Center for Strategic Assessment and forecasts

Autonomous non-profit organization

Home / Science and Society / Direction materials
Eduard projdakov: "At the time you pulled electronics personal computers. Now it will pull the robots"
Material posted : Administrator Publication date: 09-08-2019

Eduard projdakov — Director of the Virtual computer Museum, developer, expert in the field of information technology, lecturer, journalist and translator. In an interview for the Museum of DataArt Eduard Mikhailovich told about the expeditions in the deserts and swamps, the ticket reservation system of the Soviet "Aeroflot", the Oginski Polonaise on lamp M-20, programmers-romantics of the 1960s, architectural crisis and its overcoming.

Surveyor

— As far as we know, before engaging in information technology, you have mastered another profession.
— Yes, and in General I have a pretty rough biography. Let's start with the fact that I was born in Collage and the first years of life, you can say, spent. Then I was adopted with my parents I went to the Virgin lands in Biysk. Finally, he moved with them in the suburbs, with a medal graduated from high school and a few years hanging out on expeditions, as in the Moscow Institute of engineers land (MIIZ) received an engineer-surveyor. Geodesic expedition — it was something. Well, four years of living in the Dorm is also highly educated.

— What is the hostel Soviet times?
In each room there lived five people. Dropped two per week to buy food. Breakfasts and dinners were cooked on duty in the room. We had some of their chips. For example, when we are going to buy a TV KVN with a water lens, introduced a system of fines for behavior and for the week gained the required amount. Fined for a mate, for what man has brought the girl in the evening, and all were forced to leave. It was the biggest fine, almost ruble.

I had a personal and epic, because my girl lived in St. Petersburg, and I'm a year and a half every Saturday and Sunday it was gone. It cost 10 rubles 20 kopecks on a student, if the ticket is second-class. Then my student to Petersburg three years went by my best friend, who eventually got married there and moved there. He later became head gardener of the Pavlovsk Park.

— Tell us about the work of the surveyor.
Here was your problem. First, it is very difficult to find a job — people who would go with you to the mountains, the forest, the desert, shared this difficult life. Often these were the guys after the conclusion or people of unusual destiny. We, if able, satisfied qualifying exam — when 30 people were left alone, it was good.


Gennady Dmitriev. "Early summer", 1977

— You studied science?
— No. Now there are satellites and then there was aerial photography from aircraft. Here he grew wings, took off, then the pictures need to be tied to the terrain. I.e. one of the tasks is to locate the ground point, which is present in the picture, and chop her in the picture with the tip of the needle. This so-called identification mark. Next you need to determine its location is to do some measuring angles with a theodolite, and bind height. In addition, it's necessary to sketch.

Gives you some area. The plane flew, took a strip of 20 km on the 10. Need to find three points on the. And here you're walking through the desert, there's nothing. Looking for traces of the wheels of vehicles, old droppings. Quickly locate the landmarks is a great art.
We did a large — scale survey for maps of scale 1:5000 and 1:10000. At the time, was the program, when did the map of the Soviet Union in the scale of 1:25,000. It's a tremendous work, a lot of people worked. In Leningrad had its own aerial, engaged in the Baltic States — I was there for one season worked. Met a unique person in all expedition walked. He did not recognize transport except Bicycle and horse. Two weeks before the start of the season came from Peter on foot in the direction of Tartu. And then the expedition is a day from 30 miles of walking and 45. And in the swamps. Well, the summer was dry, but I lost 12 pounds.

— Paid well?
— In addition to regular salary, was supposed 30 percent — the so-called field. The money of the people lived, and when he arrived at the base, got everything else. At the end of the season, sometimes running well. The work piece — as stamped and burst.

— Why did you decide to change professions.
In 1972 I got married and realized that expedition and family life incompatible. Plus in expeditions the conditions were quite severe, and people specific. And I went on to study engineering stream of the Moscow Institute of electronic engineering, the MIEM, now it is part of the Higher school of Economics. A great University. Gave virtually the same mathematics, scientific communism was trying to flog likely to form. Well, the level of teaching, Turning out in those days was just great.

The main the Main entrance is given at the intersection is Small. Pioneer street and Stremyanny pereulok. Photo taken between 1978 and 1982.

While studying I worked for three years on distribution. Been doing some very interesting work — a survey of underground communications in Moscow. Them here the sea is all I had in this whole system to understand. Well, when we ran out of time, as a senior engineer I went down in salary, and the rank of an engineer-mathematician of the Main computing centre of civil aviation of the USSR.

