China's homemade CPUs on the way for local supercomputers

[tanlixiang28776]

Of course, they are now planning to build a more powerful Supercomputer than the previous Roadrunner.

IBM Sequoia - Wikipedia, the free encyclopedia

Obama already declared a race of Supercomputer with China in last year.

Great now they need more money from China to fund it.

 

[ao333]

not really that supercomputer become obsolete or technically old, when they have some applications to be run on. Built it when you need it, IBM did not build its supercomputers and let them idle.

These are not your lab experiments they are expensive to build and maintain.

Yeah, I agree. Running on 10% of output is a waste. They should've just made something 1/5 of the computing power but ran at half capacity. But it's a "communist" decision. Who can stop the collective "will of the people," right?

 

[Aramsogo]

Home designed, not made.

Wow, lots of sour grapes here. When China built its first fab, people like you put China down because they couldn't design. When they can design CPUs, you criticize them for being fabless. Being a fab is just buying Applied Material equipment. Costa Rica makes most of the Core i7's. AMD, ATI, Nvidia are all fabless.

 

[Kyusuibu Honbu]

1000 trillion wow!

Thats faster than prototype DNA computers i"ve heard of.

Good going China!

 

[Aramsogo]

Just like I suspected, building Super computers for ego-massaging. Bravo!!!

Do you even know what supercomputers are built for?? Ever wonder why most US supercomputers are with the Dept of Energy or the NSA. The Tianhe-1A is with a Chinese defense university. Supercomputers are used to design new bombs and test the reliability old ones or code-breaking. The Chinese are very secretive about their strategic arsenal. He can't talk about it.

 

[Kyusuibu Honbu]

Just like I suspected, building Super computers for ego-massaging. Bravo!!!

Can't the same be said about India then?

Supercomputers have lot of applications.

For example,ISRO uses Supercomputers for its rocket launches.

 

[Aramsogo]

You guys also mistake Chinese modesty for inefficiency or Loser. The Godson 16 core cpu is out in less than two years. Do you really think China is 20 years behind like he says??

 

[BlueDot_in_Space]

China good going!!

 

[Bigoren]

Just like I suspected, building Super computers for ego-massaging. Bravo!!!

Thanks for what you writen above to show ignorance of you. Go learn the important for help/design high tech weapon like a 5th gen before you say that.

 

[rcrmj]

Nope, if IBM wants to build a supercomputer more powerful than what they have, they could but perhaps being a private corporation they did not, because they spent time building WATSON having realized that Software that runs is where the catch-up needs to take place. There are no sour grapes here. May be you just need a Tickle Me ELMO to play ...

lol``you are funny, u really showed the level of stupidity, Bravo!!

 

[Anonymous user]

Yeah, I agree. Running on 10% of output is a waste. They should've just made something 1/5 of the computing power but ran at half capacity. But it's a "communist" decision. Who can stop the collective "will of the people," right?

Do you intend to reply on the thread you start i.e "Is the Libyan crackdown reminiscent of 1989 to you guys?" is it a habit for you to disappear once people start responding?

 

[qwerrty]

Can't the same be said about India then?

Supercomputers have lot of applications.

For example,ISRO uses Supercomputers for its rocket launches.

that explains indian rockets are so reliable and 20 years ahead of china.

 

[ao333]

that explains indian rockets are so reliable and 20 years ahead of china.

THey're reliable... at veering off from its trajectory, lol.

 

[gambit]

Wow, lots of sour grapes here. When China built its first fab, people like you put China down because they couldn't design. When they can design CPUs, you criticize them for being fabless. Being a fab is just buying Applied Material equipment. Costa Rica makes most of the Core i7's. AMD, ATI, Nvidia are all fabless.

Okey-dokey...There has been a lot of understandable misconceptions about the semicon industry and a lot of deliberate exaggerations about China's own semicon industry. So am going to set some facts straight.

Being 'fabless' does not mean the company does not have a 'fab', in fact, the company may have several fabs in different degrees of sophistication and sizes in order to support its R/D. The company need to verify its new design somehow so what else better than to have a fab completely under its controls? Part of that is for intellectual property security reasons before mass volume production. When a semicon company is called 'fabless' it simply mean the company does not have the facilities necessary for mass volume production. Somewhere on its campus, there is a fab working 24/7/365 to support its core business. We outsiders just are not privy to it.

