Other articles in the series:
- History of the relay
- History of electronic computers
- History of the transistor
- Internet history
So far, we have successively reminisced about each of the first three attempts to build a digital electronic computer: the Atanasoff-Berry ABC computer, conceived by John Atanasoff; the British Colossus project, led by Tommy Flowers, and ENIAC, created at the Moore School at the University of Pennsylvania. All these projects were, in fact, independent. Although John Mauchly, the main driving force behind the ENIAC project, was aware of Atanasov's work, the ENIAC scheme did not resemble ABC in any way. If there was any common ancestor of the electronic computing device, it was the humble Wynn-Williams counter, the first device to use vacuum tubes for digital storage and set Atanasoff, Flowers, and Mauchly on the path to electronic computers.
But only one of these three machines, however, played a role in the events that followed. ABC has never produced useful work, and by and large, the few people who knew about it have forgotten it. Two war machines proved capable of outperforming any other computer in existence in sheer speed, but the Colossus remained secret even after the defeat of Germany and Japan. Only ENIAC became widely known, and therefore became the holder of the electronic computing standard. And now anyone who wanted to create a computing device based on vacuum tubes could point to the success of the Moore school for confirmation. The deep-rooted skepticism of the engineering community, which met all such projects before 1945, has disappeared; the skeptics either changed their minds or fell silent.
EDVAC report
Released in 1945, a document based on the experience of creating and using ENIAC set the tone for the direction of computer technology in the world after World War II. It was called "The First Draft Report on EDVAC" [Electronic Discrete Variable Automatic Computer], and provided a blueprint for the architecture of the first computers programmable in the modern sense - that is, executing instructions retrieved from high-speed memory. And although the exact origin of the ideas listed in it remains a matter of debate, it was signed with the name of a mathematician (born Janos Lajos Neumann). In characteristic of the mind of a mathematician, the paper also made the first attempt to abstract the way a computer works from the specifications of a particular machine; he tried to separate the very essence of the structure of the computer from its various possible and random incarnations.
Von Neumann, who was born in Hungary, entered ENIAC via Princeton, New Jersey and Los Alamos, New Mexico. In 1929, as an accomplished young mathematician, with notable contributions to set theory, quantum mechanics, and game theory, he left Europe to take up a position at Princeton University. Four years later, the nearby Institute for Advanced Study (IAS) offered him a lifetime position on the staff. Because of the rise of Nazism in Europe, von Neumann happily jumped at the chance to remain indefinitely on the other side of the Atlantic - and became, after the fact, one of the first Jewish intellectual refugees from Hitler's Europe. After the war, he lamented: "My feelings for Europe are the opposite of nostalgia, for every corner I know reminds me of a vanished world and ruins that bring no consolation," and recalled "his utter disillusionment with the humanity of people in the period from 1933 to 1938."
Disgusted by the lost multinational Europe of his youth, von Neumann directed all his intellect to help the military machine that belonged to the country that sheltered him. Over the next five years, he traveled the country advising and consulting on a wide range of new weapons projects, while somehow managing to co-author a prolific book on game theory. His most secret and important job as a consultant was a position in the Manhattan Project - an attempt to create an atomic bomb - the research team of which was located in Los Alamos (New Mexico). Robert Oppenheimer recruited him in the summer of 1943 to help with the mathematical modeling of the project, and his calculations convinced the rest of the group to move in the direction of the bomb with an inward blast. Such an explosion, thanks to explosives moving the fissile material inward, should have allowed a self-sustaining chain reaction to be achieved. As a result, a huge amount of calculations had to be done in order to achieve a perfect spherical explosion directed inward with the right pressure - and any mistake would lead to an interruption of the chain reaction and the fiasco of the bomb.
Von Neumann while working at Los Alamos
At Los Alamos, there was a group of twenty human calculators who had desktop calculators at their disposal, but they could not cope with the computational load. The scientists gave them equipment from IBM to work with punched cards, but they still couldn't keep up. They demanded improved hardware from IBM, got it in 1944, but still couldn't keep up.