The icon with the cruise terminal — the emblem of the main computer center of civil aviation

MCC HA

— It was in 1975, and in MCC I had a very interesting project "Sirena-2" system of mass service for the sale of airline tickets. The Soviet Union had about 10,000 flights per week. One of the tasks that I was trying to observe was called "connecting flights". You fly where there is no direct flight and you need somewhere to change. First, we need to calculate that you have time — for a transplant is usually laid 4 hours. Secondly, it is necessary to choose the shortest route and to optimize its value.

But in those days there were no machines capable of keep a table in memory of the 10 thousand flights. Therefore, the "Siren" was used only for the Moscow air hub. The depth of the booking started with 30 days, then did 45. At the same time, in Western companies, in particular, Air France and some American, have booked flights for the year. They have a memory was much better, for example, stood the colossal magnetic drum at the time capacity is 800 megabytes. And he was very fast, 800 tracks by megabytes each.

— What is the difference between "Siren-1" and "Siren-2"?
— "Siren-1" worked on very old machines M-3000. My work began with a one-month internship in her statement.

The M-3000 was the structure and architecture of the IBM system 360, adjusted for available in the USSR details

Documentation for this system has not remained, therefore, at first the people were sitting and restored the original text, as it is now called. The system was written in Assembler, and 20 people have to take it apart to blocks. It is such research, when the content of the memory you're trying to understand what was programmed. Here I have a big love for the disassembler. I wrote them in my life very much and became a specialist in this area. Sometimes you gets a piece of some system, and we need to understand what it was. For example, burned down a very expensive installation, it is necessary to restore it. For this we need to restore the firmware. Restored.

"Siren-2" in the beginning was supposed to be done on the M-4030. It is absolutely wonderful naumovska machine (INEUM — Institute of electronic control machines). She's got a great story, almost a detective. At the time, the British company ICL made the Spectra 70 machine. When the USSR chose what to copy, IBM System/360 or ICL, was a voice that ICL better architectural. But the Union won IBM and ICL stopped in Germany. Siemens took the prototype was just 70 Spectra and produced a car System 4004 and INEUM at its base is made of M-4030.

Computer control system M-4000

M-4030 I liked much more than all the others. I have five years worked as it system programmer. He graduated from half-year courses and do a lot of things could it do. And she was very reliable. The Soviet equipment was two problems: bad plastic circuits and contacts. M-4030 for some conditions produced with gilded contacts. Others are not, because there were very strict limits on the precious metals. For example, we could decorate the car with Nickel. Nickel is a strategic metal.

"Siren"-2've wanted to do for the Olympics-80, but launched in 1981, and is 2005 has worked. Then bought to replace IBAMOSCOW system. I from the civil aviation authority has already left — in the 77th year went to work in cniika (Central research Institute of complex automation). Just realized that with these leaders of development will not work.

— M-4030 — it was a clone or their own development?
— His own. Often say, "Loggerheads!", however, loggerheads can the idea of architecture, but not the circuitry. First, we have a different circuitry. Have to do the same functionality but completely different means. We have other technology, other standards. American technology inch connectors, we have a metric. They are not compatible — just charge won't stick.

— When did you start working with large machines as it looked?
The first time we tried to work out how all the so-called batch mode. Write on paper, give it to the girls, they are stuffed, the operator starts the deck, give you a printout with your errors, you correct them, and repeat the process. But it was a very long time. And when we did the system that had to pass, got to we were allowed to go out at night when there are no operators. It went much faster.

— There was a separation: operators, programmers, technical team. How were you allowed to do the work of the operators?
— This is a very conditional division. Just ahead was the so-called batch mode. When you make a card, give operators a lot of instructions which tapes to put which discs to take. This is due to the fact that programmers didn't have access to the machine. There were plenty of cars. Initially almost, the Deputy Minister shared computer time between organizations, and then, when the machines became enough the need for it disappeared. Over time, they learned to one machine connect a sufficient number of terminals, there was a regime of collective work. In the end, the profession of "computer operator" has disappeared. All realized that much more productive when people work directly with the machine, especially the developers. Efficiency increases on the order. There are quite a few professions that are disappearing with the development of technology. Typist here disappeared. I remember the typing pool — come, order a print of a letter or document of some kind, and 20 typists in the sweat of pounding on the keys.

