Petaflop supercomputers are artifacts in this era!

The development speed of supercomputers is very fast, but in 1976 it was only 250 million operations per second, so Chen Xu guessed that the supercomputer in the IBM laboratory had a maximum of 1 billion calculations, which would not be faster than this. Pen? Interesting? Pavilion wWw. ｂiquge。 ｉｎfo

It was not until 08 that the United States imperialism spent a huge amount of money to develop mankind's first petaflop military supercomputer, which was used to simulate nuclear explosions at that time.

It is only the early eighties, and the research on supercomputing has just begun, and the level of supercomputing in various countries is quite low, and even the military-grade performance is not much better.

Chen Xu has many experiments that require the use of supercomputers, and it can even be said that it is difficult to move an inch without supercomputing. Even at the national level, nuclear weapons, aerospace, climate, astrophysics, biomedicine, and materials science all require supercomputing to carry out intensive computing tasks.

Something that can't be calculated in 20 years by manpower alone, a supercomputer can do it in a few hours, which can greatly save time and cost.

Of course, Chen Xu knows that handing over the architecture technology of supercomputers to IBM is of great benefit to the United States, but what can Chen Xu do! All walks of life in China urgently need supercomputing to solve the problem, and they can't make it themselves. Even trying to develop a supercomputer would require no less money and time than building the six large nuclear power plants, and perhaps more expensive.

Supercomputing is actually throwing money, throwing out enough banknotes, and computing power will go up in a hurry.

Chen Xu has no ability to throw money at supercomputing anymore, and after thinking about it, it is better to cooperate with IBM, the computer manufacturing giant of this era, which is the fastest shortcut to make tens of millions of supercomputers. As the richest and most valuable computer company in the world today, IBM has the ability and financial resources to digest the supercomputing technology provided by Chen Xu.

Of course, IBM never dreamed that the architecture technology that Chen Xu came up with was actually IBM's blue gene supercomputer, and used IBM's technology in 2005 to exchange for IBM's money in the 1980s.

Blue Gene is a commercial supercomputer, and it does not reach the terrifying petascale computing power, it is just a petascale supercomputer with a peak computing speed of 367 trillion times. But if Chen Xu wants to raise the price, he naturally wants to boast about the technology and exaggerate it into a supercomputer of hundreds of billions of times, so that it can be sold for several times more money.

Petaflop supercomputing is unthinkable in this era, and Barthes, as a senior executive of IBM, of course knows that his so-called world's most advanced machine has only one billion computing power, which is several orders of magnitude worse than that of a quadrillion, so he will squirt out a mouthful of water.

"It can't be! With the performance of today's CPUs, even a CPU as tall as a 50-story building cannot achieve this level of computing speed. Bart felt that Chen Xu was deceiving people, as a leader in the industry, Bart knew very well that it would not be possible to do petascale supercalculations in this era.

"The technology that Silicon Valley does not have, does not mean that China does not have it, a country always has some technology that can be used, and supercomputers happen to be the military technology that my country focuses on researching, if it were not for the limitation of R&D funds, we would not choose to sell the technology to you IBM." Chen Xu moved his butt on the chair, changed to a comfortable position, and said with a positive color.

"How did China do the quadrillion calculation? There is no physical proof of it! "As a technician, Bart can't of course listen to Chen Xu's empty words, he must see it with his own eyes to believe it.

"If a supercomputer wants to improve its computing power, it is inseparable from the central processing unit and the supporting algorithm system, which are precisely the two core technologies that Panda Group has." Chen Xu clasped his arms with both hands, and his face showed a proud look.

"As far as I know, China's semiconductor industry is very backward, especially in terms of integrated circuits, which has not even started, and many semiconductor equipment needs to be imported from the United States, so how can you be worth buying from our company?" Bart-for-tat, as a technician, he believes that technology can only develop to a certain extent in a certain era, if Chen Xu can't explain it clearly, then the so-called petaflop supercomputer in his mouth is likely to be bragging.

Chen Xu scratched the back of his head, this kind of stubborn technician is the most difficult to explain, he is too confident in his own technology, and he always doubts this and that, so he asked: "The key to CPU technology lies in the improvement of integrated circuit technology, Mr. Barth, how many transistors can be integrated into the third-generation CPU on the market?" ”

"With current technology, 100,000 transistors can be integrated on a chip less than one square centimeter, which is the most advanced semiconductor process in this era. At a reasonable rate of upgrade, we could integrate 1 million semiconductors on a single chip in 10 years, and the etching process would have to be at least 1 micron at that time. Bart replied without thinking.

"Your suspicions are indeed justified, if the chip integration technology does not break through, the computing power of the supercomputer will not be able to improve, after all, the supercomputer is composed of a central processing unit, and the chip is the foundation of the supercomputer." Chen Xu nodded, the American on the other side was not easy to coax, Bart concluded on the basis of science that Chen Xu's petascale supercomputer could not exist.

Chen Xu paused for a moment, and then said: "If you want to integrate more transistors on the chip, reduce power consumption, and improve performance, the micron process has lagged behind. Now, after ten years of research, our company has developed a nano-etching process, and the number of transistors on the chip has exceeded 10 million. ”

"Oh my God! This is impossible, how can China develop such a technology. Bart slammed the table, the United States semiconductor industry is so developed, so much money has been invested until now, and it has only done tens of micron processes so far, and China has suddenly entered the category of nano technology, which he absolutely cannot tolerate, after all, one micron is equal to 1000 nanometers, if China has done the nano process, then the gap between the United States and China in the semiconductor industry is no longer on the same order of magnitude.

This also means that the United States has lagged behind China in IT technology across the board.

"What's so strange about this, since Panda Group can design the 'Windows operating system' and make outstanding contributions to the development of the IT industry, it is normal to develop nano processes." Chen Xu looked at Bart, who was a little broken, and said with a smile.

"In that case, can you show me the real thing? It's the kind of chip that integrates 10 million transistors. Barton lifted his glasses and swallowed pleadingly.

"I'm sorry, we haven't made such a chip yet, because due to China's backward semiconductor industry, we don't have a nano-scale lithography machine to make patterns on semiconductors, and then use nano-scale etching machines to remove unwanted patterns on the chips. But I must state that Panda Group has now designed nanoscale lithography machines and etching machines, and if IBM is willing to pay enough, we are willing to transfer the technology. At this point, Chen Xu finally got to the point, in fact, he had so much foreshadowing in front of him, it was nothing more than wanting IBM to spend money to make nano-level semiconductor equipment, and then Panda Group would pick up ready-made ones.

Panda Group can also invest manpower, material and financial resources to do it itself, but the cost is too high, and it can't be done in a few years, so it is better to sell the technology directly to IBM, which can not only get a considerable amount of advanced, but also save a huge amount of R&D investment.

The most important thing in the semiconductor industry is undoubtedly semiconductor equipment, and later generations of China is very backward in the semiconductor industry, the main reason is that the core technology of semiconductor equipment has always been mastered by several giants, and more importantly, they have spent decades to develop and master thousands of technology patents related to semiconductor equipment.

Patents have long been occupied by foreign giants, and domestic companies will naturally have no motivation for research and development if they can't make money, which is also the key reason why the domestic semiconductor industry has never been able to develop. (To be continued.) )