By then, von Neumann had added another set of places to visit to his regular cruise around the country: he traveled to every possible location of computer equipment that might be useful in Los Alamos. He wrote a letter to Warren Weaver, head of the Applied Mathematics Division of the National Defense Research Committee (NDRC), and got some good leads. He went to Harvard to look at the Mark I, but it was already fully loaded with work for the Navy. He spoke with George Stibitz and considered ordering Bell's relay computer for Los Alamos, but abandoned the idea after learning how long it would take. He visited a group at Columbia University that combined several IBM computers into a larger automated system run by Wallace Eckert, but there was no noticeable improvement over the IBM computers already at Los Alamos.
However, Weaver did not include one project on the list he gave von Neumann: ENIAC. He certainly knew about it: in his position as Director of Applied Mathematics, he was required to track the progress of all computing projects in the country. Weaver and the NDRC certainly may have had doubts about the viability and timing of ENIAC, but it is quite surprising that he did not even mention its existence.
Whatever the reason, the result was that von Neumann only learned about ENIAC through a chance meeting on a railway platform. This story was told by Herman Goldstein, a liaison at the Moore School test lab where ENIAC was built. Goldstein encountered von Neumann at Aberdeen railway station in June 1944 - von Neumann was leaving for one of his consultations, which he was giving as a member of the scientific advisory committee at the Aberdeen Ballistic Research Laboratory. Goldstein knew von Neumann's reputation as a great man and struck up a conversation with him. Wanting to make an impression, he could not help but mention a new and interesting project developing in Philadelphia. Von Neumann's approach instantly changed from that of a complacent colleague to that of a tough controller, and he peppered Goldstein with questions related to the details of the new computer. He found an interesting new source of potential computer power for Los Alamos.
Von Neumann first visited Presper Eckert, John Mauchly and other members of the ENIAC team in September 1944. He immediately fell in love with the project and added another item to his long list of consulting organizations. Both sides benefited from this. It is easy to see how the potential of high-speed electronic computing attracted von Neumann. The ENIAC, or a machine similar to it, was able to overcome all the computational limitations that hampered the progress of the Manhattan Project and many other existing or potential projects (however, Say's law, which is still in effect to this day, ensured that the advent of computing power would soon cause an equal demand for them) . For the Moore school, the blessing of such a recognized specialist as von Neumann meant the end of skepticism towards them. Moreover, given his lively mind and wealth of experience throughout the country, he was unrivaled in the breadth and depth of knowledge in the field of automatic calculations.
This is how von Neumann got involved in Eckert and Mauchly's plan to create a successor to ENIAC. Together with Hermann Goldstein and another ENIAC mathematician, Arthur Burks, they began sketching the parameters for the second generation of the electronic computer, and it was from this group that von Neumann summarized the ideas in a "first draft" report. The new machine had to be more powerful, get smoother lines, and, most importantly, overcome the biggest barrier to using ENIAC - many hours of tuning for each new task, during which this powerful and extremely expensive computer simply sat idle. Designers of the latest electromechanical machines, the Harvard Mark I and Bell's relay computer, avoided this by entering instructions into the computer using a paper tape with holes punched in it - the operator could prepare the paper while the machine was solving other tasks. However, such input would negate the speed advantage of the electronics; no paper could feed data as fast as ENIAC could receive it. ("Colossus" worked with paper using photoelectric sensors and each of its five computing modules absorbed data at a rate of 5000 characters per second, but this was only possible due to the fastest possible scrolling of the paper tape. Moving to an arbitrary place on the tape required a delay of 0,5, 5000 s for every XNUMX lines).
The solution to the problem described in the "first draft" was to move the storage of instructions from the "external recording medium" to "memory" - this word was used for the first time in relation to computer data storage (von Neumann specifically used this and other biological terms in the work - he was very interested in the work of the brain and the processes occurring in neurons). This idea was later called "program storage". However, this immediately led to another problem - which puzzled Atanasov - the excessive high cost of vacuum tubes. The "first draft" estimated that a computer capable of performing a wide range of computational tasks would need a memory of 250 binary numbers to store instructions and temporary data. Vacuum tube memory of this size would cost millions of dollars and be completely unreliable.
