May 21, 2015 at 16:54Russia's technological lag in microprocessors based on the technological process in production
Against the background of the upcoming serial production, there are already 3 Russian-made processors - Elbrus 8C (MTsST), Baikal (T-Platforms) and Multiclet R1, I decided to analyze Russia's technological backlog in microprocessors based on the technological process in the production of the processor in Russia / USSR and among the leaders in this area.
Any structuring data for Russia for the entire period of the existence of microelectronics, or at least part of it, I have not found.
There were no problems with the data on the leaders - they are in a variety of sources, for example, in the English Wikipedia or in Russian . Since the correctness of these data does not raise any doubts, it is not worth dwelling on this.
The data for Russia / USSR was more difficult. Several directories were used, in particular sovietcpu.com . The development dates and those of the process were checked in various sources, in particular, these are manufacturers' websites, and reference books, for example, academic.ru and others , were used on Soviet processors .
Minimum listing criteria:
It turned out to be convenient to draw graphs on two grounds - backwardness in years and backwardness in multiple numerical terms. I’ll note right away: the graphs reflect the ability to produce (no matter where) the necessary ones. process, not the number of devices manufactured, but the condition is at least small-scale production.
The graph of backwardness in years displays the rate of lag in the transition to new technological processes. Transitions are rare and jerky, from which the processors have time to pretty old. So, in 1997, the lag reached its peak, since at that time Russia could produce processors only for those. 1 micron process, which first appeared in 1985, that is, 12 years ago relative to 1997.
The second chart is more interesting because it displays the lag for those. process in multiple numerical terms. The maximum lag falls not on the 90s as you might expect, but in 2014 when Elbrus 4C with 65 nm comes out (the leaders have already mastered 14nm), it turns out that the Russian processor was then "4.6 times" larger. However, already in 2015, with the advent of 32 nm Elbrus 8C and Baikal processors, the lag is reduced to 2 times. Minimum separation of competitors occurred in 1991-1993 (only 1.3 times) and 2004-2005 (1.4 times).
The first Soviet processors made by those. The 10 micron process appeared in 1974. This was the K587 series, which was commissioned 3 years later than its analogues.
Three years later, in 1977, a series of K580, a clone of i8080, manufactured by those. 6 micron process. One of the first models is the K580IK80. After it was released a lot of processors for this tech. process. Remarkably, almost all of them were clones of Western companies.
The transition to 3 μm was carried out only in 1982, with the advent of the processor K1801VM1.
In the 80s, when the United States imposed sanctions against the USSR, the development of the electronics industry in the country slowed down significantly. The sanctions had the following provision:
Nevertheless, the transition to 1.5 μm was made in 1989 with the advent of KR1847VM286, a copy of i286, and in 1991 they switched to 1 μm - L1876VM1 (i386). Then the USSR was inferior to competitors in those. process only 1.3 times. This was the best result in the entire history of Russia / USSR.
Unfortunately, in the next 7 years the situation began to degrade. The reasons we all know are the economic crisis and political instability.
By 1997, Russian enterprises could not produce anything more complicated than a copy of the i386, the backlog was 12 years. It was independently incapable of breaking the barrier in the form of several generations of processors by a country that had just woken up from shocks, therefore, having designed its new R80 processor for those. the 0.5 micron process on the SPARC architecture, MCST began production in France.
In large quantities, the R80 and its slightly improved version of the R100 were not produced - but still it was a real breakthrough for Russia. After 3 years, on this basis, a more modern R150 (350 nm) was developed - it went into series, production was established in Taiwan.
Further events developed much more dynamically - in 2004 the transition to 130 nm (R500), and in 2010 - to 90 microns (R1000). The production was transferred from Taiwan to Zelenograd near Moscow, which is very important since before that there were no capacities in Russia capable of working according to those needs. processes.
In April 2014, the ICTS introduced 4 nuclear Elbrus-4C, made using 65 nm technology. And although this, given the ability to produce in Russia, was a real breakthrough, it was still a product that was 8 years behind competitors.
At the beginning of 2015, the MCST introduced the 8th nuclear Elbrus-8C, made using 28 nm technology. Almost at the same time, the T-platform company announced its 28 nm Baikal processor, based on the licensed ARM Cortex A57.
Based on 2 graphs, the following conclusions can be drawn regarding the Soviet and Russian microelectronic industry:
1) Despite the fact that the pace of transition to new technological processes in Russia before 2014 compared to the USSR has slightly increased, the lag for those. process increased on average.
2) The technological lag, although quite large, is less than many people imagine.
3) If you count only with processors made in Russia, the picture will be slightly worse. However, many countries have experience in the production of their processors in other countries. In addition, the manufacturer earns only 10-15%.
Analysis of the near future suggests that the lag until 2020 by years will be fixed for 4-5 years. In the MSCT plan for the development of microelectronics for 2018, a transition to 16 nm is planned (14 nm is already being mastered by competitors), by 2020 a transition to 10 nm (the same as competitors by 2016).
Probably, for more dynamic development, much larger investments and a business plan are needed, built not on defense orders, but on an ordinary consumer.
Any structuring data for Russia for the entire period of the existence of microelectronics, or at least part of it, I have not found.
Data collection
There were no problems with the data on the leaders - they are in a variety of sources, for example, in the English Wikipedia or in Russian . Since the correctness of these data does not raise any doubts, it is not worth dwelling on this.
The data for Russia / USSR was more difficult. Several directories were used, in particular sovietcpu.com . The development dates and those of the process were checked in various sources, in particular, these are manufacturers' websites, and reference books, for example, academic.ru and others , were used on Soviet processors .