— What terminal did you work?
The first terminal of the electric typewriter. "Consul" was called, can't remember the number. What you typed, recorded on tape, as the log file. Well, the responses the machine was fixed, because the monitor was not. Machine printed rather slowly. I remember one of razvlekuh. On UCS, I had to learn the language PL/1. People called him the "Eaten one". When the first languages in 1957, Fortran, COBOL later — had the idea to make a powerful universal language. IBM did and called it a Programming Language 1 — "programming language one". They believed that it will be a single language on all occasions. Description — 600 pages. I read. But, unfortunately, it proved to be very cumbersome, although some ideas were great, they then migrated to other languages. PL/1 had a great influence on subsequent developments, and to learn it, I wrote on the game "battleship". Introduced move, she tapped, hit or missed.

Came by night and if had the time, I was out. In parallel he taught at the Moscow mathematical school. Just read them languages. The Trinity was taught Fortran, the b — PL/1, honors — Assembler.

The age of romanticism

— You remember how you were introduced UCS?
— Around a single series were persistent internal struggle. In terms of the system IBM-360 was designed perfectly. A single line, something we haven't been in any series. In the "Ural", although there are now many former developers of the machines with inflated cheeks, like "we had a BESM". Line BESM was not. Plus it is very expensive to turn it into a production car for everyone in those days was extremely difficult.

I don't know the background of all sorts of clandestine games, but the main argument of those who struck UCS, was this: guys, with this machine we get a lot of software that will cover all our needs. Because at that time a million programmers in the Soviet Union was not. It is then all sorts of ACS was involved in 700 thousand people. You can argue effectively or ineffectively they worked, but it's competent people, trained, trained. Universities began to produce a lot of programmers, and it is in some way elite.

If we talk about the machines at the time, I would say that it was the age of romanticism — the first love and so on. Because there was a huge hope: a tool which gives you the opportunity to greatly enhance the level of human intelligence and solve the most unusual tasks.

1956-th year. We are working on machine translation when the machine memory is only one kinoslovo — equivalent to 8 kilobytes. Believed: "now we will do." When buried in technical problems: there is no memory, the performance is not enough understanding how it works is also not yet formed, was born science called "mathematical linguistics". Ie — a direct consequence of the failure of the first machine translation systems.

In the 1960s, made the first chess games. I mean, that was tremendous enthusiasm and the romance was. Although just appeared first languages, all saw that it is a living thing, it evolves and each generation grows strongly. Especially when there was a transistor machine, which is much more reliable.

— What was your first contact with a computer?
— Somewhere in the senior years, year in ' 71, we have arranged an excursion to some computer center, and there stood the machine, M-20, another lamp. Each team was making its sound, and the people were programmed to it, "Oginski Polonaise". She plays, we listen. Lights flash, everything is cool. But I never imagined that I will have to work closely. The first car was tube, then a transistor, I got on the car has the third generation, when I went to small and medium degree of integration.

M-20 Soviet tube electronic computing machine designed in 1955-1958

With M-20 I had absolutely amazing experience. I have worked in INEUM head of the laboratory of translators and a group of friends travelled in the Union republics, and has lectured about computing. After the lecture in Alma-ATA, we were lucky to watch a unique solar Observatory, located at an altitude of 2900 meters. Come, and there the native M-20 is. It was the year 1985, these machines have almost gone. The Almaty said that it is amazing that does not freeze, because of lamp warm up. For them it was the most.

Electricos of all Armenians

In the early 1980s, you worked at the Institute of problems of management.
Yeah, I was probably one of the youngest of Salamov, we are engaged in automation of experiments. 14 employees had 9 different types of machines, we had to know everything.

IPU is a wonderful place. Young team, mostly graduates. During the year they are taught, and then they work quite independently. There were a few bison. Nick Nadolsky — an exceptionally competent person. Only, perhaps, in the Union, a specialist in Smalltalk is a programming language. In General, languages in the world now, probably about 5,000. I am in my life have taught about 25. There are unique languages that you know two or three people.

Once we Tomek was lucky enough to be in the team of academician Iosifyan, developed "Istra-4816". He recruited people to his dacha in Barvikha. Staring at the man and said, "take This". Such testing took place and I was very proud.


Iosifyan said: "I electricos of all Armenians. Is the Catholicos, and I electricos". He is indeed the greatest scientist in the field of electrical engineering. Created VNIIEM — Institute of electromechanics at Red gate, this including a series of satellites "meteor". During the war Andronicus Gevondovich came up with these wedges, which were driven under an enemy tank and exploded. Only, it was not radio — over wedge stretched wire.


