China wants to become a semiconductor superpower, budgeting $100B to achieve it
economist.com
economist.com
http://es.slideshare.net/ortegaalfredo/deep-submicronbackdoo...
This example is a memory inspector and poker. On a side note, I think using a LEON MCU would be smart for this kind of demonstration, the code is free and it's synthetisable in a lot of targets.
[1]: https://ece.umass.edu/news/paper-stealthy-trojans-attracting...
Any such thing as circuit decompilers? In my imagination, they look like as huge boxes with tunneling electron microscopes inside, where you input a chip and it outputs source code.
I think even with modern computer vision techniques, it would be very very difficult to do this kind of thing automatically. If you were a big state with cheap labour, you could probably hire a load of people to do nothing but map traces all day.
That's super cool. I wanted to show you this too if you hadn't seen it already:
http://zeptobars.ru/en/read/KR580VM80A-intel-i8080-verilog-r...
Edit- oh, and dropping code into memory for the CPU to run is not unpossible either.
This +1. We are supposed to be intelligent adults but this behaviour is more like schoolyard bullies that behave one way and then cry foul when another does the same.
Are they still using VHDL?
I wonder what hardware languages are being used by Intel and ARM.
Systemverilog is the de facto standard in verification, but VHDL with its strong typing remains a favorite for writing synthesizable code.
As for Intel and ARM, both are primarily Verilog shops, but I know ARM provides both Verilog and VHDL models for the licensed components.
As far as why Intel, ARM, and everyone else mostly sticks to these languages is similar reason to why C is still used... it is because it works, and works good enough. After 20 years of just having Verilog, enough people agreed that it needed a update, and SystemVerilog developed. Another big reason is the fact that Intel, ARM, etc. all rely heavily on scripts and design flows that are common between pretty much every semiconductor company... all of the EDA vendors have united around these standards, and in a world where interoperability is required and very little innovation happens on the backend tools, people stick to what works.
You usually see VHDL in .eu and some defense contractors, and Verilog everywhere else. VHDL feels similar to Ada, and Verilog is more similar to C.
An interesting history from Intel's perspective is available here: http://webee.technion.ac.il/people/kolodny/ftp/IntelCADPaper...
China became an economic superpower because it had a huge labor force to offer cheaply and provided an environment in which international corporations could exploit (not necessarily with a negative confutation but rather the direct dictionary definition of the word) it easily.
They are coming to the point in which they need to import a very large amount of the components which is ending up in the electronics which are being made in China even basic stuff (e.g. high quality capacitors are still primary made in Japan, same goes for thin film resistors and other high precision electronics).
With electronics becoming nothing more than system on chip China needs to be able to compete locally if nothing than at least to offer cheaper local alternatives (A Chinese MediaTek for example), if they'll fail they risk at severely cutting the profitability and hence the income from Chinese manufacturing and risking quite a bit of it being actually shipped overseas especially when your production can be highly automated and scaled up without the need for a large and cheap labor force.
Another lucrative aspect that China has been offering is it's "design and manufacturing engineering" expertise you could go with a rough idea and a sketch to China and get 50 engineers all of who are experts in various fields of manufacturing from injection molding to electronic assembly and get your product out of the way.
To maintain this capability China needs to be able to design bespoke silicon just as easily as they can design a bespoke plastic shell which will be injection molded and then ultrasonically welded.
For some reason - this really made it sink in...
The cold war WRT electronics in the next few years is going to be very interesting. The dystopian cyberpunk landscape we read about in the 80s was utopian vs the reality of what we will see in the future.
my guess is it will pressure companies who currently hold on to their secrets without innovating (because they are the incumbent) to start innovating. If your competitors are on an even playing field with you, the only way to survive is to innovate.
https://en.wikipedia.org/wiki/Semiconductor_Chip_Protection_...
