Russian Electronics: Semiconductor and Processors

It would be good for Russia to join the Chinese and invest/buy their DUV lithography machines. EUV machines might need to wait. But if the Chinese are after it, they will develop it and make it available commercially as well.

SMEE (Shanghai Micro Electronics Equipment) have already successfully developed their 28nm DUV machines which will mature in few years time. With these they might be able to go for 7nm with multiple exposures.

Not an experts on these matters, hope others with much expertise in this field might be able to give a better understanding on these stuffs.


@kvs

Can you please help me understand, what is it that Russia, Russian chips and Russian architecture (like Elbrus architecture) that lacks and which need to be overcome or developed further to have a proper competitor to the likes of Intel & AMD?

I wonder what company in perm that makes the gallium arsenide.

Also, they left out the current production of 65nm and 45nm. Probably because they are talking about latest and greatest tech. But as KVS said, Chiplet design is an alternative on 65nm.

Rodion_Romanovic wrote:

Dima wrote:It would be good for Russia to join the Chinese and invest/buy their DUV lithography machines. EUV machines might need to wait. But if the Chinese are after it, they will develop it and make it available commercially as well.

SMEE (Shanghai Micro Electronics Equipment) have already successfully developed their 28nm DUV machines which will mature in few years time. With these they might be able to go for 7nm with multiple exposures.

Not an experts on these matters, hope others with much expertise in this field might be able to give a better understanding on these stuffs.


@kvs

Can you please help me understand, what is it that Russia, Russian chips and Russian architecture (like Elbrus architecture) that lacks and which need to be overcome or developed further to have a proper competitor to the likes of Intel & AMD?

One of the issues with Elbrus processors is that they are manufactured in Taiwan... Unfortunately

Elbrus R-1000, 2CM, 4s was actually made locally in short supply and mass produced in TSMC. Anything lower nm was in tsmc.

Dima wrote:
@kvs

Can you please help me understand, what is it that Russia, Russian chips and Russian architecture (like Elbrus architecture) that lacks and which need to be overcome or developed further to have a proper competitor to the likes of Intel & AMD?

Im not kvs ,but will explain to you the problem with Russia high electronics and computer industry.

the big big problem with Russia high tech development is this.. is called vladimir putin.




Is the most anti high tech, anti space president of russia ever. he barely give funding to russia
most important business that was aiding russia ,to attract nations to its orbit. Cut in half for example the already low space budget , and in electronics he barely fund it.

Russia entire high tech busines could be very described as..  a Potemkin village.  

from wikipedia..

Potemkin village
In politics and economics, a Potemkin village is any construction whose sole purpose is to provide an external façade to a country that is faring poorly, making people believe that the country is faring better.

So this is exactly what the entire russia civilian high tech industry is a FACADE , to make it look that
russia is not outdated.. you can do this by painting buildings.. cleaning roads , making expensive luxury parks ,hosting olympics and so and so.. cosmetic development at best . things that are not bad ,but neither essential for a nation development .why is this?  because even though Russia design in papers the blue prints of their semiconductors.. guess who are the ones who build it in silicon ? who manufacture it?    TAIWAN..  a nation that is totally under control of US .  

Who was the genius who tought that was a good idea?
to ask taiwan help ,that will make send 20 years ago up to 15 years ago ,when russian president was starting ,so they learn the basics and later move to manufacturing..

Likely Even Russia military chips are manufactured by taiwan.  
and russia semiconductors are designed around the british designs.. 37 years ago..  
the ARM architecture family.

Russia also hired german companies to build a tiny manufacturing facility Angstrem  fab that produce (90 nm and 130 nm) this is a decade old or more technology, using european technology for old semiconductors , that was allowed to fall and went bankruptcy.   simply you can't build high tech electronics solid business withouot proper investment of the government ,and that was not going to happen with an very backward thinking  president of russia , that did not even saw the importance of internet and computers.. because he himself told years ago ,don't use the internet.  

and if that wasn't enough problems.. Russia also lacks in the software side..
computers without software aplications are useless.. the west have decades developing software.
russia experience in building software applications for commercial use is small in comparison with the west. and zero experience , if russia is forced now to build their own designs and manufacture it too.
which will be to start everything from zero.. 22 years later of putin in power.