Minimum listing criteria:
- Central Processing Units
- Original Russian architecture
- Licensed architecture
- Possibility of production in Russia or abroad
- At least small-scale production
- Turn on the first mention
From these data, we can distinguish the main years of transition to new ones. the processes
P.S. L1876VM1 was developed in the USSR, but was produced in Belarus, so it is also included in the list.
P.S. Also, the USSR in the 90s in Germany had capacities for production using 1 micron technology.
| Year of transition | Those. process | Representative |
|---|---|---|
| 1974 | 10 microns | K587 Series |
| 1977 | 6 microns | K580IK80 |
| 1982 | 3 microns | K1801VM1 |
| 1989 | 1.5 μm | KR1847VM286 |
| 1991 | 1 μm | L1876VM1 |
| 1998 | 500 nm | MCST R80 |
| 2001 | 350 nm | MCST R150 |
| 2004 | 130 nm | MCST R500 |
| 2010 | 90 nm | MCST R1000 |
| 2014 | 65 nm | Elbrus-4C |
| 2015 | 28 nm | Elbrus-8S |
P.S. L1876VM1 was developed in the USSR, but was produced in Belarus, so it is also included in the list.
P.S. Also, the USSR in the 90s in Germany had capacities for production using 1 micron technology.
It turned out to be convenient to draw graphs on two grounds - backwardness in years and backwardness in multiple numerical terms. I’ll note right away: the graphs reflect the ability to produce (no matter where) the necessary ones. process, not the number of devices manufactured, but the condition is at least small-scale production.
The graph of backwardness in years displays the rate of lag in the transition to new technological processes. Transitions are rare and jerky, from which the processors have time to pretty old. So, in 1997, the lag reached its peak, since at that time Russia could produce processors only for those. 1 micron process, which first appeared in 1985, that is, 12 years ago relative to 1997.
The second chart is more interesting because it displays the lag for those. process in multiple numerical terms. The maximum lag falls not on the 90s as you might expect, but in 2014 when Elbrus 4C with 65 nm comes out (the leaders have already mastered 14nm), it turns out that the Russian processor was then "4.6 times" larger. However, already in 2015, with the advent of 32 nm Elbrus 8C and Baikal processors, the lag is reduced to 2 times. Minimum separation of competitors occurred in 1991-1993 (only 1.3 times) and 2004-2005 (1.4 times).
A small perspective on history and the reasons for the ups and downs
The first Soviet processors made by those. The 10 micron process appeared in 1974. This was the K587 series, which was commissioned 3 years later than its analogues.
Three years later, in 1977, a series of K580, a clone of i8080, manufactured by those. 6 micron process. One of the first models is the K580IK80. After it was released a lot of processors for this tech. process. Remarkably, almost all of them were clones of Western companies.
The transition to 3 μm was carried out only in 1982, with the advent of the processor K1801VM1.
In the 80s, when the United States imposed sanctions against the USSR, the development of the electronics industry in the country slowed down significantly. The sanctions had the following provision:
Prohibit technology shipments: nothing more complicated than IBM 360 (1964)
Nevertheless, the transition to 1.5 μm was made in 1989 with the advent of KR1847VM286, a copy of i286, and in 1991 they switched to 1 μm - L1876VM1 (i386). Then the USSR was inferior to competitors in those. process only 1.3 times. This was the best result in the entire history of Russia / USSR.
Unfortunately, in the next 7 years the situation began to degrade. The reasons we all know are the economic crisis and political instability.
By 1997, Russian enterprises could not produce anything more complicated than a copy of the i386, the backlog was 12 years. It was independently incapable of breaking the barrier in the form of several generations of processors by a country that had just woken up from shocks, therefore, having designed its new R80 processor for those. the 0.5 micron process on the SPARC architecture, MCST began production in France.
In large quantities, the R80 and its slightly improved version of the R100 were not produced - but still it was a real breakthrough for Russia. After 3 years, on this basis, a more modern R150 (350 nm) was developed - it went into series, production was established in Taiwan.
Further events developed much more dynamically - in 2004 the transition to 130 nm (R500), and in 2010 - to 90 microns (R1000). The production was transferred from Taiwan to Zelenograd near Moscow, which is very important since before that there were no capacities in Russia capable of working according to those needs. processes.
In April 2014, the ICTS introduced 4 nuclear Elbrus-4C, made using 65 nm technology. And although this, given the ability to produce in Russia, was a real breakthrough, it was still a product that was 8 years behind competitors.
At the beginning of 2015, the MCST introduced the 8th nuclear Elbrus-8C, made using 28 nm technology. Almost at the same time, the T-platform company announced its 28 nm Baikal processor, based on the licensed ARM Cortex A57.
Summary
Based on 2 graphs, the following conclusions can be drawn regarding the Soviet and Russian microelectronic industry:
1) Despite the fact that the pace of transition to new technological processes in Russia before 2014 compared to the USSR has slightly increased, the lag for those. process increased on average.
2) The technological lag, although quite large, is less than many people imagine.
3) If you count only with processors made in Russia, the picture will be slightly worse. However, many countries have experience in the production of their processors in other countries. In addition, the manufacturer earns only 10-15%.
Analysis of the near future suggests that the lag until 2020 by years will be fixed for 4-5 years. In the MSCT plan for the development of microelectronics for 2018, a transition to 16 nm is planned (14 nm is already being mastered by competitors), by 2020 a transition to 10 nm (the same as competitors by 2016).
Probably, for more dynamic development, much larger investments and a business plan are needed, built not on defense orders, but on an ordinary consumer.