Let's go back to Lebedev's adventures in Moscow. He went there not as a savage, but at the invitation of the aforementioned M. A.
The Institute of Precision Mechanics and Computer Science was originally organized in 1948 to calculate (mechanically and manually!) Ballistic tables and perform other calculations for the Department of Defense (in the United States, by that time, ENIAC was working on similar tables, and there were several more machines in the project) … Its director was Lieutenant General N. G. Bruevich, a mechanic by profession. Under him, the institute was focused on the development of differential analyzers, since the director did not represent any other technique. In mid-1950, Bruyevich (according to Soviet tradition, directly through a letter to Stalin) was replaced by Lavrentyev. The displacement took place through a promise to the leader to create a machine for calculating nuclear weapons as soon as possible.
For this, he lured the talented Lebedev from Kiev, where he had just completed the construction of the MESM. Lebedev brought 12 notebooks filled with drawings of an improved version of the machine, and immediately got into work. In the same 1950, Bruevich struck Lavrentiev in retaliation, offering ITMiVT "fraternal assistance" from the USSR Ministry of Mechanical Engineering and Instrumentation. The ministers "advised" (as you understand, there was no option to refuse) ITMiVT to cooperate with SKB-245 (the same where later director V. V. Aleksandrov did not want to "see and know" the unique Setun machine and where from Brook Rameev), Scientific Research Institute "Schetmash" (previously developing adding machines) and the SAM Plant, which produced these adding machines. Satisfied assistants, having studied Lebedev's project, immediately made a proposal, telling Minister PI Parshin that they themselves would master the creation of a computer.
Strela and BESM
The minister immediately signed an order on the development of the Strela machine. And the three competitors somehow managed to complete its prototype just by the time the BESM was tested. SKB had no chances, Strela's performance was no more than 2 kFLOPS, and BESM-1 produced more than 10 kFLOPS. The ministry was not asleep and told Lebedev's group that only one copy of RAM on fast potentioscopes, which was vital for their computer, was given to Strela. The domestic industry allegedly did not master the larger party, and BESM is already working well, it is necessary to support colleagues. Lebedev urgently remakes the memory for obsolete and bulky mercury delay lines, which reduces the prototype's performance just to the level of "Strela".
Even in such a castrated form, his car utterly breaks a competitor: 5 thousand lamps were used in BESM, almost 7 thousand in "Strela", BESM consumed 35 kW, "Strela" - 150 kW. The presentation of data in the SKB was archaic - BDC with a fixed point, while BESM was real and completely binary. Equipped with advanced RAM, it would have been one of the best in the world at that time.
There is nothing to do, in April 1953 BESM was adopted by the State Commission. But … it was not put into series, it remained the only prototype. For mass production, the "Arrow" is chosen, produced in the amount of 8 copies.
In 1956, Lebedev knocks out potentioscopes. And the BESM prototype becomes the fastest car outside the United States. But at the same time, the IBM 701 outperforms it in technical specifications, using the latest memory on ferrite cores. The famous mathematician MR Shura-Bura, one of the first programmers of Strela, did not remember her very warmly:
The “Arrow” was put in the Department of Applied Mathematics. The machine worked poorly, it had only 1000 cells, an inoperative magnetic tape drive, frequent malfunctions in arithmetic and a host of other problems, but, nevertheless, we managed to cope with the task - we made a program to calculate the energy of explosions when simulating nuclear weapons …
Almost everyone who had the dubious happiness of touching this miracle of technology made such an opinion about her. Here is what A. K. Platonov says about Strela (from the interview we have already mentioned):
The director of the institute, which was making computing equipment, which was used at that time, did not cope with the task. And there was a whole story: how they persuaded Lebedev (Lavrentyev persuaded him), and Lavrentyev became the director of the institute, and then Lebedev became the director of the institute instead of that "unsuccessful" academician. And they made BESM. How did you do it? Collected graduate students and term papers of the physics departments of several institutes, and the students made this machine. First, they made projects on their projects, then they made iron in the workshops. The process began, aroused interest, the Ministry of Radio Industry joined in …
When I came to this car with BESM, my eyes went up to my forehead. The people who made it just sculpted it out of what they have. There was no idea there, that is, I could hardly do anything with it! She knew how to multiply, add, divide, had a memory, indeed, and she had some kind of tricky code that you can't use … You give the IF command and you have to wait eight commands until the path under the head fits there. The developers told us: just find what to do in these eight commands, but because of this it turned out eight times slower … SCM in my memory is a kind of freak … BESM had to give 10,000 operations … But, because of the replacement [memory], BESM on tubes gave only 1000 operations. Moreover, all calculations on them were carried out 2 times, necessarily, because these mercury tubes often got lost. When we later switched to electrostatic memory … the whole team of young guys - after all, Melnikov and others were still boys - rolled up their sleeves and redid everything. We did our 10 thousand operations per second, then increased the frequency and they got 12 thousand. I remember that moment. Melnikov says to me: “Look! Look, I'll give the country another Strela now! " And on this oscillator turns the knob, just increasing the frequency.
