Soviet computer technology. The story of take-off and oblivion

2024 Author: Seth Attwood | [email protected]. Last modified: 2023-12-16 15:55

Complete and comprehensive information about the development of Soviet electronics. Why did Soviet electronics at one time significantly surpass foreign "hardware"? Which Russian scientist embodied Soviet know-how in Intel's microprocessors?

How many critical arrows have been fired in recent years on the state of our computing technology! And that it was hopelessly backward (at the same time it was sure to mention the "organic vices of socialism and the planned economy"), and that it is pointless to develop it now, because "we are forever behind." And in almost every case, the reasoning will be accompanied by the conclusion that "Western technology has always been better", that "Russian computers do not know how to do it" …

Usually, criticizing Soviet computers, attention is focused on their unreliability, difficulties in operation, and low capabilities. Yes, many "experienced" programmers probably remember those "ES-ki" "hanging" endlessly from the 70s and 80s, they can talk about how the "Sparks", "Agatha", "Robotrons" looked like, " Electronics "against the background of the IBM PCs that had just begun to appear in the Union (not even the latest models) in the late 80s - early 90s, mentioning that such a comparison does not end in favor of domestic computers. And this is so - these models were really inferior to their Western counterparts in their characteristics.

But these listed brands of computers were by no means the best domestic developments, despite the fact that they were the most widespread. And in fact, Soviet electronics not only developed at the world level, but sometimes outstripped a similar Western industry!

But why, then, now we use exclusively foreign "hardware", and in Soviet times, even the hard-won domestic computer seemed like a heap of metal compared to its Western counterpart? Isn't the statement about the superiority of Soviet electronics unfounded?

No, it's not! Why? The answer is in this article.

The glory of our fathers

The official "date of birth" of Soviet computer technology should probably be considered the end of 1948. It was then that in a secret laboratory in the town of Feofania near Kiev, under the leadership of Sergei Aleksandrovich Lebedev (at that time - director of the Institute of Electrical Engineering of the Academy of Sciences of Ukraine and also head of the laboratory of the Institute of Precise Mechanics and Computing Technology of the Academy of Sciences of the USSR), work began on the creation of a Small Electronic Counting Machine (MESM) …

Lebedev put forward, substantiated and implemented (independently of John von Neumann) the principles of a computer with a program stored in memory.

In his first machine, Lebedev implemented the fundamental principles of building computers, such as:

availability of arithmetic devices, memory, input / output and control devices;

coding and storing a program in memory like numbers;

binary number system for encoding numbers and commands;

automatic execution of calculations based on the stored program;

the presence of both arithmetic and logical operations;

the hierarchical principle of building memory;

using numerical methods to implement computations.

Design, installation and debugging of MESM were carried out in record time (about 2 years) and carried out by only 17 people (12 researchers and 5 technicians). The test run of the MESM machine took place on November 6, 1950, and regular operation on December 25, 1951.

In 1953, a team headed by S. A. Lebedev created the first mainframe - BESM-1 (from the Big Electronic Counting Machine), released in one copy. It was created already in Moscow, at the Institute of Precision Mechanics (abbreviated as ITM) and the Computing Center of the Academy of Sciences of the USSR, the director of which was S. A. Lebedev, and was assembled at the Moscow Plant of Calculating and Analytical Machines (abbreviated as CAM).

After the BESM-1 RAM was equipped with an improved element base, its performance reached 10,000 operations per second - at the level of the best in the USA and the best in Europe. In 1958, after another modernization of the RAM, the BESM, which had already received the name BESM-2, was prepared for serial production at one of the factories of the Union, which was carried out in the amount of several dozen.

In parallel, work was going on in the Moscow Region Special Design Bureau No. 245, which was headed by M. A. Lesechko, also founded in December 1948 by order of I. V. Stalin. In 1950-1953 the team of this design bureau, but already under the leadership of Bazilevsky Yu. Ya. developed a general-purpose digital computer "Strela" with a speed of 2 thousand operations per second. This car was produced until 1956, and a total of 7 copies were made. Thus, "Strela" was the first industrial computer - MESM, BESM existed at that time in only one copy.