ET-1-627. Electrotinned-torpedo

The Germans quickly realized that I need to kill these leads and effectively learned to deal with wedgies. So on they go, but some number of tanks destroyed.
Iosifyan invented synchros, and all people unique. He, by the way, in the Caucasus the only winner of the Stalin prize. Just had to change the identity for the State prize, and he has managed to write Khrushchev at the 20th party Congress a note saying if I could leave as is, and all the note waved: "I Khrushchev allowed". Then some sort of reception he wore a badge of the winner. Shevardnadze said to him: "you sly Armenian!" Iosifyan was happy — washed! Because in Georgia no winner of the Stalin prize left.

Missed opportunities

— What's the last car in which you participated?
— If to consider what was actually done, it was not even a machine, and the controller 51-m processor. 51 I once wrote a command interpreter, the debugger like this, which was surrendered soft. This CPU was produced in Kiev, just on my software they have debugged the program. In the end we did a good controller. Already the collapse of everything in the world was I three years was with the Union letters — sell. To make a million dollars on it it was possible. But some of the guys from the team ran away, without fulfilling the promises. When I found the electronics, which helped it to make in gland, shaft orders passed. Something we have automated, somewhere to put. But it wasn't the same.

— What design impressed you the most?
I have used this in many projects involved, it is difficult to say. If you do something, the idea is to capture, to like. You and the team needs to believe in it. When all it is, is a great idea.

In General any project in computing is a trade — off between future and current capabilities. Each machine should be the idea, then she lives.


"Istra-4816", the Polytechnical Museum in Moscow

Here "Istra-4816", which is now in the Polytechnic Museum at the exhibition. Mass, it released 45 thousand pieces — at that time a good circulation. There were a few ideas. The first is a large circuit Board. Gilded contacts does not — much, so we need to make contacts was low. The fee is huge, 300 to 400 mm, it all fits, and it printed. The technology of printed circuit boards fairly well established, it is immediately guaranteed the high reliability of the machine.

Because there were few chips, we made the machine multiprocessor. These processors shared functions between them. Put an 8-bit processor that was doing I / o, but it has emulated a Western schema that physically we were not.
The second idea was that the next processor must be with your operating system. He was also 8-bit operating system CP/M-80, emulated bios (a basic input system output) and any functions associated with the instrument. We didn't know which operating system will win. In those days we had DOS, was Xenix is Unix, but on 86-m processor. And one of them will be ahead of the game, at the stage of development was not obvious. So we have made the basis on which it was possible to pull on any of these systems. Want MS DOS — it and put — I want Xenix, Linux, and so on. All this was possible due to the fact that there was a level that was based on a lower level, and which emulated any environment that needs.

Finally, the top was already 86 processor. Because we thought that it will be used in CAD, already had 4 megabytes of video memory. It was the 1982-1983 year, in the world of such machines was not. So it is everywhere met with enthusiasm, and immediately began to ruin and slow down. To the factory it was only after 6 years. But even on the decline of 45 thousand copies of the series survive.

— Why did you ruin?
Because there was a big fight. There were groups that believed that there should be good computer engineering, and organized sabotaged its implementation. I know a lot of examples of not only our development, but others when these things happened.

— As microelectronics have dealt with the problems of perestroika?
— We had a "Club 80", which consisted of the leading circuits of the country. Once a year in Riga we had gatherings. It worked quite ingenious developer Andrey Kolesnikov. He started with microprocessors, all systems, and then there was a huge project of digital exchanges.

If you can not read, he has a Stirlitz after the war he worked in Spain. And not aby where, and in ITT, which has developed the most ambitious in the time system of digital exchanges. It was called System 12, then they sold it to "Alcatel", that put the system in Russia, and we went to her opening in Surgut. Brilliant station, there are about 40 million lines of code written. We developed a special language of CHILL, the system wrote 8,000 people.

AT&T invented the "mode of eternal summer" for programmers. That is, they are worldwide built a camp, and people following the season moved from one to the other. A lot of what they invented, and in this work, I believe, was the first in the history of mankind an extremely challenging task. Then went soft for the American lunar programs. For the titans it was written 10 thousand programmers from the new York office of IBM, if I remember correctly. And the third is a modern operating system, the same Windows that too for 40 million rows exceeded. That is, a system that one person is unable to calculate in my entire life, the level of complexity exceeded a certain threshold. Therefore, qualification of managers of such projects is non-linear. It is quite unique people should be.