Simply russia don't have a robust high tech industry ,because its backwards thinking president don't saw how important is  at all and high tech independent russia. but russia do have a lot of scientific talent.. but if not used is wasted talent.  So when he had the opportunity to start a high tech revolution ,he instead preffered to waste a fortune in olympics , upto 100 billions $ dollars spent from 2010 to 2018. and a fortune in other distractions,  potemkin cosmetic development in very expensive nature parks he build in moscow for tourist that will not arrive today because of sanctions.
and disney like parks he build in moscow ,full of american business.

So there is no better enemy for russia development that its own president , when you have a backward manager at the top , the entire nation becomes not far from a Potemkin village when it comes to the nation business ,culture and politics too.    the only thing that holds russia as a nation together is the military modernization and the social aid and the government promotion of patriotism..  but Russia could face the same problems of populist leaders in latin america.. that promised villages and castles to population and patriotism , a development in words.. just cosmetic development , but once they face sanctions from the West , the entire dream of independence collapse as it happened to cuba and venezuela.. and  for me there is no doubt ,that russia without help of china ,will become another failed state.. not as bad as cuba or venezuela.. but more like iran.

so to answer your question, it will be easier to say , that russia problems with high tech industry
is that have no industry at all.  It depends on the west , or china if they choose to help.  The only
realistic development i see for russia, is that russian engineers go to work for china that invested  trillion $$ dollars in their high tech electronics industry.  That way it can speed up china catch up
with the west in just a few years with a bit of luck. but also hire south korean retired engineers from samsung who worked building their own microprocessors..

why a nation so tiny like taiwan rule the world in semiconductors and high tech? because their leadership contrary to putin understand how important is high tech in nations development ,and when putin focused on pipelines economy with the west.. taiwan focused in high tech electronics.
so you get ,what you invest..  invest in building a banana african gas station economy and this is what you will get. invest heavily in high tech economy and sooner or later it will produce it.
Russia focuses most of its nation energies in gas stations ,, taiwan focused in high end electronics.. this is why russia have so much problems to influence the developed world.. because of bad leadership with outdated development of the country.

what other hopes there is for russia ?
to skip semiconductors and try graphene instead.. or optical computers , atomic computers ,or magnetic too..or that russia get lucky and their scientist manage to build complete quantum computers. but this are very long term things ,that takes decades to show up from scientific theory to manufacturing to consumer hands to show up. In the inmediate future ,next decade , things
are heading to cloud computing , that is to use the internet , like a virtual computer online ,that feeds data in real time , from a super computer in some place. but for such kind of tech you need super high speed internet.. this is why CHINA 5G technology is interesting. it can provide a virtual super fast computer to everyone for free as long they have an smartphone.

The last time there was any substantial investment in the chip industry in Russia was when Medvedev was in power.
Russia basically bought some outdated production equipment from the West. Processes were licensed. This was a relatively minor investment and in my opinion was good enough.
Europe's semiconductor industry is barely in better shape than the one in Russia having vanished into near non-existence.
Only companies which sell to markets with many hundreds of millions of customers can invest in the huge production facilities in use today. Everyone else cannot compete in the open market.

Even Japan collapsed from being in a position of leadership in the late 1980s, early 1990s, to being a has been today.
Japan still has a tools industry which is a remnant of that semi industry but it has been outcompeted by the rest of the West for so long that its days are, I think, numbered.
As Samsung clawed the consumer electronics business from Sony, Japan lost the scale to compete.
South Korea and Taiwan are close to the assembly lines for the consumer electronics devices and have year around port access. Russia does not have these conditions.

Given these harsh realities I think you expect too much from the Russian government.