In general, the end of 1948 was an extremely productive time for the creators of the first Soviet computers. Despite the fact that both of the computers mentioned above were among the best in the world, again, in parallel with them, another branch of the Soviet computer industry developed - M-1, "Automatic digital computing machine", which was headed by IS Brook.

M-1 was launched in December 1951 - simultaneously with MESM and for almost two years was the only operating computer in the USSR (MESM was geographically located in Ukraine, near Kiev).

However, the speed of the M-1 turned out to be extremely low - only 20 operations per second, which, however, did not prevent it from solving the problems of nuclear research at the IV Kurchatov Institute. At the same time, M-1 took up quite a bit of space - only 9 square meters (compare with 100 square meters for BESM-1) and consumed significantly less energy than Lebedev's brainchild. M-1 became the ancestor of a whole class of "small computers", of which its creator IS Brook was a supporter. Such machines, according to Brook, should have been intended for small design bureaus and scientific organizations that do not have the means and premises to purchase machines of the BESM type.

Soon the M-1 was seriously improved, and its performance reached the level of "Strela" - 2 thousand operations per second, at the same time, the size and power consumption increased slightly. The new car received the natural name M-2 and was put into operation in 1953. In terms of cost, size and performance, the M-2 has become the best computer in the Union. It was M-2 that won the first international chess tournament between computers.

As a result, in 1953, serious computing tasks for the needs of the country's defense, science and the national economy could be solved on three types of computers - BESM, Strela and M-2. All these computers are the first generation computers. The element base - electronic tubes - determined their large dimensions, significant energy consumption, low reliability and, as a result, small production volumes and a narrow circle of users, mainly from the world of science. In such machines, there were practically no means of combining the operations of the program being executed and parallelizing the operation of various devices; the commands were executed one after another, the ALU ("arithmetic-logic device", a unit that directly performs data conversions) was idle in the process of data exchange with external devices, the set of which was very limited. The volume of the BESM-2 RAM, for example, was 2048 39-bit words; magnetic drums and magnetic tape drives were used as external memory.

Setun is the first and only ternary computer in the world. Moscow State University. THE USSR.

Manufacturing plant: Kazan Plant of Mathematical Machines of the USSR Ministry of Radio Industry. The manufacturer of logic elements is the Astrakhan plant of electronic equipment and electronic devices of the USSR Ministry of Radio Industry. The manufacturer of magnetic drums is the Penza Computer Plant of the USSR Ministry of Radio Industry. The manufacturer of the printing device is the Moscow Plant of Typewriters of the USSR Ministry of Instrument Industry.

Completion year of development: 1959.

Year of the beginning of production: 1961.

Discontinued production: 1965.

Number of cars built: 50.

In our time, "Setun" has no analogues, but historically it happened that the development of informatics went into the mainstream of binary logic.

But the next development of Lebedev was more productive - the M-20 computer, the serial production of which began in 1959.

The number 20 in the name means high-speed performance - 20 thousand operations per second, the amount of RAM twice exceeded the OP BESM, some combination of the executed commands was also envisaged. At that time it was one of the most powerful and reliable machines in the world, and it was used to solve many of the most important theoretical and applied problems of science and technology of that time. In the M20 machine, the possibility of writing programs in mnemonic codes was implemented. This greatly expanded the circle of specialists who were able to take advantage of the benefits of computing. Ironically, exactly 20 M-20 computers were produced.

The first generation computers were produced in the USSR for a long time. Even in 1964, the Ural-4 computer, which was used for economic calculations, was still being produced in Penza.

Victory gait

In 1948, a semiconductor transistor was invented in the USA, which began to be used as an element base for a computer. This made it possible to develop computers with significantly smaller dimensions, power consumption, and significantly higher (in comparison with lamp computers) reliability and productivity. The problem of programming automation became extremely urgent, since the gap between the time for developing programs and the time for the actual calculation was increasing.

