Imaginarium of Science. Part 2

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

After the introduction of the copying system for American samples and the appearance of a series of EU machines - copies of the American IBM360 / IBM370, the USSR's own developments in the field of computer technology did not stop. However, they almost completely went into the framework of military projects - the military did not want to use only copies, and even worse than their own developments. Import did not suit them because of possible "bookmarks" - undocumented features of electronics that could disable electronics in the interests of a potential enemy. ITM and VT, whose director was Academician Lebedev, although he continued to be listed as an academic institute, became essentially a military department and work continued there towards improving the BESM-6 and military M-40, M-50. The result of such work was the Elbrus line, the main tasks of which were the tasks for the anti-missile defense system. First, on the basis of the military computers 5E261 and 5E262, a multiprocessor computer complex "Elbrus-1" with a productivity of 15 million operations / s was created. At the second stage, the Elbrus-2 MVK was created with a capacity of 120 million operations / s. Elbrus-3, the development of which was completed by the end of the 80s, had a performance of 500 MFLOPS (millions of floating point operations per second).

Performance indicators for a computer are a very relative thing, depending both on architectural features and on the efficiency of compilers from programming languages. Therefore, benchmarks are often used to compare real-world performance. In 1988, S. V. Kalin measured the performance of the CPU of the MVK "Elbrus-2" at 24 "Livermore cycles" and, according to the results of these tests, the average harmonic value of performance was 2.7 MFLOPS. For comparison, the Cray-X MP processor (the most famous development of Seymour Kray in 1982) has a similar indicator - 9.3 MFLOPS (at a clock frequency 5 times higher than that of the Elbrus-2 MVK). This ratio indicates the high efficiency of the Elbrus architecture, which allows more operations to be performed per processor cycle.

The architecture of the Elbrus processors was already significantly different from the old BESM-6 and was very different from the traditional one. The core of "Elbrus 3-1" was a modular conveyor processor (MCP), designed by Andrey Andreevich Sokolov. Sokolov was a participant in all the most significant projects of the Lebedev Institute, from BESM-1 to AS-6. And it was Sokolov's engineering talent that colleagues have often compared with the talent of Seymour Krey - Lebedev's constant rival in the super-speed computing competition. "The MCP was a powerful processor capable of processing two independent streams of instructions. The pipeline devices of the processor worked with two types of objects - vectors and scalars. Scalars seemed to be wedged into a vector pipeline and processed between two adjacent vector components. Several access channels provided up to 8 parallel calls to memory in one cycle. " Almost all the architectural features of Elbrus were absolutely original, but they are often called borrowing principles from CDC and Burroughs, which is an obvious lie. Lebedev began to use both the pipeline and the principles of parallel computing earlier.

The Lebedev Institute is still at its best, having gone through the era of Yeltsinism, albeit with significant losses, but without losing its creative potential. True, in a new incarnation - in April 1992, on the basis of the departments of the Lebedev Institute of Precision Mechanics and Computer Technology, MCST was created, which continued the development of the Elbrus architecture. That year, one of the leading employees of the institute B. A. Babayan and most of the MCST specialists were hired by the giant Intel corporation to work in its Russian branch. It may seem ridiculous, but it was Intel then that allowed us to retain domestic personnel in electronics, borrowing, of course, significant developments of the institute along with a part of the personnel. On the basis of the architecture of the Elbrus MVK, the specialists of the new company in 2007 created the Elbrus microprocessor, which served as the basis for the Elbrus-3M1 computing systems, with a clock frequency of 300 MHz and a performance of 4.8 GFLOPS (for comparison, Intel Core2Duo 2.4 GHz has only 1.3 gigaflops). At the same time, the Russian microprocessor does not even require a radiator for cooling. The two-processor version of the computer complex, called UVK / S, has a peak performance of 19 GFLOPS (for 32-bit data). This is the answer to those who think that our military today has to use personal computers from IBM with microprocessors from Intel. Fortunately, this is not the case. Although for this I had to purchase imported equipment for the production of microcircuits.

System module with two microprocessors "Elbrus" and computing complex "Elbrus-3M1":

The microprocessor is made using the 0.13 micron technology, which is not a technological record for today, but it does not lag far behind them either (the technology was considered a novelty about 5 years ago). Now the development of the Elbrus-S microprocessor is underway on the technology of 0.09 microns, which is already a "system on a chip", that is, it includes peripheral equipment controllers. It is designed to create high-performance single-board computers for "wearable and embedded" applications, which means our aircraft and missiles will not be equipped with imported components.

But let's go back to the 60s. The USSR was then the first in many technical developments in the field of electronics, most of which were carried out within the framework of military projects and therefore were secret. And due to the secrecy, these achievements remained outside the attention of historians. The creator of BESM-6, an outstanding Soviet designer of computer technology, Sergei Alekseevich Lebedev, also designed purely military computers for the first, still experimental, anti-missile defense (ABM) system:

"Specialized computers, created under the leadership of S. A. Lebedev for the anti-missile defense system, became the basis for achieving strategic parity between the USSR and the United States during the Cold War." specialized computers "Diana-1" and "Diana-2" were developed for automatic data retrieval from the radar and automatic tracking of targets. -40, and a little later M-50 (floating point). The possibility of hitting ballistic missiles, provided by missile defense, forced the United States to look for ways to conclude an agreement with the USSR on the limitation of missile defense, which appeared in 1972."

