Importance of indigenization of defense industry.

[below_freezing]

Loongson aka. LongXin/Godson processor is around the level of 2006-2007 Pentium chips. The upcoming Loongson 3B 8-core processor is slated for release this year and is expected to be comparable to state of the art CPUs of 2008-2009. Loongson 3B is planned as the CPU for China's follow-on supercomputer to Tianhe-IA (current world's fastest supercomputer using Intel CPUs+Nvidia GPUs) and is expected to dispense of the CPU+GPU design in favor of only the Loongson 3B processor in a 100% indigenous supercomputer.

The Loongson 3C 16-core 28nm processor is expected 2nd half of 2011 or 2012 and is expected to be comparable to state of the art CPUs of 2010-2011. Currently, the best China fabs can only produce 45nm semiconductor parts so China is almost assuredly outsourcing production of this part, probably to STM Microelectronics, but there is probably minor consideration of sourcing through TSMC in Taiwan.



China isn't anywhere near 32nm fab tech. It's best parts are made at SMIC at 45nm and that's not even for logic circuits. There are international technology sanctions against China (Wassenaar Arrangement) across the board including state of the art fab tech that restricts it to getting older generation semiconductor manufacturing equipment. The only reason China is getting any fab equipment better than 65nm is because it's own research & development is able to research and produce this themselves. So, the Wassenaar Arrangement as a voluntary international technology export restrictions cartel allows such technology transfer once China has clearly ALREADY ACHIEVED the ability to produce such prior restricted technology themselves anyways.

I was aware that there were severe sanctions against china in semiconductor technology. However you said that we can manufacture 45 nm; that implies that we can manufacture photolithography equipment. i haven't found any on sale though. do you know where to buy a china made photolithography chamber?

 

[tanlixiang28776]

Tianhe-1A supercomputer used American Intel CPUs and Nvidia GPUs with Chinese interconnect.

Along with 2048 NUDT processors

 

[tanlixiang28776]

I was aware that there were severe sanctions against china in semiconductor technology. However you said that we can manufacture 45 nm; that implies that we can manufacture photolithography equipment. i haven't found any on sale though. do you know where to buy a china made photolithography chamber?

It's not something private citizen's often ask to buy I'm guessing.

 

[VelocuR]

UK? Germany? Japan?

US borrowed from Germany WWII to re-invent or improve it. For instance, F-117 was the idea of Soviet Union. US has trillions dollars ! 700 Billion Defence Budget compared to the rest.

 

[below_freezing]

It's not something private citizen's often ask to buy I'm guessing.

You can even buy heavy machine guns on alibaba, but not photolithography machines...

 

[tanlixiang28776]

You can even buy heavy machine guns on alibaba, but not photolithography machines...

Please help us understand what photolithography machines imply about a countries defense industries.Is it something truly difficult to make?

 

[below_freezing]

Please help us understand what photolithography machines imply about a countries defense industries.Is it something truly difficult to make?

photolithography is the critical step in creating IC devices. this is how the chip is physically manufactured. If you can't design a chip but can make chips, it's fine, just steal other people's designs. but if you don't have this equipment you can't make a chip even if you can design it.

 

[tanlixiang28776]

photolithography is the critical step in creating IC devices. this is how the chip is physically manufactured. If you can't design a chip but can make chips, it's fine, just steal other people's designs. but if you don't have this equipment you can't make a chip even if you can design it.

That is certainly a problem. So does China have to buy equipment from abroad to make all their chips or what

 

[CardSharp]

photolithography is the critical step in creating IC devices. this is how the chip is physically manufactured. If you can't design a chip but can make chips, it's fine, just steal other people's designs. but if you don't have this equipment you can't make a chip even if you can design it.

My roommate was an electrical engineer and when I asked him about this, he said it was about doping and photo-exposing. So where is the technological bottleneck?

 

[below_freezing]

My roommate was an electrical engineer and when I asked him about this, he said it was about doping and photo-exposing. So where is the technological bottleneck?

I'm not too sure about the specifics of IC manufacture since I'm a chemist but you should learn an introduction in solid state chemistry. the technological bottleneck is probably we can't make the machine. why can't we make the machine? i don't know. controls or light source is the likely problem.

the operation of the photolithography machine, hopefully i said it right. first you have a perfect crystal of silicon. this is the substrate. then using chemical vapor deposition you add layers atom by atom on top of it (atoms diffuse into the silicon, maybe atoms such as arsenic or gallium to create PN junctions; this is doping). then you cover a wafer with a sort of polymer that acts as a photoresist. it must be perfectly smooth. You take a photomask, which is just an opaque piece of rock with holes in it in the shape of the circuit you want to make. shine a UV laser through it and hit the polymer on top of the doped wafer. photochemical reactions occur that make that specific exposed part more reactive to certain reagents, maybe let's say acid. then pour acid over the photoresist, and the exposed parts selectively dissolve away and carry a controlled amount of substrate with them. Now you have a tiny printed circuit on a nanometer scale thickness. Repeat the process for all relevant levels of substrate and you can have an extremely complex circuit.