The second stage in the development of computing technology in the late 50s - early 60s is characterized by the creation of advanced programming languages (Algol, Fortran, Cobol) and the development of the process of automating task flow control using the computer itself, that is, the development of operating systems. The first operating systems automated the user's work on completing a task, and then tools were created for entering several tasks at once (a batch of tasks) and distributing computing resources between them. A multiprogramming mode of data processing has appeared. The most characteristic features of these computers, commonly referred to as "second generation computers":

combining input / output operations with calculations in the central processor;

an increase in the amount of RAM and external memory;

use of alphanumeric devices for data input / output;

"closed" mode for users: the programmer was no longer allowed into the computer room, but handed over the program in the algorithmic language (high-level language) to the operator for its further admission on the machine.

At the end of the 50s, the serial production of transistors was also established in the USSR.

This made it possible to start creating a second-generation computer with higher performance, but less space and power consumption. The development of computer technology in the Union went almost at an "explosive" pace: in a short time, the number of different computer models put into development began to count in dozens: this is the M-220 - the heir to the Lebedev M-20, and the "Minsk-2" with subsequent versions, and the Yerevan "Nairi", and many military computers - M-40 with a speed of 40 thousand operations per second and M-50 (which still had tube components). It was thanks to the latter that in 1961 it was possible to create a fully functional anti-missile defense system (during the tests, it was repeatedly possible to shoot down real ballistic missiles with a direct hit into a warhead with a volume of half a cubic meter). But first of all, I would like to mention the BESM series, developed by the team of developers of ITM and VT of the USSR Academy of Sciences under the general leadership of S. A. Lebedev, whose work pinnacle was the BESM-6 computer created in 1967. It was the first Soviet computer to achieve a speed of 1 million operations per second (an indicator surpassed by domestic computers of subsequent releases only in the early 80s, with significantly lower operating reliability than that of BESM-6).

In addition to high speed (the best indicator in Europe and one of the best in the world), the structural organization of BESM-6 was distinguished by a number of features that were revolutionary for their time and anticipated the architectural features of the next generation computers (the element base of which was made up of integrated circuits). So, for the first time in domestic practice and completely independently of foreign computers, the principle of combining the execution of instructions was widely used (up to 14 machine instructions could be simultaneously in the processor at different stages of execution). This principle, named by the chief designer of BESM-6 academician S. A. Lebedev as the "water pipeline" principle, later became widely used to increase the productivity of general-purpose computers, having received the name "command conveyor" in modern terminology.

BESM-6 was mass-produced at the Moscow plant SAM from 1968 to 1987 (a total of 355 vehicles were produced) - a kind of record! The last BESM-6 was dismantled today - in 1995 at the Mil helicopter plant in Moscow. BESM-6 was equipped with the largest academic (for example, the Computing Center of the USSR Academy of Sciences, the Joint Institute for Nuclear Research) and industry (Central Institute of Aviation Engineering - CIAM) research institutes, factories and design bureaus.

In this regard, an article by the curator of the Museum of Computer Science in Great Britain, Doron Sweid, about how he bought one of the last working BESM-6 in Novosibirsk is interesting. The title of the article speaks for itself:

Information for specialists

The operation of the RAM modules, the control unit and the arithmetic logic unit in BESM-6 was carried out in parallel and asynchronously, thanks to the presence of buffer devices for intermediate storage of commands and data. To speed up the pipelined execution of instructions in the control device, a separate register memory for storing indexes, a separate address arithmetic module, providing fast address modification using index registers, including the stack access mode, were provided.

Associative memory on fast registers (of the cache type) made it possible to automatically store the most frequently used operands in it and thereby reduce the number of accesses to the main memory. The "layering" of the random access memory provided the possibility of simultaneous access to its different modules from different devices of the machine. Mechanisms for interrupting, protecting memory, converting virtual addresses into physical and privileged operating modes for the OS made it possible to use BESM-6 in multiprogram and time-sharing modes. In the arithmetic-logic device, accelerated algorithms for multiplication and division were implemented (multiplication by four digits of a multiplier, calculation of four digits of the quotient in one clock cycle), as well as an adder without end-to-end carry chains, representing the result of the operation in the form of a two-row code (bitwise sums and transfers) and operating on the input three-row code (the new operand and the two-row result of the previous operation).