Edit: The context for the Chip Protection Act is actually in Intel's primary wikipedia page:
"During the late 1980s and 1990s (after this law was passed), Intel also sued companies that tried to develop competitor chips to the 80386 CPU.[32] The lawsuits were noted to significantly burden the competition with legal bills, even if Intel lost the suits.[32] Antitrust allegations had been simmering since the early 1990s and had been the cause of one lawsuit against Intel in 1991. In 2004 and 2005, AMD brought further claims against Intel related to unfair competition."
Of course, now the effect applies to those two where people wanting iOS were stuck with ObjectiveC and ARM while people targeting ChromeOS had to support its ecosystem/model. Legacy effect just shifts here. Truly portable and self-contained software is rare these days. Mostly only OSS stuff.
Also the Chinese might catch up faster than people think. Two small examples:- Xiaomi's MIUI has really been one of the most innovative android ROMS out there balancing usability and functionality. Another one, we use multitouch IR-based components. Right now the top 2 most innovative and competitive providers are both founded by Chinese, although one is headquartered in the US.
Basically, the whole "Chinese know only how to copy" argument is an excuse to be complacent, made by people thinking their culture or genes or whatever will protect them.
Then when they have parity and we have real competition, it will accelerate progress.
Maybe, but by that time Intel will be on 7nm technology. Intel isn't holding still, they are improving technology rapidly.
btw this year Intel switched to selling radio modules below cost, so you can expect china forcing them to another deal in a year or two.
Sure, you can copy the designs, but will you have the knowledge to modify them? Just playing catch up all the time has a major cost.
Think about code. Sure, people copy designs and code, but it's still hard. And even if you have problem X, and Intel doesn't, it could be for many reasons apart from transistor Y they have in the same general vicinity that you don't.
In a way, the blatant copying is good because it means that the Chinese industry has to go through less iterations to get up to speed. Hopefully the end result is a healthy environment with multiple players trying to out-innovate each other.
There was plenty of industrial espionage, but the Tu-144 is no more a copy than the DC-8 is a copy of the 707.
[0] no the F-22 didn't invent supercruise!
According to that Wikipedia article, the Russian plane was built and flown before the Concorde.
Intel's advantage isn't their design (which was always generally one or two generations behind the best). Consequently, there is no point is reverse engineering it.
Intel's primary advantage is their manufacturing has always been stellar.
Whether it is a good thing or not, hard to say. If it succeeds, China will dominate the market, at least the chips will be become cheaper. But as asian tradition goes, those investments may never find its way back in face of next tech revolution, just like how Japan dominate the analogue electronics but fail to keep up with digital ones.
There's one every season.
I think that money would be waste. While money is important human potential is the far more important thing. If we look closely a lot of governments,people claim that they will build a world class university which will soon be among top 10. Rarely anyone succeeds and lack of money is not the reason.
Same is true for cities. A lot of governments and countries have tried to build next Silicon Valley or next NY but none has succeeded. I do not doubt China's potential to create amazing things, but this sort of plan probably wont work.
Two detailed articles explaining this:
I don't know any brand off the top of my head from China that is top-quality. Most are blatent, cheap, ripoffs of better brands.
Also, this is only an example of copying relatively primitive designs, not expensive R&D.
They spend only enough to get ahead of their competitors, which is why I said it's a completely different model. Progress happens more incrementally.
I've been to China many times over the years for business and I haven't seen anything revolutionary.
China's lax IP laws will only get them so far. If they want to actually become a real tech. center, they need to have protections in place for inventors/companies (and actually enforce them).
The other problem is government corruption, but this is for another discussion entirely.
If China tried to knock off major chip designs Intel or AMD, they could be instantly found out and banned from import to major economies. And that would include any products that used those chips. So no major manufacturer would be interested in it.
China would almost definitely use some industrial espionage and trade secret misappropriation. But they couldn't just do a knock off version. But its not like other competitors in this space haven't been doing the same thing.
Plus, you can't fully reverse engineer semiconductor devices. A lot of steps in the fabrication process don't leave any clues.