TK
In general, the architectural solutions of this machine are now practically forgotten, but in vain - they perfectly demonstrate a kind of technical schizophrenia, which the developers had to follow largely through no fault of their own. For those who are not in the know, in the USSR (especially in the military field, which included all computers in the Union until the mid-1960s), it was impossible to officially build or invent anything, acting freely. For any potential product, a group of specially trained bureaucrats would first issue a technical assignment.
It was basically impossible not to meet the TK (even the strangest, from the point of view of common sense) - even an ingenious invention would not have been accepted by a government commission. So in the technical assignment for "Strela" was indicated the requirement of the obligatory possibility of working with all machine units in thick warm gloves (!), The meaning of which the mind is not able to comprehend. As a result, the developers were as perverted as they could. For example, the notorious magnetic tape drive used reels not of the global 3⁄4”standard, but 12.5 cm, so that they could be charged in fur mittens. In addition, the tape had to withstand a jerk during a cold start of the drive (according to TZ –45 ° C), so it was super thick and very strong to the detriment of everything else. How a storage device can have a temperature of -45 ° C, when a 150 kW lamp battery is running a step away from it, the compiler of the technical specification definitely did not think.
But the secrecy of SKB-245 was paranoid (in contrast to the BESM project, which Lebedev did with the students). The organization had 6 departments, which were designated by numbers (before that they were secret). Moreover, the most important, 1st department (according to tradition, later in all Soviet institutions this very "1st part" existed, where specially trained people from the KGB sat and secret everything that was possible, for example, in the 1970s, the "first departments" were responsible for access to a strategic machine - a copier, otherwise employees will suddenly begin to propagate sedition). The entire department was engaged in daily checks of all other departments, every day the SKB employees were given suitcases with papers and stitched, numbered, sealed notebooks, which were handed over at the end of the working day. Nevertheless, for some reason, such an outstanding level of bureaucratic organization did not allow the creation of an equally outstanding machine.
It is striking, however, that "Strela" not only entered the pantheon of Soviet computers, but was also known in the West. For example, the author of this article was sincerely surprised to find, in C. Gordon Bell, Allen Newell, Computer Structures: Reading and Examples, published by the McGraw-Hill Book Company in 1971, in a chapter on various command set architectures, a description of Arrow commands. Although it was cited there, as is clear from the preface, rather, for the sake of a curiosity, since it was rather intricate even by tricky domestic standards.
M-20
Lebedev learned two valuable lessons from this story. And for the production of the next machine, the M-20, he moved to the competitors favored by the authorities - the same SKB-245. And for patronage he appoints as his deputy a high rank from the Ministry - M. K. Sulima. After that, he begins to drown the competing development - "Setun" with the same ardor. In particular, not a single design bureau undertook to develop documentation vital for mass production.
Later, the vindictive Bruevich dealt the last blow to Lebedev.