The BESM-6 computer had a random access memory on ferrite cores - 32 KB of 50-bit words, the amount of random access memory increased with subsequent modifications to 128 KB.

Data exchange with external memory on magnetic drums (hereinafter also on magnetic disks) and magnetic tapes was carried out in parallel via seven high-speed channels (a prototype of future selector channels). The work with the rest of the peripheral devices (element-by-element data input / output) was carried out by the operating system driver programs when the corresponding interrupts from the devices occurred.

Technical and operational characteristics:

Average performance - up to 1 million unicast commands / s

The word length is 48 binary bits and two check bits (the parity of the whole word had to be “odd”. Thus, it was possible to distinguish commands from data - some had half-words parity “even-odd”, while others had “odd-even” . The transition to data or erasure of the code was caught elementary, as soon as there was an attempt to execute a word with data)

Number representation - floating point

Working frequency - 10 MHz

Occupied area - 150-200 sq. m

Power consumption from the network 220 V / 50 Hz - 30 kW (without air cooling system)

BESM-6 had an original system of elements with paraphase synchronization. The high clock frequency of the elements demanded from the developers new original design solutions to shorten the lengths of the element connections and reduce parasitic capacitances.

The use of these elements in combination with original structural solutions made it possible to provide a performance level of up to 1 million operations per second when operating in 48-bit floating point mode, which is a record in relation to a relatively small number of semiconductor elements and their speed (about 60 thousand units). transistors and 180 thousand diodes and a frequency of 10 MHz).

The BESM-6 architecture is characterized by an optimal set of arithmetic and logical operations, fast address modification using index registers (including the stack access mode), and a mechanism for expanding the opcode (extracodes).

When creating BESM-6, the basic principles of a computer design automation system (CAD) were laid. The compact recording of the machine diagrams by the formulas of Boolean algebra was the basis of its operational and commissioning documentation. The documentation for the installation was issued to the plant in the form of tables obtained on an instrumental computer.

The creators of BESM-6 were V. A. Melnikov, L. N. Korolev, V. S. Petrov, L. A. Teplitsky - the leaders; A. A. Sokolov, V. N. Laut, M. V. Tyapkin, V. L. Lee, L. A. Zak, V. I. Smirnov, A. S. Fedorov, O. K. Shcherbakov, A. V. Avayev, V. Ya. Alekseev, O. A. Bolshakov, V. F. Zhirov, V. A. Zhukovsky, Yu. I. Mitropolsky, Yu. N. Znamensky, V. S. Chekhlov,. A. Lebedev.

In 1966, an anti-missile defense system was deployed over Moscow on the basis of a 5E92b computer created by the groups of S. A. Lebedev and his colleague V. S. Burtsev with a capacity of 500 thousand operations per second, which has existed until now (in 2002 it should be with the reduction of the Strategic Missile Forces).

A material base was also created for the deployment of missile defense over the entire territory of the Soviet Union, but subsequently, according to the terms of the ABM-1 treaty, work in this direction was curtailed. V. S. Burtsev's group took an active part in the development of the legendary anti-aircraft anti-aircraft system S-300, creating in 1968 for it the 5E26 computer, which was distinguished by its small size (2 cubic meters) and the most careful hardware control that tracked any incorrect information. The performance of the 5E26 computer was equal to that of the BESM-6 - 1 million operations per second.

Betrayal

Probably the most stellar period in the history of Soviet computing was the mid-sixties. There were many creative collectives operating in the USSR at that time. The institutes of S. A. Lebedev, I. S. Bruk, V. M. Glushkov are only the largest of them. Sometimes they competed, sometimes they complemented each other. At the same time, many different types of machines were produced, most often incompatible with each other (perhaps with the exception of machines developed at the same institute), for a wide variety of purposes. All of them were designed and made at the world level and were not inferior to their Western competitors.