So, we were in Riga, assess what you can do, what architectural decisions. Just was the year when the Americans released the 386. And here it became clear that they do not catch up. There was a gap, which will continue to increase.

Why did this happen? The sector has been chronically underfunded. Somehow, it was believed that on old investments, we catch up with the West small improvements. Nothing of the sort. It was necessary to build new industries, to spend money on research and development. It did not, and the industry began a systemic crisis. The gap was growing, and at some point the process becomes irreversible.

The path of development

— Your first personal computer was a?
— The laptop in the early 90s. At work I have always had computers in large quantity and of any type. Working day on the development lasted 10-12 hours, so the computer was not really needed. And then, in the 90s, the head of one of the IT companies needed a technical book about the products that they released or going to release. I for laptop contracted her to write. Good computer "Toshiba". Healthy, thick, heavy. In those days happiness was.

— 12-hour working day. What about free time?
— It almost was not. My record is 23 hours behind the keyboard. The day before the surrender, when the last hours. Already the head of the Department went and drove everyone home. Worked with me master in karate, black belt. Look: sitting and head to the keyboard and Bang! Fell asleep.


Adored projdakov with co-author Leonid Teplitzky

Free time — often work too. One of my Hobbies — English-Russian explanatory dictionary on computer technology and programming, I'm writing for 28 years. He is now the biggest in the world, somewhere 40500 entries. I know of no other such in Russia it is exactly the one. He was transferred to the Ukrainian language, two editions came out. If you use the package Lingvo computer dictionary it our. This old hobby. Dictionary to start writing, but impossible to finish. Everyday masochism is about 10 entries in the dictionary. Moreover, the number of computer terms is only growing. By what words appear, you can see where there is development.

— And where is it coming from?
Computing equipment is based mainly on achievements in microelectronics. Still it was quite obvious — until acted Moore's law. Now it does not work, in my opinion. Just technology came to the border of physical possibilities, we began limited. To improve performance in these conditions you need to look for new architectural solutions.

Architectural crisis the people felt for a long time, and one solution was the creation of multicore processors. Not to build a system of caches, you make kernel simple and work thrown in between. Thus there was a 8-core CPUs, in fact, there are 60 nuclear — Intel they were shown. For graphics co-processors there is already 512, 1024 and more cores. But all this causes headaches because the person is extremely hard to think in parallel processes.


Eduard projdakov the program systems Institute of RAS

Remember, the ISP did EU-2000 — a parallel machine. When I found out that the people writes the sine to the library for three months, once precipitated. Exactly because of the difficulty of parallel programming. It is clear that, when the main library is written, then a little easier, but still.

There are other a lot of problems. For example, a million nodes, they need each other somehow to sync. They need somewhere to wait. And from a certain point, the challenge hanging in the compounds: the nodes are waiting for each other. Something is solved, but dramatically slowed. There is a risk to hang the task in the cloud formations. This way, in my opinion, is now exhausted. Moore's law expires, to make, for example, the processor, in which more than 512 cores, normal, not reduced, are already extremely difficult. Now the total capacity of the chip somewhere in the 40 billion valves is selected, and then there is the problem of consumption. Look at the legs of the chip. There's also a half feet is food. Ie supplied a pretty serious current. Is heat, there are a lot of restrictions. Then we must find another way.

What could it be? Now the main thing — artificial intelligence. It is necessary to make some special kind of calculation. In this way there were different solutions, but the biggest is what is called neuromorphic electronics. Electronics-internal architecture resembles a neural network of a person. Neuromorphic computing is now such an icon. IBM is moving there, some chips, they have already made. Is the idea of these neuromorphic chips is to collect something similar on the brain at least a mammal. This is an interesting and correct way, but difficult.

The second way in which trying to go — replace the electronic connections of the photon. Photon processor and other troubles, but some successes have been. I read about the switches that is made of pure optical. Will there be breakthroughs, I don't know.
In my opinion, there are still several areas that are associated with completely different mathematics. We were doing Alexander S. narinyani, which was created at the time under defined math. In his best works he was ahead of the Americans for 25 years, but have not received any support or funding. On his ideas the country could move forward, definitely.

Modern companies still pray for the Internet of things, but when you have a billion sensors, and that gives it a tremendous load on the network. It seems to me that the future still goes a little out there. It will go into robotics, artificial intelligence. At the time, personal computers had pulled the electronics. Now it will pull the robots.

Source: https://habr.com/ru/company/dataart/blog/463023/


RELATED MATERIALS:Science and Society
Возрастное ограничение