The work of the M-20 team was nominated for the Lenin Prize. However, the work was rejected for unspecified reasons. The fact is that Bruevich (who was then an official of the Gospriyemka) wrote down his dissenting opinion in addition to the act on the acceptance of the M-20 computer. Referring to the fact that the US is already running a military computer IBM Naval Ordnance Research Calculator (NORC), allegedly producing more than 20 kFLOPS (in reality, no more than 15), and "forgetting" that the M-20 has 1600 lamps instead of 8000 NORC, he expressed great doubts about the high quality of the machine. Naturally, no one began to argue with him.
Lebedev learned this lesson too. And Sulim, already familiar to us, became not just a deputy, but a general designer of the following machines M-220 and M-222. This time everything went like clockwork. Despite the numerous shortcomings of the first series (by that time, a poor ferrite-transistor element base, a small amount of RAM, an unsuccessful design of the control panel, high labor intensity of production, a single-program console mode of operation), 809 sets of this series were produced from 1965 to 1978. The last of them, 25 years old, were installed back in the 80s.
BESM-1
It is interesting that BESM-1 cannot be considered purely lamp-based. In many blocks, ferrite transformers rather than resistance lamps were used in the anode circuit. Lebedev's student Burtsev recalled:
Since these transformers were made in an artisanal way, they often burned out, while giving off a pungent specific smell. Sergei Alekseevich had a wonderful sense of smell and, sniffing the rack, pointed to the defective one up to a block. He was almost never wrong.
In general, the results of the first stage of the computer race were summed up in 1955 by the Central Committee of the CPSU. The result of the pursuit of academicians' chairs and foundations was disappointing, which is confirmed by the corresponding report:
The domestic industry, which produces electronic machines and devices, does not make sufficient use of the achievements of modern science and technology and lags behind the level of a similar industry abroad. This lag is especially clearly manifested in the creation of high-speed calculating devices … The work … is organized on a completely insufficient scale, … not allowing to catch up and, moreover, to outstrip foreign countries. SKB-245 MMiP is the only industrial institution in this area …
In 1951, there were 15 types of universal high-speed digital machines in the USA with a total of 5 large and about 100 small machines. In 1954, the United States already had over 70 types of machines, totaling over 2,300 pieces, of which 78 were large, 202 were medium, and over 2,000 were small. At present, we have only two types of large machines (BESM and "Strela") and two types of small machines (ATsVM M-1 and EV) and only 5-6 machines are in operation. We are lagging behind the USA … and in the quality of the machines we have. Our main serial machine "Strela" is inferior to the serial American machine IBM 701 in a number of indicators … Part of the available manpower and resources is spent on performing unpromising work that lags behind the level of modern technology. Thus, the electromechanical differential analyzer with 24 integrators manufactured in SKB-245, which is an extremely complex and expensive machine, has rather narrow capabilities in comparison with digital electronic machines; abroad from the manufacture of such machines refused …
Soviet industry also lags behind foreign industry in technology for the production of computers. So, abroad, special radio components and products are widely produced, which are used in calculating machines. Of these, germanium diodes and triodes should be indicated in the first place. The production of these elements is being successfully automated. The automatic line at the General Electric plant produces 12 million germanium diodes per year.
At the end of the 50s, squabbles and discord among designers associated with an attempt to get more funding from the state for their projects and drown others (since the number of seats in the Academy of Sciences is not rubber), as well as a low technical level, which hardly allows the production of such complex equipment, led to the fact that at the beginning of the 1960s, the park in general of all lamp machines in the USSR was:
In addition, until 1960, several specialized machines were produced - M-17, M-46, "Kristall", "Pogoda", "Granit", etc. In total, no more than 20-30 pieces. The most popular Ural-1 computer was also the smallest (100 lamps) and the slowest (about 80 FLOPS). For comparison: the IBM 650, the former more complex and faster than almost all of the above, was produced by that time in more than 2,000 copies, not counting other models of this company alone. The level of lack of computer technology was such that when in 1955 the country's first specialized computing center was created - the Computing Center of the USSR Academy of Sciences with two whole machines - BESM-2 and Strela, computers in it worked around the clock and could not cope with the flow of tasks (one is more important than the other).