Finally, China is starting to develop a robust IP system in their country. The number of Chinese patents is skyrocketing. As they try to transition from dumb factory of the world to a leading innovation center, they'll want IP protection.
It means if a S Korean semiconductor engineer joins a Chinese firm in China, he's guaranteed to be paid 7 times of his current salary for 3 years.
So let's say you make $10,000 at current job in S Korea, you will be making $70,000 a year, for 3 years. That's supposedly 'guaranteed'.
I may be off with the formula. It may have been 1:3:9. But I am sure it was 1:3:x, so pretty sure that it was supposed to be for 3 years.
A few dozen engineers did accept the offer so far. In the story I heard some S Korean HR person commenting such poaching went on recently for Display engineers who were recruited away from S Korea to Chinese firms. But he claimed none achieved much success, which I took as their career in the industry didn't last longer or they didn't make the money they expected to make.
China also made massive investment in commercial airliner, resulting in Comac C919. I read however that the C919 mostly had US/EU components though.
The reason for 3 years per contract? Seems to be a sweet spot. After 2 contract renewals an employee is 'permanent' and essentially unfireable. 5 years per contract seems a bit long and a big buy-in by the employer, and 1 year way too short to expect someone to have the level of assurance and employer buy-in.
So, the 7/9 sounds realistic because of the 3. But I honestly don't expect much to get done, even in 3 years. Someone coming from outside a firmly entrenched organisation has to be a management master to achieve anything.
Such a setup could make for a HW equivalent of what's happening in embedded with Arduino's etc. A slower pace due to costs but still better than now. Plus, an existing SOC platform at 45nm and 28nm would be great for cloud datacenters that are investigating custom hardware for accelerators and cost reduction. We'd see shit cranking out constantly for that market.
So, it's now up to China. They can have that blueprint of mine for free. I'd also consult on the specifics since nobody here is doing it outside a few academics and smaller firms. Gotta create the end result somehow.
Note to Chinese involved reading: Also, do a FPGA at 28nm and 28SLP with open bitstream and free tooling at low cost. Make ASIC conversions easy with paid, Chinese I.P. as an upsell. Probably sell the heck out of those FPGA's and academics will do most of R&D on their EDA just for fun. Acquire Achronix while you're at it to do the same with their asynchronous stuff. Acquire eASIC and/or Triad for RAD hardware at a discount for Chinese firms. Same advice I give here should help there.
Open-source FPGA http://www.eecs.berkeley.edu/Pubs/TechRpts/2014/EECS-2014-43...
Open-source, bitstream generation for Xilinx http://drop.isi.edu/sites/default/files/users/nsteiner/soni-...
Open-source tooling for lattice http://www.clifford.at/icestorm/
Open-source EDA flow http://opencircuitdesign.com/qflow/
Older, academic tooling for FPGA's http://www.eecg.toronto.edu/~jayar/pubs/rose/rosefpga12.pdf
"[T]he home countries of major semiconductor companies ban the export of leading-edge manufacturing technologies to China."
http://www.mckinsey.com/client_service/semiconductors/latest...
Doubt it.
https://semiaccurate.com/2015/11/18/arm-charts-path-printed-...
These aren't better than silicon, but they probably will be close enough shortly.
After that, Moore's Law should go operational. And these have a huge advantage in terms of non-recurring expense.
Think taking all of the "maker" stuff but creating a chip instead of a PCB.
You missed the point. The point isn't to replace 14nm transistors.
The point is to allow engineers to do VLSI design like PCB design.
There are lots of interesting things that can even be done with 5um or 10um transistors, but we can't get there because the non-recurring expense is too high.
Here is an example: guitar pedals--specifically the analog delay ones. These pedals all use an ancient MN3205 bucket brigade CCD chip. The chip is dead simple to make, but since the volume is too low to offset the NRE, nobody is willing to make it.