A sapphire crystal ban would not be the panacea you hear about. Sapphire crystal wafers are only used in some limited niche applications. AFAIK Intel and AMD do not use it. Neither does TSMC. But those niche applications are precisely those that matter like military grade, radiation hardened chips, and tiny GPS radio chips the US holds a near monopoly over. So a ban would hit the West semi industry hard you can bet on it.

The neon gases export ban would be more difficult for companies like TSMC as they would be forced to get their neon from China which I think is the only other major supplier right now.
China's own semiconductor industry would be in direct competition with them and this would squeeze them.

Japan and Taiwan are heavily dependent on LNG for their energy generation. So if Russia wants to hit them directly this is a good way to do it. Japan US and South Korean nickel imports should be hit too. This would hurt their production of militarily relevant metal alloys and lithium-cobalt-nickel batteries.

Russia should also ban exports of helium gas for Western companies. Helium is critical for modern hard disk drive manufacture. SpaceX's rockets use it and so do other Western launchers. It is also used in magnetic resonance medical equipment market which is mostly controlled by European and US companies like Siemens and GE. Plus other things which use superconductors.

anyone can comment on this?

how real is that US can "shut down" chinese own semiconductor industry ?

China’s Largest Chipmaker Will Be ‘Shut Down’ If It Sells To Russia Says U.S. Official

https://wccftech.com/chinas-largest-chipmaker-will-be-shut-down-if-it-sells-to-russia-says-u-s-official/

The west claims it can shutdown china SMIC fab plant in shangai..  but tha'ts seems wishful thinking.


Either way i hope china blocks Russia too from their semiconductors , to block the parasite
president so he stop being lazy and finally take the decision to develop the high tech industry for real.

Given these harsh realities I think you expect too much from the Russian government.

Not really ,because the semiconductor industry have more than 3 decades already and russia wasted the last 22 years focused exclusively in pipelines business to the west, because president putin did not took seriously the high tech semiconductors industry. South korea.. and tiny small nation is galaxies away superior than russia in semiconductors ,and their budget is much inferior than the russian one.. even singapur a tiny island is more advanced than russia in electronics..
simply there is no excuse for Russia , other than his president was totally out of touch with reality ,
and did not saw the danger he was ,for russia extreme dependence on the west for high electronics.

by the way japan and south korea buy liquid gas from australia. So russia have nothing to offer them, and this is a major problem. when your country is not important at all for any developed country. By Russia going the most lazy way of building its economy ,it allowed the west to consolidate their power in high tech industry.

They can. SMIC uses Western tools. They had to stop supplying chips to Huawei because of this. You heard that right SMIC cannot manufacture 14nm chips for Huawei smartphones.
Smartphones. Not military chips. Which Huawei never manufactured in the first place.

SMIC itself is in a US government entity list where all of their Western imports have to be screened and approved first and they cannot buy any modern EUV tools.
This means SMIC cannot manufacture at 5nm and under that. If they do get 7nm to work at all it will not be cost competitive against TSMC. They are stuck at 10nm.

The Chinese government has paid SMIC since to build a new fab in Beijing with tools and materials 100% independent of the West. But you cannot build a new fab just like that.
It is still under construction. Expected to be complete in 2024.

I told you to forget anything better than 28nm. Even Chinese 28nm production chain is not 100% independent yet. They are working on it.
The Chinese are ramping up production of 28nm class photoresist and mask material to cut imports of either of those consumables for example.
They in the process of testing Chinese tools for 28nm mass production in a test line at a Shanghai government lab.

The US will also do the same thing to Russian semiconductor fabs. They will cut their supply of parts and materials.
But like I said the lucky thing is equipment is so old it will be much easier either to smuggle or to make replacement products than for newer equipment.