Here is a May 2009 US International Trade Commission (USITC) report on China's semicon industry...With relevant information...

Goofernmental interference pre-2000 retarded China's own struggling semicon industry despite the government's declaration that an indigenous semicon industry is vital for national security under the 'four modernization' programs outlined back in the 1970s. Taiwan Semiconductor Manufacturing Company (TSMC) was founded by American educated Morris Chang back in 1986-7 and it took nearly 20 yrs before the company was considered to be 'world class' in semicon manufacturing. The PRC government finally realized its folly and changed its ways.

Figure 1 showed China's share of the world's semicon industry. Hardly a cause for those deliberate exaggerations, especially the ones exalting China's semicon industry over those of South Korea's and Japan's who are genuinely global power players from R/D to mass volume production.

Table 2 shows China's current status among the world's semicon players. Do not dismiss the fact that this is a 2009 report. A fab takes about 3-4 yrs to construct with all tools just delivered, not yet installed, calibrated and cleaned for production.

Here is a guide on China's semicon fabs...

A Bright or Bleak Future: Semiconductor Fabs in China | SEMI.org

Notice the majority are 150-200 mm wafer facilities. All the major players in the industry are 300 mm and moving towards 450 mm. Mass volume production and profitability are greater at 300 mm. These 150-200 mm fabs are best for 'low end' semicon devices such as in household appliances, toys and perhaps in other 'low end' multi-media products.

Back to Table 2...We see that China's participation in R/D and 'front-end' manufacturing is limited. Part of the cause is because of excessive goofernment's interference as shown in Table 1 and its accompanying paragraph. Part of the cause is...

SEMI F47: Increased Power Quality Concerns in Chinese Fabs | SEMI.org

SEMI F47: Increased Power Quality Concerns in Chinese Fabs

Due to the sensitivity of equipment and process controls, semiconductor fabs require high levels of power quality. Voltage sags are the number one power quality problem for equipment, because a 200 millisecond voltage sag could easily trigger EMO (emergency off) on various tools and cause productions lines to go down for hours. SEMI F47-0706: Specification for Semiconductor Processing Equipment Voltage Sag Immunity defines the voltage sag immunity required for semiconductor processing, metrology, and automated test equipment. However, for semiconductor fabs, the decision whether or not to adopt F47-compliant solutions must strike a balance between reduced vulnerability to voltage sags and increased equipment cost. Based on feedback recently received by SEMI China, awareness of SEMI F47 is steadily increasing across the Chinese semiconductor industry.

The cost of voltage sag mitigation is not the only big concern. Tool owners have found it difficult to quantify losses attributable to voltage sags, as local industry still lacks statistical data to characterize power quality. While it is known that voltage sag triggered EMO shutdowns in Beijing and in Shanghai’s Zhangjiang Industry Park, where China’s top three foundries are located, there is little data to show how strongly correlated these were to poor power quality, or what the main characteristics of the poor power quality were.

September 1, 2009

That report is not even 2 yrs old.

Back-end testing, die extraction, and packaging assembly equipments are not as sensitive to power fluctuations as front-end equipments. At worst, one wafer in a batch or 'lot' might be physically ruined by a misbehaving robot arm moving the wafer around inside an equipment. At the front-end where the entire lot is bathed in hot chemical or in a bake oven, the incorrect temperature or prolonged exposure will destroy the whole lot. The Chinese government knows the country's shortcomings in this. A batch of 150-200 mm wafers loss at the front-end when the industry is moving towards 450 mm wafer size throughout is a financial disaster if the product is high value commodity items like fast DRAM and NAND memory which the world is hungry for. Disrespect for intellectual property rights (IPR) does not help and as long as the Chinese government ignore this, China's semicon industry will remain 'limited' in R/D and front-end manufacturing despite this quite laudable achievement of an indigenous processor design.

 

[ViXuyen]

Being able to design and fabricate the microchip is one thing, being able to produce those hi-tech machineries to produce the chip is another level; this is the difference between South korea and Japan semiconductor industry.