If, however, you could design that chip with an NRE of $1,000 instead of $100,000, you could make a tidy profit. In addition, you would probably pull all of the other functions of the pedal into the chip as well.
There are also other nice benefits to using older and larger transistors. One of them is voltage tolerance. Modern transistors can't take voltages at even 3.3V in many cases, while the old 10um transistors could go to 18V (old school 4000 series CMOS was specified from 3V-18V for most chips).
The point isn't to replace silicon transistors. It's to create a completely separate market at a different volume point.
> These aren't better than silicon, but they probably will be close enough shortly.
If you had said:
> The point isn't to replace silicon transistors. It's to create a completely separate market at a different volume point.
We would have no disagreement. You changed your initial entire statement.
It boggles my mind that China went communist because of perceived gap between haves and have-nots. But now, the gap is even greater and severe, under a communist government. The more I learn about history, the more I find how history is full of ironies.
Like the singularity, quantum computing is always around the next corner, I think it's more than 15 years off and there are a whole lot of questions.
At a glance, China seems like they could excel at fabrication. You need to invest a lot to get going but it's been repeated, there are models to bootstrap that process. there are also more fabless chip makers than ever and they want to build stuff as efficiently as possible. As for design, that seems more competitive and there are more factors on its success (like the actual software and developers adopting your stuff) their schedule seems aggressive but there are so many middle areas. License power or arm and tweak it some, who designed that? Do a couple generations and it starts to be original work, nobody in the west will think that but I don't know if that matters
For channel materials, Intel has been using SiGe as the channel of their pmos transistors for two generations now, so in a sense elemental Si has already disappeared from the channel.
The wafers are still made of Si, half the oxide stack is still Si-O-Ni. The rest is Hf-O.
Polysilicon is still used a resistor material.
The semiconductor industry has changed a whole lot in the past, there is no "Silicon is dead moment", rather gradual shifts toward different materials used for different purposes.
Satellite radio transistors (HEMTs) already use InP and GaAs and have been for decades.
If you want an actual prediction:
Gallium Nitride and Silicon Carbide are going to be the two actively worked on materials in the near future. Both have a higher bandgap than Si, which means less leakage current and lower static power but they both come with their own problems.
There is a lot of research going in semiconductor industry, people are well aware of the problems and there tons of possible avenues of improvements on current technology (such as better lithography).
You can stack chips, make architectural improvements, etc but the exponential scaling from moore's law will be over.
That being said, someone should re-invent the antiquated world of EDA, electronic design automation software, the software that you can use to design chips.
Several major problems of Cadence/Synopsis....
* No Forking of designs(not easily) * Min Cost is too high. --Should be browser based solution. * Version Control is lacking in chip design teams today.
It's hard to get single digit cost licenses for chip design.
If you thought Encounter was slow, wait until you see the browser version!
I've actually been in an EDA startup. The problem is that changing the toolchain is percieved to increase design risk, and you have to overcome the usual problem that the people paying for "enterprise" software aren't the users. It's quite hard to make headway in this environment.
Why, oh why? Browser is a crappy environment and its unsuitable for any kind of bigger application. It's too bloated and too constrained. Google is barely able to make WordPad equivalent run in the browser - I don't see any actually usable EDA as a browser solution any time soon.
Agreed that Synopsys/Cadence tools are severely outdated and terrible.
They need to be rewritten from scratch and designed by actual UX/UI designers for the front end. Their backend algorithms are already state of the art for the most part (e.g. Place and route and simulation).
A huge hurdle in making a Virtuoso or Encounter replacement is that foundries are not going to work with you because they already have such deep relationships with Cadence/Synopsys. It's an extremely nontrivial task to get that up and running.
Please no. That's exactly the type of suggestion I would expect to see a startup looking to do things "differently" to get attention, without bringing any really value to the overall system.
If they want to catch Intel, they will want to at least spend as much as them, and also benefit from IP-leakage. So I guess it depends on how far you think they are behind the state of the art?