Russia is mostly stuck at 65nm node. China has 100% native production chain for 65nm node. Both tools and materials. So there is no issue.
The Chinese company CETC is in the US blacklist and they make 65nm tools. They are a Chinese state electronics company and they don't give a shit about US sanctions.
CETC is part of Chinese MIC proper. Not a commercial company like SMIC. CETC makes AESA radars for Chinese air force for example.

i read somewhere that china use different methods than western EUV tools for its semiconductors , designed by China that achieve comparable results , so more info will be need on this , to really understand . china response will be interesting.  China investments in the semiconductors industry
is superior than US and taiwain combined.. is on the trillions $USD.. and the head of taiwan semiconductor lab ,told that chinese progress is closing the gap more faster than they expected.. just 3 years behind in manufacture their own high end 5nm chips and 5 years in production of lithographics machines.  he told in 2021 ,that in just 5 years china will catchup taiwan.  

There are alternatives to electronic ultra violate tools that taiwan and west use, and china is using it , but forgot the name. the next video is not the video . but shows china advances in locally done euv tools.

Unafraid of U.S blockade, China successfully developed lithography machine, chip independent hope.



The days of west monopoly are soon to be over , in near future.

China is way behind but they have a huge market which can easily do the hundred billion dollar investments (that is right hundreds of billions) to get this working in large scale.
The Chinese can also afford the continuous and escalating investments to keep up with this technological arms race.

Compare this with investing ten billions in a pipeline which can operate for 20-30 years. In the lithography space you spend twenty billion on a factory which will be obsolete in 8 years.
Then you have the factories for tools, materials, etc. It is just pointless. The Japanese had to give up for the most part and they have similar population base to Russia.
Medium and small countries like South Korea and Taiwan can get on this game because they can buy and sell worldwide.
Had those countries been put into an Iran like trade sanctions hole they would be dead in the chip space.

Russia should not bother getting into this hole. Which is why I suggested the e-beam litho machine. It will be more than enough for the MIC.
But this is a long term project since there is nothing in the market like it. This is a breakthrough asymmetric technology.

As I said, Russia has:

250nm
180nm
90nm
65nm
45nm

Now, I know that their development rates are very low compared to the big companies.  Obvious as to why as most of their production was specialized for military production chips.  Reality is that most chips on 180nm is fine for standard chips like guidance systems on missiles, control systems like industrial printers and machinery (company I worked for used Pentium 3's for every industrial equipment including advanced CNC machines), only recently upgraded to a ARM chips because couldn't source the parts anymore for the older chips.

For mass production, 90nm is best they can do for now.  65nm and 45 dunno (I read it was rather very small production batches). But as I linked before (please stop spreading fud guys and actually read my links), Baikal electronics is working with GS Nanotech for production of their processors.  So if this is the case, then I assume it is because GS Nanotech somehow managed to get production of their chips up to a standard that a company like Baikal Electronics (Subsidiary of HPC company T-Platforms), found it sufficient enough to go with them for localized production.

I am wondering if it would be possible to try and reverse engineer the lithography equipment and start producing their own equipment and ultimately improve upon it in order to go smaller?  In mean time, as KVS said, do chiplet designs.

Now I am also interested in which Perm company is making GaN modules that could possibly be used.  In this case, only company I know of that makes GaN modules and GaAS modules is NPP Istok which is not in Perm.  These are used for Russian airborn and ship based AESA.

Source:

http://gsnanotech.ru/news/gs-group-i-baykal-elektroniks-nachinayut-sotrudnichestvo-v-oblasti-razvitiya-otechestvennoy-elektron/
http://gsnanotech.ru/news/gs-nanotech-voshel-v-reestr-rossiyskikh-dizayn-tsentrov-mikroelektroniki/

Honestly, between Mikron and GS Nanotech, the government should offer a big amount. Maybe get MCST and Baikal involved, to work with either one of these in a form of competition for development of Lithography tech to advance the development at least down to 28nm or lower and for further chip development.

Once the money starts to trickle in, there will be incentives. Massive opportunity now since these companies will be even more relied upon after these events.

The above article goes off the deep end with talk of 5 nm and smaller process resolutions. The 7 nm process is physically 10 nm.
The "7 nm" FinFet is a marketing concoction. It does not exist. People think that materials are infinitely divisible. BS. Below
3 nm you enter the molecular cluster regime. You cannot make regular ICs at this resolution. Perhaps you can make quantum
computer components using semi-contiguous molecular cluster aggregates but just stop with the crap about making standard
transistors. There are no meaningful P-N junctions and classical thermodynamics limit material to define different transistor domains.