The achievements of the USSR in computer technology were of the greatest importance for defense and served as an important argument for the conclusion of a treaty on the limitation of missile defense … And just when we had a significant advantage in this. The USSR practically already had its own anti-missile defense by the mid-60s, when the United States could only dream of it. The treaty limited primarily the USSR, not the United States - as a result of the treaty, the missile defense system was deployed only around Moscow. When the United States was finally able to do something in this area (this is 30 years later!), It immediately withdrew from the treaty. The question is - was there any point for the USSR to sign such an agreement? We gave up the missile defense shield and got nothing in return! The United States simply could not create its own back then. Did the USSR leadership know about this? If she knew, then the ABM Treaty can already be considered an act of betrayal of the country's interests. The situation is very reminiscent of 1987, when the Soviet Union was ready to put into orbit the components of a space missile defense system - satellites with laser weapons "SKIF". Then Gorbachev, convinced of the possible success of the program, immediately imposed a unilateral moratorium on it, announcing from the UN rostrum that the USSR was abandoning the "arms race in space." The United States plans to launch similar satellites into orbit only in 2012, 25 years after the closure of a similar Soviet program. Not because they suddenly had such a desire. Because their technologies, not without the help of Russian specialists, have only now allowed it. Why did the USSR leadership make unilateral concessions? There is no official version of the answer to this question.

Back in the early 60s, our computers managed to calculate the trajectories of ballistic missiles, despite the fact that initially our missile defense system worked on rather slow computers. The M-40 and M-50 machines had a productivity of only 40 thousand and 50 thousand operations per second, respectively. However, the 5E92b, a military modification of the M-50, had a productivity of 500 thousand operations per second, which for 1966, from which its production began, was close to a world record, if not. And there is another little-known detail here.

Among the many often mentioned Soviet computer models, the names of a very important series of computers that were produced in the second half of the 60s - the beginning of the 70s and were wholly used for the acquisition of the USSR Armed Forces are rare. These are machines of the 5E series (5E51, 5E92b, etc.), developed by the Lebedev Design Bureau. BESM-6 is widely known, but few people know that BESM-6 became famous only because it lost the tender for supplies to the Armed Forces of the USSR - the tender won by "5E". The military, having opted for "5E", sort of "rejected" BESM-6 and the latter went into open distribution for civilian industries. And the 5E series was classified and was shipped only to the military. Machines of the 5E series were united by "intermachine exchange" channels into local networks, which in the first half of the 70s constituted a multiprocessor computing environment as the basis for space control and space objects control systems. Several computers put together in such a computing environment made up a single computing complex, which had several times higher performance than BESM-6. The same principle now serves as the basis for the creation of modern supercomputers - these are individual processors assembled into a single network by fast communication channels. And this requires special means. The machines of the M series (M-40, M-50) also had a developed interrupt system, they could receive and transmit data over seven duplex asynchronously operating channels with a total bandwidth of 1 Mbit / s. Modification M-50 - 5E92 was specially designed for use in such data processing complexes.

For the first time in the world, multiplex channels were used in a computer network and parallel operation of control devices, random access memory, external devices and communication channels was carried out. In terms of structure and principle of operation, it was the world's first multiprocessor system … In 1959, a computer network was built from computers that were hundreds of kilometers apart - there were no similar complexes abroad at that time. The main command and computer center of the "A" system was built on the basis of the 5E92 computer. The computer network itself was unique, it was she who served as the starting point of research, which subsequently led to the creation of other global information and computer networks. Of course, this network itself did not resemble, for example, the modern Internet, but as a set of independent machines solving independent fragments of a common problem and exchanging information using unified protocols, it can be considered the forerunner of today's global networks. The first similar network, connecting two TX-2 computers in Massachusetts and Q-32 in California via a telephone line, was tested only in 1965 … On March 4, 1961, an experimental anti-missile defense system was successfully tested - the warhead of an R-12 missile was destroyed. The experiment showed that the task of combating paired ballistic targets consisting of a ballistic missile body and a nuclear warhead separated from it has been technically solved. Similar tests took place in the United States 21 years later.

System A is a missile defense system. Work on missile defense (system "A") played a huge role in the development of computer technology in the USSR: by order of the military, using a relatively slow element base, specialists from the Lebedev Design Bureau (ITMiVT) created computing facilities that were superior in their parameters to foreign ones. They also created mobile versions of such systems, for example 5E261 - a mobile multiprocessor high-performance control system built on a modular basis. It was she who was used as part of the S-300PT air defense systems for land and sea-based:

But most importantly, means of interfacing individual computers into a computing environment - fast asynchronous multiplex communication channels and corresponding software - were created. And here we come to another very important project for the country, the system OGAS - "National automated system of accounting and information processing", a system of automated economic management in the USSR, based on the principles of cybernetics. This system, developed by Academician Viktor Mikhailovich Glushkov, was based precisely on such technical means.

Author - Maxson