 

[marshall]

I was aware that there were severe sanctions against china in semiconductor technology. However you said that we can manufacture 45 nm; that implies that we can manufacture photolithography equipment. i haven't found any on sale though. do you know where to buy a china made photolithography chamber?

China can manufacture 45nm scale semiconductors but the equipment they use is a mixture of domestic and imported equipment, mostly imported.

There are 2 distinctions here, one being semiconductor manufacturing equipment, the other production processes for various semiconductor generations, albeit not bleeding edge. China has mastery of semiconductor manufacturing processes but its own semiconductor manufacturing equipment industry is lagging due to commercial disadvantages. I liken China's situation with that of semiconductor manufacturers like Intel and TSMC. They both have technology to produce 18nm and 20nm scale semiconductors but they are not considered semiconductor equipment manufacturers even though the process technology they researched and developed is significantly more complicated than the closely related R&D of semiconductor manufacturing equipment.

 

[CardSharp]

I'm not too sure about the specifics of IC manufacture since I'm a chemist but you should learn an introduction in solid state chemistry. the technological bottleneck is probably we can't make the machine. why can't we make the machine? i don't know. controls or light source is the likely problem.

the operation of the photolithography machine, hopefully i said it right. first you have a perfect crystal of silicon. this is the substrate. then using chemical vapor deposition you add layers atom by atom on top of it (atoms diffuse into the silicon, maybe atoms such as arsenic or gallium to create PN junctions; this is doping). then you cover a wafer with a sort of polymer that acts as a photoresist. it must be perfectly smooth. You take a photomask, which is just an opaque piece of rock with holes in it in the shape of the circuit you want to make. shine a UV laser through it and hit the polymer on top of the doped wafer. photochemical reactions occur that make that specific exposed part more reactive to certain reagents, maybe let's say acid. then pour acid over the photoresist, and the exposed parts selectively dissolve away and carry a controlled amount of substrate with them. Now you have a tiny printed circuit on a nanometer scale thickness. Repeat the process for all relevant levels of substrate and you can have an extremely complex circuit.

Yep that's exactly what he told me. Almost exactly.

 

[Akasa]

The only way innovation will succeed in the arms industry is if the government loosens control and boosts private companies. With many companies working en masse, innovation will leap forward.

For example, the LFC-16 and CY-1 was privately-funded

 

[Developereo]

the operation of the photolithography machine

Isn't it approaching the twilight of this technology, though?
People are already looking at the next technologies to replace photolithography.

http://www.lsi.usp.br/~acseabra/pos/5838_files/TheLithoLimits.pdf

Lithography technology has been one of the key enablers and
drivers for the semiconductor industry for the past several decades.
Improvements in lithography are responsible for roughly half of the
improvement in cost per function in integrated circuit (IC) technology.
The underlying reason for the driving force in semiconductor
technology has been the ability to keep the cost for printing
a silicon wafer roughly constant while dramatically increasing the
number of transistors that can be printed per chip. ICs have always
been printed optically with improvements in lens and imaging
material technology along with decreases in wavelength used fueling
the steady improvement of lithography technology. The end
of optical lithography technology has been predicted by many and
for many years. Many technologies have been proposed and developed
to improve on the performance of optical lithography, but so
far none has succeeded. This has been true largely because it has
always been more economical to push incremental improvements
in the existing optical technology rather than displace it with a new
one. At some point in time, the costs for pushing optical lithography
technology beyond previously conceived limits may exceed the cost
of introducing new technologies. In this paper, I examine the limits
of lithography and possible future technologies from both a technical
and economic point of view.

This paper is already ten years old. There's probably a lot more reserach by now.

 

[below_freezing]

Isn't it approaching the twilight of this technology, though?
People are already looking at the next technologies to replace photolithography.

http://www.lsi.usp.br/~acseabra/pos/5838_files/TheLithoLimits.pdf



This paper is already ten years old. There's probably a lot more reserach by now.

With better light sources lithography should be able to arbitrarily etch smaller features. The only question is how do we make those light sources. I'm not an expert on these processes though.