A 28 nm process resolution would be good enough for Russia's short term needs. A 14 nm process would be ideal. TSMC can
make its "2 nm" components on Twitter.

kvs wrote:The above article goes off the deep end with talk of 5 nm and smaller process resolutions.   The 7 nm process is physically 10 nm.
The "7 nm" FinFet is a marketing concoction.   It does not exist.    People think that materials are infinitely divisible.   BS.  Below
3 nm you enter the molecular cluster regime.   You cannot make regular ICs at this resolution.   Perhaps you can make quantum
computer components using semi-contiguous molecular cluster aggregates but just stop with the crap about making standard
transistors.   There are no meaningful P-N junctions and classical thermodynamics limit material to define different transistor domains.

A 28 nm process resolution would be good enough for Russia's short term needs.   A 14 nm process would be ideal.    TSMC can
make its "2 nm" components on Twitter.

I know Mikron wanted to get 28nm tech. GS nanogroup kinda does its own thing and reports things after they got the equipment. I guess they make their own mem modules and make various SoC's

https://rostec.ru/news/rostekh-kaskad-i-mtusi-sozdayut-zashchishchennye-kriptoprotsessory/

Rostec, Kaskad and MTUCI create secure cryptoprocessors

Automatika Concern of Rostec State Corporation, Moscow Technical University of Communications and Informatics (MTUCI) and Kaskad Company are jointly developing a line of secure cryptoprocessors for the Internet of Things. Separate chips for data encryption will allow you to secure data and protect devices from unauthorized entry.

A cryptoprocessor is a separate system on a chip ,a "computer within a computer". It converts data in such a way that it can only be decrypted on the source computer, running the same software that was used initially. This ensures the maximum level of security and the impossibility of unauthorized access to information from outside. The presence of a cryptoprocessor eliminates the need for additional physical security features of the system.

Cryptographic protection of information is a necessity when implementing IoT systems in various industries. It blocks attackers from accessing IoT networks, protects critical data from interception, and eliminates unauthorized access to technological processes.

Cryptoprocessors are used in the production of intelligent sensors for the oil, mining and other industries. Also, the products can be used in sensors for controlling the engines of cars or aircraft, for creating critical infrastructure equipment.

"Experts and specialists from three organizations are accumulating competencies in order to develop a line of high-performance and energy-efficient cryptoprocessors. This equipment will be used to protect information in sensors to organize a secure channel to the network core. The relevant agreement between the parties has already been signed, "said Andrey Motorko, General Director of the Automatika Concern.

As part of the project, MTUCI engineers develop software and provide integration of cryptographic algorithms, Automatika Group certifies a family of cryptoprocessors, conducts case studies, promotes and sells products, Kaskad LLC develops topologies of cryptoprocessors, tests products and performs final configuration.

"MTUCI is one of the leaders in the field of applied research and development. The high level of results of scientific research and engineering developments at MTUCI is ensured by historically established and developing research teams – MTUCI scientific schools. The total number of employees involved in scientific research is about 300 people, including more than 20 doctors of sciences and more than 70 candidates of sciences. I am sure that combining the research potential of our university and leading developers and manufacturers of solutions in the field of the Internet of Things will allow us to bring new products to the market in the shortest possible time," said Yuri Leokhin, Vice – Rector for Research at MTUCI.

The development of the chip line is accompanied by experts in the field of information security. In the course of joint activities, specialists take into account and correct the shortcomings of cryptoprocessors already existing on the market, which allows us to maximize the potential market for using the developed chips.

When Russia realised it needed to completely overhaul its military, it focused on C4IRSTAR first, but also looked across the board at all the existing technologies and weapons and equipment and opened two main focuses... the first was to take existing designs and equipment and look at and try to eliminate all their faults and problems and issues. The second was to start work on a next generation of systems and equipment and weapons that didn't have the faults to begin with and would be better than anything the west currently had.