A solution to the dilemma was proposed by Eckert, who worked on radar research in the early 1940s under a contract between the Moore School and the Rad Lab of MIT, the central research center for radar technology in the United States. Specifically, Eckert was working on a radar system called the Moving Target Indicator (MTI), which solved the problem of “ground flare”: any noise on the radar screen created by buildings, hills and other stationary objects that made it difficult for the operator to isolate important information – size, location and speed of moving aircraft.
At MTI, the backlight problem was solved with a device called . It converted the electrical impulses of the radar into sound waves, and then sent these waves through a mercury tube so that the sound would arrive at the other end and be converted back into an electrical impulse the moment the radar rescanned the same point in the sky (delay lines for propagation sound can also be used by other media: a different liquid, solid crystals, and even air (according to some sources, their idea was invented by Bell Labs physicist William Shockley, about which later). Any signal that came from the radar at the same time as the signal from the tube was considered a signal from a stationary object, and was removed.
Eckert realized that the sound pulses in the delay line could be considered binary numbers - 1 indicates the presence of sound, 0 - its absence. One mercury tube can contain hundreds of such digits, each of which passes through the line several times per millisecond, that is, a computer would have to wait a couple of hundred microseconds to access the digit. In this case, access to consecutive digits in the handset would be faster, since the digits were separated by only a few microseconds.
Mercury delay lines in the British EDSAC computer
After resolving major problems with the computer's design, von Neumann compiled the entire group's ideas into a 101-page "first draft" report in the spring of 1945 and distributed it to key figures in the second-generation EDVAC project. Pretty soon, he infiltrated other circles as well. The mathematician Leslie Comrie, for example, took a copy home to Britain after visiting Moore's school in 1946 and shared it with colleagues. The report's circulation angered Eckert and Mauchly for two reasons: first, it gave much of the credit to the draft's author, von Neumann. Secondly, all the main ideas contained in the system were, in fact, published from the point of view of the patent office, which interfered with their plans to commercialize the electronic computer.
The very basis of Eckert and Mauchly's resentment caused, in turn, the indignation of mathematicians: von Neumann, Goldstein and Burks. In their view, the report was important new knowledge that needed to be disseminated as widely as possible in the spirit of scientific progress. In addition, this entire enterprise was financed by the government, and therefore at the expense of American taxpayers. They were repelled by the commercialism of Eckert and Mauchly's attempt to cash in on the war. Von Neumann wrote: “I would never have accepted a university consulting position knowing that I was advising a commercial group.”
The factions parted ways in 1946: Eckert and Mauchly opened their own company based on a seemingly safer patent based on ENIAC technology. At first they named their business Electronic Control Company, but the following year they renamed it Eckert-Mauchly Computer Corporation. Von Neumann returned to IAS to build the EDVAC computer, and was joined by Goldstein and Burks. To prevent a repeat of the situation with Eckert and Mauchly, they made sure that all the intellectual property of the new project became public domain.
Von Neumann in front of an IAS computer built in 1951.
Digression dedicated to Alan Turing
Among the people who saw the EDVAC report in a roundabout way was the British mathematician Alan Turing. Turing was not among the first scientists to create or invent an automatic computer, electronic or otherwise, and some authors have greatly exaggerated his role in the history of computing. However, we must give him credit as the first person who guessed that computers can do more than just "calculate" something by tritely processing large sequences of numbers. His main idea was that the information processed by the human mind can be represented in the form of numbers, so any mental process can be turned into a calculation.