The variety of computers produced and their incompatibility with each other at the software and hardware levels did not satisfy their creators. It was necessary to put in the slightest degree order in the entire set of computers produced, for example, taking any of them as a certain standard. But…

In the late 60s, the country's leadership made a decision, which, as the course of further events showed, had disastrous consequences: to replace all different-sized domestic developments of the middle class (there were about half a dozen of them - "Minsky", "Urals", different versions of the architecture of the M-20 etc.) - on the Unified Family of computers based on the architecture of the IBM 360, - the American counterpart. At the level of the Ministry of Instrumentation, a similar decision was not made so loudly with regard to the mini-computer. Then, in the second half of the 70s, the PDP-11 architecture of the foreign firm DEC was also approved as the general line for mini- and micro-computers. As a result, manufacturers of domestic computers were forced to copy outdated samples of IBM computers. It was the beginning of the end.

Here is the assessment of Boris Artashesovich Babayan, Corresponding Member of the Russian Academy of Sciences:

It is by no means worth thinking that the teams of ES EVM developers did their job poorly. On the contrary, creating fully functional computers (albeit not very reliable and powerful) similar to their Western counterparts, they coped with this task brilliantly, given that the production base in the USSR lagged behind the Western one. It was precisely the orientation of the entire industry towards "imitation of the West" and not towards the development of original technologies that was erroneous.

Unfortunately, it is now unknown who exactly in the country's leadership made the criminal decision to curtail the original domestic developments and develop electronics in the direction of copying Western counterparts. There were no objective reasons for such a decision.

One way or another, but from the beginning of the 70s, the development of small and medium-sized computer technology in the USSR began to degrade. Instead of further development of well-developed and tested concepts of computer engineering, the huge forces of the country's computer science institutes began to engage in "stupid" and, moreover, semi-legal copying of Western computers. However, it could not be legal - the "cold war" was on, and the export of modern "computer-building" technologies to the USSR in most Western countries was simply prohibited by law.

Here is one more testimony of B. A. Babayan:

The most important thing is that the way of copying overseas decisions turned out to be much more complicated than previously thought. Compatibility of architectures required compatibility at the element base level, which we did not have. In those days, the domestic electronics industry was also forced to take the path of cloning American components, in order to provide the possibility of creating analogs of Western computers. But it was very difficult.

It was possible to get and copy the topology of microcircuits, find out all the parameters of electronic circuits. However, this did not answer the main question - how to make them. According to one of the experts of the Russian Ministry of Economic Development, who at one time worked as the general director of a large NGO, the advantage of the Americans has always been in huge investments in electronic engineering. In the United States, it was not so much the technological lines for the production of electronic components that were and remain top secret, but the equipment for the creation of these very lines. The result of this situation was that the Soviet microcircuits created in the early 70s - analogs of the Western ones - were similar to the American-Japanese ones in functional terms, but did not reach them in terms of technical parameters. Therefore, boards assembled according to American topologies, but with our components, turned out to be inoperative. I had to develop my own circuit solutions.

Sweid's article cited above concludes:. This is not entirely true: after the BESM-6 there was the Elbrus series: the first of the machines of this series, the Elbrus-B, was a microelectronic copy of the BESM-6, which made it possible to work in the BESM-6 command system and use the software written for it.

However, the general meaning of the conclusion is correct: due to the order of incompetent or deliberately harmful leaders of the ruling elite of the Soviet Union at that time, Soviet computer technology was closed the way to the top of the world Olympus. Which she could well achieve - the scientific, creative and material potential quite allowed to do this.

For example, here are some of the personal impressions of one of the authors of the article:

However, by no means all original domestic developments were curtailed. As already mentioned, VS Burtsev's team continued to work on the Elbrus computer series, and in 1980 the Elbrus-1 computer with a speed of up to 15 million operations per second was put into mass production. Symmetric multiprocessor architecture with shared memory, implementation of secure programming with hardware data types, superscalarity of processor processing, a unified operating system for multiprocessor complexes - all these capabilities implemented in the Elbrus series appeared earlier than in the West. In 1985, the next model of this series, Elbrus-2, was already performing 125 million operations per second. "Elbrus" worked in a number of important systems associated with the processing of radar information, they were counted in license plates Arzamas and Chelyabinsk, and many computers of this model still provide the functioning of anti-missile defense systems and space forces.