It took time and money and effort but they did it for a fraction of what the US and HATO were spending on their military over the same period.

Clearly they need to look at their core industries... food, energy, computers, transport and electronics... and start to do the same across the board.

Apply new technology to try to leap ahead, photonic computers and radar, hypersonic platforms, new technology super computers...

The west is broken... time to shine and make yourself an alternative to the rest of the world.

A lot of half wits going off about Russia being stuck at 60nm and cant make the smaller nodes necessary for machine learning.
And how even China is stuck on 14nm with terrible yields.

Are those smaller chips so critical for machine learning??

And what's Russia's progress on the machine learning research?

AlfaT8 wrote:A lot of half wits going off about Russia being stuck at 60nm and cant make the smaller nodes necessary for machine learning.
And how even China is stuck on 14nm with terrible yields.
Are those smaller chips so critical for machine learning??
And what's Russia's progress on the machine learning research?

This is kind of overblown to be honest.

The Chinese 14nm at SMIC has not had terrible yields for over a year already. Right now it should have similar yields to other players. SMIC also started production and made deliveries at 10nm like half a year ago. What the Chinese lack is the ability to ramp up production of 14nm because it depends on Western tools supply (European/US or Japanese). The supply and maintenance of which could be cut at any minute. So, rather wisely, the Chinese cut their plans for further expansion of this node and focused on lower resolution nodes which are more sanctions resistant.

The whole idea you will have everyone do machine learning is also kind of BS. Machine learning, at least for the next decade, will mostly happen in centralized data centers. You also do not need FP32 or even FP16 for a lot of loads like what is being used now. For the client nodes people will use other hardware architectures with much less transistors. These are both possible and being actively worked upon.

Russia has had hardware DSPs since the late 1990s which would be suitable for both AI machine learning and for active deployment in weapons systems. If they ever wanted to do something like that.

A lot of the discussion in the public toilet information space assumes Russia has not been investing in research and development of
its own lithography equipment. The hear about hand me down purchases (e.g. from AMD's long gone Dresden fab) and assume that
is all that Russia can do. Rosnano has been going at it for a long time and we have the the lack of Russian PR being treated as
total absence.

The notion that Putin's government would not have been aware of the strategic importance of Russian IC manufacturing since 1999
is delusional. Expect surprises in the short term on the IC manufacturing front. All of the excuses why outsourcing is cheaper and
more economically efficient have been destroyed. Tech asses will be kicked and resistant heads will roll.

I would not say there was no investment in semiconductors in Russia since.
You have the Crocus Nano MRAM memory fab. It uses 300mm wafers with 90/55nm process.
You also have those LED lighting fabs they made at Kaliningrad.
You have motherboard manufacturing plants.
More Elbrus and Baikal processors were designed.

Unfortunately a lot of the more modern chip designs the Russian industry made rely on use of external foundries in fabless model.
This made economic sense, given the volumes involved, but is now biting the industry back.

First thing the Russian semi industry should work on, I think, is make a common platform used in all existing fabs. This will require collaboration between the different corporations. You should be able to easily move a chip design between factories. Once the design is done you should be able to manufacture it without changes at any of those plants. The industry should also work on ways to smartly bunch up different chip designs on the same wafer so small production runs will be more cost effective.

Last but not least all relevant chip designs which were targeted at foreign production, use manufacturing nodes which can be done at existing plants, and are still relevant for military or industrial applications should be moved to these processes.

For example if the Elbrus-4S (65nm), MCST-R1000 (90nm), KOMDIV-64 series (65nm), NeuroMatrix (65nm), still don't have a 100% Russian production chain even after the ban on space grade semiconductors happened then they should get it. And it should have happened yesterday.