Alan Turing in 1951
At the end of 1945, Turing published his own report, where he mentioned von Neumann, called "electronic calculator proposal", and intended for the British National Physical Laboratory (NPL). He did not go into the specific details of the design of the proposed electronic computer so much. His scheme reflected the mind of a logician. It was not intended to have special hardware for high-level functions, since they can be composed from low-level primitives; it would be an ugly growth on the beautiful symmetry of the machine. Also, Turing did not allocate any linear memory for a computer program - data and instructions could co-exist in memory, since they were just numbers. An instruction only became an instruction when it was so interpreted (Turing's 1936 work "On Computable Numbers" had already explored the relationship between static data and dynamic instructions. He described what later became known as a "Turing machine" and showed how it could be turned into a number and be fed as input to a universal Turing machine capable of interpreting and executing any other Turing machine). Since Turing knew that numbers could represent any form of neatly defined information, he included in the list of tasks for solving on this computer not only the construction of artillery tables and the solution of systems of linear equations, but also the solution of puzzles and chess studies.
The Turing Automatic Computing Machine (ACE) was never built in its original form. It was too slow and had to compete with more zealous British computing projects for the best talent. The project stalled for several years, and then Turing lost interest in him. In the 1950s, NPL made the Pilot ACE, a smaller and slightly different machine, and several other computer designs were inspired by the ACE architecture in the early 1950s. But she failed to expand her influence, and she quickly went into oblivion.
But all this does not diminish the merits of Turing, it simply helps to place him in the right context. The importance of his influence on the history of computers is based not on the designs of computers in the 1950s, but on the theoretical basis he prepared for computer science that emerged in the 1960s. His early work in mathematical logic, which explored the boundaries of the computable and the non-computable, became the fundamental texts of the new discipline.
Slow Revolution
With the spread of news about ENIAC and the EDVAC report, Moore's school has become a place of pilgrimage. Many visitors came to learn "at the feet of the masters", especially from the US and Britain. In order to streamline the flow of applicants, in 1946 the dean of the school had to organize a summer school on automatic computers, working by invitation. Lectures were given by luminaries such as Eckert, Mauchly, von Neumann, Burks, Goldstein, and Howard Aiken (designer of the Harvard Mark I electromechanical computer).
Almost everyone now wanted to build machines from the instructions in the EDVAC report (ironically, the first machine to run a stored program was ENIAC itself, which was redesigned to use stored instructions in 1948. Only after that did it begin to work successfully in its new house, Aberdeen Proving Ground). Even in the design names of new computers created in the 1940s and 50s, the influence of ENIAC and EDVAC was traced. Even if you do not take into account UNIVAC and BINAC (created in the new company of Eckert and Mouchli) and EDVAC itself (finished at the Moore school after its founders left), there are still AVIDAC, CSIRAC, EDSAC, FLAC, ILLIAC, JOHNNIAC, ORDVAC , SEAC, SILLIAC, SWAC and WEIZAC. Many of them directly copied the freely published IAS construct (with minor modifications), taking advantage of von Neumann's policy of openness regarding intellectual property.
However, the electronic revolution developed gradually, step by step changing the existing order. The first EDVAC style machine did not appear until 1948, and it was only a small proof-of-concept project, the Manchester "baby", designed to prove the viability of memory on (most computers switched from mercury tubes to another type of memory, which also owes its origin to radar technology. Only instead of tubes it used a CRT screen. British engineer Frederick Williams was the first to figure out how to solve the problem with the stability of this memory, as a result of which drives are named after him). In 1949, four more machines were created: the full-size Manchester Mark I, EDSAC at the University of Cambridge, CSIRAC in Sydney (Australia) and the American BINAC - although the latter never worked. Small but stable continued for the next five years.
Some writers have described ENIAC as if it had covered the past with a veil and instantly brought us into the era of electronic computing. Because of this, the real evidence was greatly distorted. “The advent of the all-electronic ENIAC made the Mark I obsolete almost immediately (although it ran successfully for another fifteen years),” wrote Katherine Davis Fishman [The Computer Establishment (1982)]. This statement is so obviously self-contradictory that one might think that Miss Fishman's left hand did not know what her right was doing. You can, of course, write it off as the notes of a simple journalist. However, we find a pair of real historians picking the Mark I again as the whipping boy and write: “Not only was the Harvard Mark I a technical dead end, it didn’t do anything very useful at all in its fifteen years of operation. It was used on several Navy projects, and there the machine proved useful enough for the Navy to order more computers for Aiken's lab" [Aspray and Campbell-Kelly]. Again, a clear contradiction.
