods could be applied.
Stibits with his first invention
The interpolator worked around the clock, was very reliable due to the binary-bi-quinary decimal digit encoding system. Each decimal place was represented by two digits. One of them was the digit of the five-fold system and took values ​​from 0 to 4. The other was the digit of the binary system. As a result, seven relays were required to represent any decimal digit, although only two were switched on at a time. This coding system made it possible to carry out a simple hardware control of the correctness of the interpolator at each step of the calculations. In subsequent years, it was used in all Bell Labs relay machines and in a number of computers from other companies.
Relay computers were less popular than electrical and electromechanical analog devices, which outperformed the former in speed. This point was taken into account in the Model III machines (also known as the “Ballistic computer”) and Model IV. They were also relay, but with an increased number of relays (up to 1,400). In addition, the machines became more efficient and reliable, including ten memory registers. Up to seven teletypes could be connected to them. Both machines performed the work of one hundred calculators with desktop calculators. The devices were able to read tables of several variables from punched tapes and interpolate. Model III still solved ballistic equations describing the path of an aerial target.
Machines Model III and Model IV operated for almost 15 years.
In 1946, the Model V universal computer was developed, which has six processors with 9000 relays each. This was the most significant development of the company Bell Labs.
Relay Model V machine
The Model V was an extremely reliable and precise machine. The storage device consisted of thirty 8-bit registers. Data input and output was made through punched tapes, numbers were represented in the form of a floating point. One could even extract the square root and calculate functions like sin (x), log (x), 10x. For this, there were special blocks in the car. Execution time of arithmetic operations: division - 2.7 seconds; square root - 4.5 seconds; calculation of the logarithm - 15 seconds. In the car there were two identical arithmetic devices (AU), each of which was associated with 15 memory registers. Thanks to this, it was possible to solve two problems simultaneously. Or combine both AUs to perform more complex calculations. While working, a new program could be loaded into the machine, the execution of which was performed by the free AU. In addition, it was possible to simultaneously use multiple software punched tapes. Depending on the results of intermediate calculations, the control device connected one of them. Thus, a semblance of the program branching was created.
The car weighed about 10 tons and cost customers $ 500,000.
Model V worked until 1956, after which it passed into the possession of the Polytechnic Institute of the city of Brooklyn.
Source: https://habr.com/ru/post/392103/
All Articles