Baikal is probably pretty much screwed though. Without 28nm or better fab access there is little they can do. Their whole business relies on set top box and server chips. Their volume requirement is way too high for Russian industry to handle right now. Chinese Phytium has same issue. It is also kind of pointless to make a 65nm ARM chip design unless you want to go back to dumb phones. Well close to it. 65nm is basically the SoC of the iPhone 3GS. They also license chip IP from UK based companies. So they might be in deep shit. Although ARM China insulates the Chinese ARM chip designs Russia has no such scheme that I know of.

lancelot wrote:They also license chip IP from UK based companies. So they might be in deep shit. Although ARM China insulates the Chinese ARM chip designs Russia has no such scheme that I know of.

As per counter-sanctions, Russia will not respect western IP anymore, they may ever start building x86 processors if they wish.

lancelot wrote:I would not say there was no investment in semiconductors in Russia since.
You have the Crocus Nano MRAM memory fab. It uses 300mm wafers with 90/55nm process.
You also have those LED lighting fabs they made at Kaliningrad.
You have motherboard manufacturing plants.
More Elbrus and Baikal processors were designed.

Unfortunately a lot of the more modern chip designs the Russian industry made rely on use of external foundries in fabless model.
This made economic sense, given the volumes involved, but is now biting the industry back.

First thing the Russian semi industry should work on, I think, is make a common platform used in all existing fabs. This will require collaboration between the different corporations. You should be able to easily move a chip design between factories. Once the design is done you should be able to manufacture it without changes at any of those plants. The industry should also work on ways to smartly bunch up different chip designs on the same wafer so small production runs will be more cost effective.

Last but not least all relevant chip designs which were targeted at foreign production, use manufacturing nodes which can be done at existing plants, and are still relevant for military or industrial applications should be moved to these processes.

For example if the Elbrus-4S (65nm), MCST-R1000 (90nm), KOMDIV-64 series (65nm), NeuroMatrix (65nm), still don't have a 100% Russian production chain even after the ban on space grade semiconductors happened then they should get it. And it should have happened yesterday.

Baikal is probably pretty much screwed though. Without 28nm or better fab access there is little they can do. Their whole business relies on set top box and server chips. Their volume requirement is way too high for Russian industry to handle right now. Chinese Phytium has same issue. It is also kind of pointless to make a 65nm ARM chip design unless you want to go back to dumb phones. Well close to it. 65nm is basically the SoC of the iPhone 3GS. They also license chip IP from UK based companies. So they might be in deep shit. Although ARM China insulates the Chinese ARM chip designs Russia has no such scheme that I know of.

Edited down to be nice

Please read the thread first.

As I pointed out and you have now ignored 3 or 4 times, Baikal electronics is moving to GS nanogroup. As well, It is well known they produce both MIPS and ARM processors. Lastly, it's been mentioned that they don't care about IP from enemy states which UK which holds patent ownership of ARM is one of said states.

Do your research

Dima wrote:It would be good for Russia to join the Chinese and invest/buy their DUV lithography machines. EUV machines might need to wait. But if the Chinese are after it, they will develop it and make it available commercially as well.

SMEE (Shanghai Micro Electronics Equipment) have already successfully developed their 28nm DUV machines which will mature in few years time. With these they might be able to go for 7nm with multiple exposures.

Not an experts on these matters, hope others with much expertise in this field might be able to give a better understanding on these stuffs.


@kvs

Can you please help me understand, what is it that Russia, Russian chips and Russian architecture (like Elbrus architecture) that lacks and which need to be overcome or developed further to have a proper competitor to the likes of Intel & AMD?

The Elbrus is the only fully viable VLIW architecture presently available in the world.  It does not need adaptation and changes.  Unlike
Intel's hack EPIC (Itanium) Russia has developed optimal compiler and binary translator tools for the Elbrus architecture.    

Competition for x86 software requires an x86 CPU.   Any software where the source code is not available for recompilation implies
the need for a binary translator or emulator software.   This introduces a performance hit.   The Russian binary translators are very
good but there is no free lunch.   The "competition" is constrained by monopoly rackets.   To produce and x86 CPU you needed a
license from Intel.   Apparently, Intel is opening up the x86 instruction set but we will see.    Russia has no plans to produce x86
CPUs.   Maybe China can fill the gap since it does manufacture its own x86 CPUs.  