Designing chips makes money, manufacturing them in $10B and rising fabs that stop making profit after a few years not so much.
In general I think it will accelerate humanity's technological progress. Looking on an individual level the most impactful inventors weren't motivated by monetary reward (e.g. Tesla). If you are a true inventor it is a tap you cannot turn off. If you are a bro who is in it for the money chances are you will not invent anything particularly ground breaking. The major concern here is that it's the bros who control the capital.
It is not there yet but not much success is too negative imho. Looking at Apple and other western brands here I think it even went too far. Mao would not be happy.
You mean kind like you're free to go wherever you want in prison, as long as you don't try to walk out of the door of your cell? :P
Here's a really sobering thought though:
> You can have a lot of political 'change' in the United States, but will it really change that much? Will it change the amount of money in someone’s bank account? Will it change contracts? Will it void contracts that already exist? And contracts on contracts? And contracts on contracts on contracts? Not really.
> So I say that free speech in many Western places is free not as a result of liberal circumstances but rather as a result of such intense fiscalization that it doesn’t matter what you say. The dominant elite doesn’t have to be scared of what people think, because a change in political view is not going to change whether they own their company or not; it is not going to change whether they own a piece of land or not. But China is still a political society, although it is rapidly heading toward a fiscalized society. And other societies, like Egypt, are still heavily politicized. Their rulers really do need to be concerned about what people think, so they expend proportionate efforts on controlling freedom of speech.
-- Julian Assange
My sister's home town in Michigan says it's illegal to spit on the ground. America threatens jail time for possessing weed!
Despite leaking state secrets, Snowden's life was never threatened by US officials. Yes, he's a whistleblower, but his actions are far more serious than developing software. Even then, US officials talk of imprisonment at worst.
http://www.foxnews.com/politics/2013/06/11/ron-paul-us-gover...
http://www.businessinsider.com/us-spies-talk-about-killing-s...
http://www.theguardian.com/world/2013/oct/03/edward-snowden-...
Your second link cites Buzzfeed as a source, and the Buzzfeed article[1] claims some unnamed low-level officials would like to kill Snowden. Yet these anonymously-quoted people also admit that they wouldn't do it unless they had approval from higher up. Pretty close, but still not a threat.
Your last link is about a retired government official making a joke in very poor taste.
None of these are actual threats, let alone by US officials. They're either idle musings, or bad jokes, or "I want to, but I'm not going to." I certainly don't agree with the people making these despicable statements, but they're qualitatively different from, "If you don't stop leaking state secrets, you will die."
1. http://www.buzzfeed.com/bennyjohnson/americas-spies-want-edw...
What seems to be the case is that they're shooting for slow incremental improvement, moving from less touchy areas to more touchy areas. They've gone from not talking about corruption to cracking down on it higher and higher up the ranks. It's a very slow sleigh of hand, with the distraction being economic growth, and various distinctly communist tasting programs (the current one should be 中国梦(Chinese Dream), if I recall correctly).
But their fear doesn't seem unjustified. There are still spontaneous uprisings which are usually nipped in the bud before they can escalate by means of information and communication monitoring. That's what they care about. Expats in China using VPNs? They don't give one.
Whilst I'm all for democracy, introducing it to China right now would work about as well as introducing it to the Middle East. Groups feeling under-represented would fraction off, there would likely be civil war, widespread devastation to one of the most populous countries of the world, and worst of all, whilst Iran getting nuclear weapons seems like a looming threat, China does have them. The sheer pandemonium would make Syria look like a backwater stand-off. I may be exaggerating, but then look at the bloodshed of the cultural revolution.
I grew up in Germany, so it's not like I'm opposed or unfamiliar with the idea of democracy, it's an incredible system. But for China right now? Well meant, but not practical. It'd be a less bloody affair to turn the US into a monarchy (to make a comparison). In 20 years, maybe, though it might take longer.
