What's new

Russia wants to replace US computer chips with local processors

Russia Launches ARM Chip Effort To End X86 Dependence
June 25, 2014 by George Leopold
arm-system-logo-370x258.jpg



The Russian government is reportedly preparing to say “nyet” to American-made microprocessors, planning instead to replace Intel and AMD chips with a homegrown CPU based on an alternative architecture offered by ARM Holdings.

According to Russian media reports, Russia’s Ministry of Industry and Trade plans to replace the U.S.-made processors used in government computers with a domestic processor design codenamed “Baikal.” According to reports, the new 2 GHz CPU based on ARM’s 64-bit kernel Cortex-A57 design will run in both government PCs and servers; it will have eight cores, like other Cortex-A57 designs.

Baikal (also the name of the region around Lake Baikal, the world’s deepest freshwater lake) will be designed by the Russian supercomputer company T-Platforms. The company was blacklisted by the U.S. Commerce Department in March 2013 for alleged involvement in diverting technologies to military programs. T-Platformsannounced in January that its named had been removed the U.S. Export Restrictions Entity List, allowing it to “resume normal trade activities in compliance with all U.S. export control requirements.” Among its services is “compute system design,” according to the company’s Web site.

The first microprocessors based on the ARM designed will be designated “Baikal M” and “M/S,” the Russian news agency reported. No specific timeframe was given for when the processors would be ready and it is not clear who will actually manufacture the chips in the 28 nanometer design process that Baikal Electronics, the division of T-Platforms that is doing the design, has chosen for the first chip.

Beyond reports in state-controlled media, the Russian government has said little or nothing about the project, which was first reported by the Russian business daily Kommersant.

However, senior Russian officials have recently expressed concerns about using U.S.-made technology inside sensitive government systems. Referring to the need for a national card payment system, Russian Prime Minister Dmitry Medvedev recently told government officials: “For us it is a priority that the future system rely on Russian technology. This will ensure the independence of our system, expand opportunities for customers, and resolve issues related to security and protecting information from unauthorized access.”

The ITAR-TASS news agency reported that state-run companies are expected to pour “dozens of millions of dollars” into the Baikal project. Among the agencies reportedly bankrolling the effort are state military conglomerate Rostec along with technology giant Rosnano. The latter was formed to invest in nanotechnology.

The news agency said Russian state-run firms and government agencies purchase about 700,000 PCs each year worth $500 million. About 300,000 servers are purchased by the government annually at a cost of $800 million. The overall Russian market for microprocessors is estimated to be $3.5 billion.

Moscow’s plan for domestic microprocessor development likely comes in response to revelations of widespread eavesdropping on international communications by the U.S. National Security Agency. One fear is that U.S.-made components might contain “back doors” that would aid the NSA in spying on Russia.

Tensions between Russian and the United States mounted when NSA contractor and whistleblower Edward Snowden was given asylum by Moscow. Those tensions have increased with Russia’s annexation of Crimea from Ukraine.








June 23, 2014
Russia to Develop Home-Grown Chips
Tiffany Trader
russia-flag-200x149.jpg


Russian officials want to replace US microprocessors from Intel and AMD with locally-grown parts. According to recent reports from local news services, the Russian government is planning to create its own line of microprocessors using 64-bit ARM chips for use in governmental computers and servers.

The new microprocessors will be designed by Baikal Electronics, a unit of supercomputing vendor T-Platforms, with funding from state-owned defense giant Rostec and tech-focused state-run venture capital firm Rusnano. The Baikal nomenclature was inspired by the largest freshwater lake in the world, Lake Baikal, located in southern Siberia.

The project, worth tens of millions of dollars, is being managed by Russian’s Industry and Trade Ministry. In place of current chips from US companies, the Baikal line of processors will be built using ARM Holdings’ 64-bit Cortex-A57 design, which runs at 2 GHz. The first products, designated as Baikal M and M/S, will be manufactured using 28nm process technology. These 8-core variants are scheduled to debut early next year, with a more powerful 16-core server processor, based on 16-nm process technology, to follow in late 2016.

Russian news agency Kommersant noted that currently Russia does not have the capability to produce 28nm processors, and the project’s price tag, estimated at tens of millions of dollars, does not include the cost of production.

The new chips will power Linux-based personal computers and micro servers and will mainly be used by the defense industry and other government-aligned operations. According to ITAR-TASS, the Russian government purchases about 700,000 personal computers each year worth $500 million as well as 300,000 servers worth $800 million. The total market volume is estimated at 5 million devices worth $3.5 billion.