In fact, relay computers had their advantages, and they continued to work simultaneously with their electronic cousins. Several new electromechanical computers were created after World War II, and even in the early 1950s in Japan. Relay machines were easier to design, build, and maintain, and didn't require as much electricity and air conditioning (to dissipate the enormous amount of heat emitted by thousands of vacuum tubes). ENIAC used 150 kW of electricity, 20 of which went to its cooling.
The US military continued to be the main consumer of computing power and did not neglect the "obsolete" electromechanical models. In the late 1940s, the Army had four relay computers and the Navy had five. The Aberdeen Ballistics Research Laboratory had the largest concentration of computing power in the world, running ENIAC, relay calculators from Bell and IBM, and an old differential analyzer. In the September 1949 report, everyone was given their place: ENIAC worked best with long, simple calculations; Bell's Model V calculator was better at handling complex calculations due to the almost unlimited length of instruction tape and the ability to work with floating point, and IBM could handle very large amounts of information stored in punched cards. In the meantime, certain operations, such as extracting cube roots, were still easier to perform manually (combining the use of spreadsheets and desktop calculators) and save machine time.
The best marker for the completion of the electronic computing revolution will not be 1945, when ENIAC was born, but 1954, when the IBM 650 and 704 computers appeared. These were not the first commercial electronic computers, but they were the first, produced by the hundreds, and determined IBM's dominant position in computer industry that lasted thirty years. In terminology , electronic computers were no longer a strange anomaly of the 1940s, existing only in the dreams of outcasts like Atanasoff and Mauchly; they have become normal science.
One of the many IBM 650 computers, in this case a copy of Texas A&M University. Magnetic drum memory (below) made it relatively slow, but also relatively inexpensive.
Leaving the nest
By the mid-1950s, the circuitry and design of digital computing equipment had become untied from its origins in analog switches and amplifiers. The computer designs of the 1930s and early '40s relied heavily on ideas from physics and radar laboratories, and especially ideas from telecommunications engineers and research departments. Now computers had organized their own field, and experts in the field were developing their own ideas, vocabulary, and tools to solve their own problems.
The computer in its modern sense appeared, and therefore our coming to an end. However, the world of telecommunications had another interesting trump card up its sleeve. The vacuum tube has surpassed the relay due to the lack of moving parts. And the last relay in our history had an advantage in the complete absence of any internal parts. The innocuous-looking lump of matter with a few wires sticking out of it comes from a new branch of electronics known as "solid-state."
Although vacuum tubes were fast, they were still expensive, large, hot, and not particularly reliable. On them it was impossible to make, say, a laptop. Von Neumann wrote in 1948 that "it is unlikely that we will be able to exceed the number of switches of 10 (or perhaps several tens of thousands) so long as we are forced to apply current technology and philosophy)." The solid state relay gave computers the ability to push these limits again and again, breaking them repeatedly; come into use in small businesses, schools, homes, household appliances and fit into pockets; to create a magical digital land that permeates our existence today. And to find its origins, we need to rewind the clock fifty years ago, and go back to the interesting early days of wireless technology.
What else to read:
- David Anderson, “Was the Manchester Baby conceived at Bletchley Park?”, British Computer Society (June 4th, 2004)
- William Aspray, John von Neumann and the Origins of Modern Computing (1990)
- Martin Campbell-Kelly and William Aspray, Computer: A History of the Information Machine (1996)
- Thomas High, et. al., Eniac in Action (2016)
- John von Neumann, “First Draft of a Report on EDVAC” (1945)
- Alan Turing, "Proposed Electronic Calculator" (1945)
Source: habr.com