A very interesting feature of "Elbrus" was the fact that the system software for them was created in a high-level language - El-76, and not in traditional assembler. Before execution, El-76 code was translated into machine instructions using hardware, not software.

Since 1990, Elbrus 3-1 was also produced, which was distinguished by its modular design and was intended for solving large scientific and economic problems, including modeling physical processes. Its performance reached 500 million operations per second (on some commands). A total of 4 copies of this machine were produced.

Since 1975, a group of I. V. Prangishvili and V. V. Rezanov in the research and production association "Impulse" began to develop a computer complex PS-2000 with a speed of 200 million operations per second, put into production in 1980 and used mainly for processing of geophysical data, - search for new deposits of minerals. In this complex, the possibilities of parallel execution of program commands were maximized, which was achieved by an ingeniously designed architecture.

Large Soviet computers, like the PS-2000, in many ways even surpassed their foreign competitors, but they cost much less - so, only 10 million rubles were spent on the development of PS-2000 (and its use made it possible to get a profit of 200 million rubles). However, their scope was "large-scale" tasks - the same missile defense or space data processing. The development of medium and small computers in the Union was seriously and for a long time slowed down by the betrayal of the Kremlin elite. And that is why the device that is on your table and which is described in our magazine was made in Southeast Asia, and not in Russia.

Catastrophe

Since 1991, hard times have come for Russian science. The new government of Russia has taken a course towards the destruction of Russian science and original technologies. Financing of the overwhelming majority of scientific projects was stopped, due to the destruction of the Union, the interconnection of computer manufacturing plants that ended up in different states was interrupted, and efficient production became impossible. Many developers of domestic computer technology were forced to work outside their specialty, losing their qualifications and time. The only copy of the Elbrus-3 computer developed back in Soviet times, twice as fast as the most productive American supercar of that time, the Cray Y-MP, was disassembled and put under pressure in 1994.

Some of their creators of Soviet computers went abroad. So, at present, the leading developer of microprocessors from Intel is Vladimir Pentkovsky, who was educated in the USSR and worked at ITMiVT - the Lebedev Institute of Precision Mechanics and Computational Engineering. Pentkovsky took part in the development of the above-mentioned computers "Elbrus-1" and "Elbrus-2", and then headed the development of the processor for "Elbrus-3" - El-90. As a result of the targeted policy of destruction of Russian science pursued by the ruling circles of the Russian Federation under the influence of the West, funding for the Elbrus project was cut off, and Vladimir Pentkovsky was forced to emigrate to the United States and get a job at Intel. He soon became a senior engineer of the corporation and under his leadership in 1993 Intel developed the Pentium processor, rumored to be named after Pentkovsky.

Pentkovsky embodied in Intel's processors the Soviet know-how that he knew himself, thinking out a lot during the development process, and by 1995 Intel released a more advanced Pentium Pro processor, which had already come close in its capabilities to the Russian microprocessor of 1990 El- 90, although he did not catch up with him. Pentkovsky is currently developing the next generation of Intel processors. So the processor on which your computer may be running was made by our compatriot and could have been made in Russia if not for the events after 1991.

Many research institutes have switched to the creation of large computing systems based on imported components. Thus, the research institute "Kvant" under the leadership of V. K. Levin is developing computing systems MVS-100 and MVS-1000, based on Alpha 21164 processors (manufactured by DEC-Compaq). However, the acquisition of such equipment is hampered by the current embargo on the export of high technologies to Russia, while the possibility of using such complexes in defense systems is extremely doubtful - no one knows how many "bugs" can be found in them that are activated by a signal and disable the system.

On the personal computer market, domestic computers are completely absent. The maximum that Russian developers go to is assembling computers from components and creating separate devices, for example, motherboards, again from ready-made components, while placing orders for production at factories in Southeast Asia. However, there are very few such developments (one can name the firms "Aquarius", "Formosa"). The development of the ES line has practically stopped - why create your own analogs when it is easier and cheaper to buy originals?

Of course, all is not lost. There are also descriptions of technologies, sometimes even on

over the past ten years, superior Western and current models. Fortunately, not all the developers of domestic computer technology went abroad or died. So there is still a chance.

Whether it will be implemented depends on us.