The fact that the chips are being targeted for government systems indicates the project is motivated by security concerns. However, if the processors are successful, their market share could spread to the Russian public sector, and even potentially outside Russia’s borders.




@rmi5


:rofl::rofl::rofl:

Russian government dumps Intel and AMD in favor of homemade processors

BY SHARIF SAKR @SHOTSHERIFF 2 DAYS AGO




Russia's policy on Western technology is clear: The country can live without it, especially if key issues like economic sanctions, NSA spying and GPS cooperation aren't resolved to its leader's satisfaction. It looks like this tough stance extends to US-designed computer chips too, as a Russian business newspaper is reporting that state departments and state-run companies will no longer purchase PCs built around Intel or AMD processors. Instead, starting in 2015, the government will order up to one million devices annually based on the "Baikal" processor, which is manufactured by a domestic company called T-Platforms. An interesting twist, however, is that the Baikal processor is actually based on an ARM (Cortex-A57) design, which means the East / West divorce isn't quite as complete as it might sound. It could also mean that many Russian bureaucrats won't get the chance to be a Mac or a PC: they'll have to use some sort of ARM-compatible, presumably Linux-basedoperating system instead.
 
Last edited:
power consumption of tianhe-2 is 24 mw.
you are probably right and the article in Wikipedia is wrong about the power consumption of titan which they made it 1000 time ,but this also apply to the titan power consumption and so the calculation remain the same.

both tianhe and Titan use processors for distributing the data between computing units and the real work hours there is the tesla units in the titan and Xeon Phi's in Tianhe both of them are computing units . the difference is that tianhe use twice as much processing units and provide 2 time the performance while using 4 time the energy and it's due to the fact that while Intel made Xeon Phi is based on x86 is easier to program it's not as efficient as Nvidia made Tesla Units that use Cuda architecture .

it's shown in the performance while theoretically tianhe-2 only have 33.86pf and titan has 27pf of performance , tianhe can achieve the peak theoretically performance while in linpack test titan only reached 17.5pf

Uhhm, not necessarily. Even if you look up the comments in your provided link, you would see that energy consumption is not much different. About the heating, ... what do you mean by maintenance costs? BTW, using multiple CPUs is only useful when your codes can be run on a multi-thread basis. ;) About the heating, and faults, just simply buy good fans!!! The temperature on a chip is much higher than what you think and it does not cause failure, unless the heat gets accumulated for a long time, and temperature gets very very high. simple solution >> buy big good fans!!!
the time is used is 1/4th and no the fan only help but don't change the amount of the heat produced and show me a problem that need HPC and is not threaded
 
the time is used is 1/4th and no the fan only help but don't change the amount of the heat produced and show me a problem that need HPC and is not threaded
:what:
I believe you did not get what I said. some softwares, simulators, ... can be run on a multi-thread basis, and some of them cannot. Anyway, what I said was a technical term. Never mind if you did not get it.
 
Isn't China producing its own chips? Their latest supercomputer is powered by indigenous processors, is it not? Intel/AMD duopoly on processing must be broken.

I remember a time when nobody in his/her right mind could have imagined that NOKIA/BlackBerry would become a distant 4th/5th in the mobile industry but it happened. How many could have imagined that SONY/Panasonic would be playing 2nd fiddle to a new company SAMSUNG? If the Koreans can do it, why not the Chines/Russian??
 
search for early leaks of Denver chips , its not a vanilla core like what Russia plan to do but it do better than what you said and do it by consuming just 6-7 w of power while having a better GPU than what Intel sandy bridge could dream of providing and also being on a 28nm node
Yes I know about he Denver project that would be implemented on second gen maxwell. Those ARM64 cores along with the maxwell GPU would thrash the x64 CPU only computing but some folks here believe that GPU computing would never be mature.
 