The Elbrus fully serves the purpose for HPC.   HPC software is all custom compiled.   For science and engineering the code is all custom
and nobody uses any closed commercial software.    So x86 is utterly irrelevant.   The only reason you see it propagate into HPC is
because there has been a shift to commodity components for scalar parallel supercomputers.   IBM makes superior CPUs (Power series)
but they are bloody expensive since they are low volume and also very big so suffer from yield limitations due to silicon wafer defects.  

The only issue right now for Russia is consumer x86 CPUs and the lack of high resolution process manufacturing for its own chips such
as the Elbrus.   I think that Russia can source AMD and Intel CPUs via gray channels.   It will have to work with China to produce 14 nm
Elbrus chips.   The Elbrus does not need "7" nm process for production.   It has less transistor density than AMD and Intel CPUs due to
its VLIW and compiler centric nature.

kvs wrote:

Dima wrote:It would be good for Russia to join the Chinese and invest/buy their DUV lithography machines. EUV machines might need to wait. But if the Chinese are after it, they will develop it and make it available commercially as well.

SMEE (Shanghai Micro Electronics Equipment) have already successfully developed their 28nm DUV machines which will mature in few years time. With these they might be able to go for 7nm with multiple exposures.

Not an experts on these matters, hope others with much expertise in this field might be able to give a better understanding on these stuffs.


@kvs

Can you please help me understand, what is it that Russia, Russian chips and Russian architecture (like Elbrus architecture) that lacks and which need to be overcome or developed further to have a proper competitor to the likes of Intel & AMD?

The Elbrus is the only fully viable VLIW architecture presently available in the world.  It does not need adaptation and changes.  Unlike
Intel's hack EPIC (Itanium) Russia has developed optimal compiler and binary translator tools for the Elbrus architecture.    

Competition for x86 software requires an x86 CPU.   Any software where the source code is not available for recompilation implies
the need for a binary translator or emulator software.   This introduces a performance hit.   The Russian binary translators are very
good but there is no free lunch.   The "competition" is constrained by monopoly rackets.   To produce and x86 CPU you needed a
license from Intel.   Apparently, Intel is opening up the x86 instruction set but we will see.    Russia has no plans to produce x86
CPUs.   Maybe China can fill the gap since it does manufacture its own x86 CPUs.  

The Elbrus fully serves the purpose for HPC.   HPC software is all custom compiled.   For science and engineering the code is all custom
and nobody uses and closed commercial software.    So x86 is utterly irrelevant.   The only reason you see it propagate into HPC is
because there has been a shift to commodity components for scalar parallel supercomputers.   IBM makes superior CPUs (Power series)
but they are bloody expensive since they are low volume and also very big so suffer from yield limitations due to silicon wafer defects.  

The only issue right now for Russia is consumer x86 CPUs and the lack of high resolution process manufacturing for its own chips such
as the Elbrus.   I think that Russia can source AMD and Intel CPUs via gray channels.   It will have to work with China to produce 14 nm
Elbrus chips.   The Elbrus does not need "7" nm process for production.   It has less transistor density than AMD and Intel CPUs due to
its VLIW and compiler centric nature.  

What about moving to 28nm and doing 3d chip and chiplet design?

The issue with 3D is it is new design work. I am all for different approaches but they require spin up time. The fastest
way to maintain continuity is to find a substitute for TSMC. Since the move to 28 nm has been discussed for years I assume
that it was not all just hot air. But moving to 28 nm means rolling back to previous Elbrus models.

kvs wrote:The issue with 3D is it is new design work.   I am all for different approaches but they require spin up time.    The fastest
way to maintain continuity is to find a substitute for TSMC.   Since the move to 28 nm has been discussed for years I assume
that it was not all just hot air.   But moving to 28 nm means rolling back to previous Elbrus models.  

I dont think it would be hard to believe that they may re-work 8c and 4c based Elbrus chips down to what they can produce or will produce locally. And future advanced Elbrus on lower lithography will be through China.