Bureaucratic absurdity
It came, again, to the bureaucratic absurdity - so that the academics would not fight over the overvalued machine time (and, according to tradition, for total party control of everything and everyone, just in case), the plan of calculations on the computer was approved, and on a weekly basis, personally by the chairman of the USSR Council of Ministers N. A. Bulgarin. There were other anecdotal cases as well.
For example, academician Burtsev recalled the following story:
BESM began to consider tasks of particular importance [that is, nuclear weapons]. We were given security clearance, and the KGB officers very meticulously asked how information of special importance could be extracted and removed from the car … We understood that every competent engineer can extract this information from anywhere, and they wanted it to be one place. As a result of joint efforts, it was determined that this place is a magnetic drum. A plexiglass cap was built on the drum with a place to seal it. The guards regularly recorded the presence of a seal with the entry of this fact into the journal … Once we began to work, having received some, as Lyapunov said, an ingenious result.
- And what to do next with this brilliant result? “He’s in RAM,” I ask Lyapunov.
- Well, let's put it on the drum.
- Which drum? He was sealed by the KGB!
To which Lyapunov replied:
- My result is a hundred times more important than anything written and sealed there!
I recorded his result on a drum, erasing a large pool of information recorded by atomic scientists….
It was also lucky that both Lyapunov and Burtsev were necessary and important enough people not to go to colonize the Kolyma for such arbitrariness. Despite these incidents, the most important thing is that we had not yet begun to lag behind in production technology.
Academician N. N. Moiseev got acquainted with the US tube machines and wrote later:
I saw that in technology we practically do not lose: the same tube computing monsters, the same endless failures, the same magician engineers in white coats who fix breakdowns, and wise mathematicians who are trying to get out of difficult situations.
A. K. Platonov also recalls the difficulty of gaining access to BESM-1:
An episode is recalled in connection with BESM. How everyone was kicked out of the car. Her main time was with Kurchatov, and they were told not to give anyone time until they finish all the work. This greatly angered Lebedev. Initially, he allocated time himself, and did not agree with such a requirement, but Kurchatov knocked out this decree. Then I ran out of time at eight o'clock, I have to go home. The Kurchatov girls with punched tapes just enter. But behind them enters an angry Lebedev with the words: "This is wrong!" In short, Sergei Alekseevich sat down at the console himself.
At the same time, the battle of academics for lamps took place against the backdrop of the amazing literacy of the leaders. According to Lebedev, when, in the late 1940s, he met with representatives of the Central Committee of the Communist Party in Moscow to explain to them the importance of financing computers, and spoke about the theoretical performance of MESM in 1 kFLOPS. The official thought for a long time, and then gave out a brilliant:
Well, here, get the money, make a car with it, she will instantly recount all the tasks. Then what will you do with it? Throw it away?
After that, Lebedev turned to the Academy of Sciences of the Ukrainian SSR and already there he found the necessary money and support. By the time when, according to tradition, looking to the West, domestic bureaucrats saw their sight, the train almost left. We managed to produce no more than 60–70 computers in ten years, and even then up to half of experimental ones.
As a result, by the mid-1950s, an amazing and sad situation had developed - the presence of world-class scientists and the complete absence of serial computers of a similar level. As a result, when creating missile defense computers, the USSR had to rely on traditional Russian ingenuity, and the hint as to which direction to dig came from an unexpected direction.
There is a small country in Europe that is often ignored by those with a superficial knowledge of the history of technology. They often recall German weapons, French cars, British computers, but they forget that there was one state, thanks to its uniquely talented engineers, which in 1930-1950 achieved no less, if not great success in all these areas. After the war, fortunately for the USSR, it firmly entered its sphere of influence. We are talking about Czechoslovakia. And it is about Czech computers and their main role in creating the missile shield of the Country of the Soviets that we will talk about in the next article.