:what:
I believe you did not get what I said. some softwares, simulators, ... can be run on a multi-thread basis, and some of them cannot. Anyway, what I said was a technical term. Never mind if you did not get it.
and single threaded operation have nothing to do with this discussion by the way a simple web browsing can heavily benefit a help from GPU , a simple office program also benefit a lot from using threaded algorithm , single threaded problems are becoming a things of past at a dangerous rate


what is this it's ages since PC and Macs used different CPU .
 
you are probably right and the article in Wikipedia is wrong about the power consumption of titan which they made it 1000 time ,but this also apply to the titan power consumption and so the calculation remain the same.

both tianhe and Titan use processors for distributing the data between computing units and the real work hours there is the tesla units in the titan and Xeon Phi's in Tianhe both of them are computing units . the difference is that tianhe use twice as much processing units and provide 2 time the performance while using 4 time the energy and it's due to the fact that while Intel made Xeon Phi is based on x86 is easier to program it's not as efficient as Nvidia made Tesla Units that use Cuda architecture .

it's shown in the performance while theoretically tianhe-2 only have 33.86pf and titan has 27pf of performance , tianhe can achieve the peak theoretically performance while in linpack test titan only reached 17.5pf
True but what mandrom said was titan did cost 900 million which used perofessional k20x cards and Tianhe2 did cost 390 millions which used Xeon Phi co-processor units(x64 cores) and the latter have double performance while costs less than half hence CPU>GPGPU.
My reason for being that much costly is expensive manufacturing in USA and high wages of associated engineers.

@vostok
1.)Would these chips be available outside Russia with full support?
2.) Are you guys also making GPUs? (To end graphics monopoly)
 
True but what mandrom said was titan did cost 900 million which used perofessional k20x cards and Tianhe2 did cost 390 millions which used Xeon Phi co-processor units(x64 cores) and the latter have double performance while costs less than half hence CPU>GPGPU.
My reason for being that much costly is expensive manufacturing in USA and high wages of associated engineers.

@vostok
1.)Would these chips be available outside Russia with full support?
2.) Are you guys also making GPUs? (To end graphics monopoly)
Unfortunately, I have no information to give you accurate answers. All I know they will deliver those computers to government and military structures. Maybe in a few years they will appear on the market for ordinary users.
About the development of the GPU - I know nothing.
 
How many could have imagined that SONY/Panasonic would be playing 2nd fiddle to a new company SAMSUNG? If the Koreans can do it, why not the Chines/Russian??

Actually Samsung has been around for a long time (longer than Sony). They were probably behind Hitachi on the popularity list. I think they took advantage of the Japanese bubble burst to jump ahead by moving into chip making,cellphones, and being more aggressive in their exporting of home appliances (also add LG to that). Hmm...Hitachi seems to have fallen off the map.
 
Last edited:
True but what mandrom said was titan did cost 900 million which used perofessional k20x cards and Tianhe2 did cost 390 millions which used Xeon Phi co-processor units(x64 cores) and the latter have double performance while costs less than half hence CPU>GPGPU.
My reason for being that much costly is expensive manufacturing in USA and high wages of associated engineers.

@vostok
1.)Would these chips be available outside Russia with full support?
2.) Are you guys also making GPUs? (To end graphics monopoly)
you see Xeon Phi is not a processor , ts a coprocessor based on X86 , and about twice performance well it has twice processing unit so it's natural that it has more performance.
by the way if you compare a single tesla unit with Xeon Phi unit you get something like this in real world problems not heavily optimized benchmark like linpack which is used to rank supercomputers .
Benchmarks: Intel Xeon Phi vs. NVIDIA Tesla GPU


and about the being cheaper , if you consider 4 time being more power hungry and the maintenance bill , it's not that much more cheaper. and as you said it's cheaper because of the cheaper Chinese work horse and other component like cooling systems not the chips itself
 
Any country trying to break free from that repulsive Yankee regime has my approval. Well done Russia, hope you succeed. I think anyone supporting the Yankees are disgusting pigs.
 
Russia has launched the production of 8-core processors "Elbrus"
349c474260b901d88c2b80f5d20d65ab.jpg

The Russian company MCST has announced about launch of a pilot batch processor "Elbrus-8S". As noted in the company, the chip is fully Russian development. On the basis of new products is planned to collect servers, workstations and computers for high performance computing.

"Elbrus-8S" consists of 8 cores, 64-bit architecture has a cache memory of the second level of 4 MB and third level 16 MB, technology of production - 28 nanometers, the operating frequency of the chip is 1.3 GHz processing power - 250 gigaflops.

As previously reported Onliner.by Russia is ready to spend tens of millions of dollars to develop processor "Baikal", which is to replace the American microchip Intel and AMD.
В России стартовало производство 8-ядерных процессоров «Эльбрус» - Технологии onliner.by
 
Good luck to them.

I think Russia has has considered security aspect too.
 

Back
Top Bottom