Thinking China can be fixed with a fair heaping of democracy and human rights ignores a lot of very significant details in a country that works contrary to a lot of accepted wisdom in the west.
/rant
PS: There's lots of interesting things going on here, playing churned lover with western ideals is literally the least interesting thing to talk about. The day before yesterday they announced changes to the Visa regimen, coming into effect in March. 72-hour visa free visit in 4 yangtze river cities, and some streamlining of visa/green card programs for tech people and entrepreneurs working in China. Maybe I should have chosen a less ironic user name, because this is starting to sound like a sales pitch.
The new visa regimes are interesting, but it hasn't seem to trickle down yet to us tech PhD people's. Really, China doesn't like foreigners, and they merely tolerate us, making us register at police stations after every trip, making us renew our visas yearly, not giving us access to the same services as natives.
The regulation I'm referring to seems like a pilot program in scope. It allows foreign students to work for and start tech businesses in the Zhongguancun area, as well as allowing people that have worked for 4 years or more for a technology company here to receive a residency permit. That's been exciting me the last 3-4 days since I heard of it, since I've been playing with the thought of doing exactly that.
It does feel like people well versed in technology are the bottleneck for startups over here. Investment is plenty, people with ideas are a RMB a dozen, but tech people knowing more than some PHP or how to set up Bootstrap? Not nearly as many, at least where I'm hanging out.
It's certainly an interesting couple of years ahead, as long as they keep tolerating us here.
If you are a true inventor it is a tap you
cannot turn off.
You would need a mechanism to catch that extra value and redistribute it to real R&D. I'm not saying its impossible, but it's not as easy as just saying lets do it.
I completely agree. However, I propose we put a serious damper on growth. Now, I know I cannot effectively or efficiently dictate how capital is allocated but we can do something almost as good. We can try to disassociate risk and reward.
It is my sincere beliefs that the inventions, the discoveries, and the creation of useful arts will take place sooner or later even without the motivation of the 100x gains. It might take a few years, decades, or centuries but we will get there. Yes, the plan requires a hint of central planning but I think it is acceptable.
The more I think about this the more I am convinced that an effective maximum yearly individual income is pivotal to the success of our society going forward. If adding the value of perks would amount to the annual maximum income would push someone over the maximum individual income then they should rightfully pay the correct income tax on the sum over the maximum income (which I tentatively set at 100 * 2000 * minimum wage per hour, it ought to be lower but we need to be pragmatic).
The reason we need to spend so much money on everything and we need 100x gains is that our enterprises have become too top heavy. The only "simple" way to curtail it that I can think of is to impose a maximum personal income (above which we tax away a large part, lets say 90% of the income). We can make it something more palatable. We have the ability to do it. We just need the will power to effect this tweak in the way our society rewards people.
> Meanwhile billionaire Larry Ellison leased his own aircraft to the company he runs for $1.5 million. (source http://www.bloomberg.com/news/articles/2014-05-02/ceos-take-... )
Now this idea isn't perfect. Right away I can think of a way people will try to game the system by "packing" the board of directors with nephews, nieces, children, grandchildren, butlers, and maids of the corporate officers. I don't have a solution to this yet but I am sure we can fix these bugs as we see them in production. The idea is to get the ball rolling and iterate as we go.
If karma were true, then the West getting Chinese tech was also deserved.
And how much more would we have gotten from Tesla if he hadn't been in financial ruins? He's probably proof positive that without the financial incentive/backing, the world loses out on countless inventions.
Even at the time, Tesla's greatest contributions were being simultaneously discovered by contemporaries like Galileo Ferraris. Discovery marches on, and is more a product of the existing body of scientific knowledge than of any one man.
There have been too many other great minds in physics and electrical engineering for it to be remotely plausible that we've missed out on "countless inventions" that would be available to us today, if only Tesla had more money for his experiments.
How will you know where the semiconductors are made?
If anyone else is curious: http://appleinsider.com/articles/15/12/07/apple-supplier-tsm...