The news that Lianchuang Technology intends to engage in CPU and supercomputer research and development at the same time caused an uproar. Pen, fun, pavilion www. biquge。 info

Countless scientific research units were excited, among them, the most excited were 56 institutes.

On the one hand, they bought the Alpha processor technology license from PAQ in 1999, and they can develop the processor they need on the basis of the Alpha architecture; But on the other hand, when the entire army does not have much money, it is actually pitiful to be able to allocate funds to 56 institutes for supercomputer research and development.

How pitiful is it, the huge 56 institutes, last year's financial allocation was less than 20 million, this 20 million not only includes the research and development expenses of Shenwei supercomputer, but also includes the entire research institute's personnel to eat horses, travel subsidies, vehicle maintenance...... In the end, the funds that can really be used for the research and development of Shenwei's supercomputer are less than 5 million.

5 million for supercomputers, pitiful, right? The unit is still RMB......

When the Computer Research Institute of the National University of Defense Technology was still hesitating, the Chinese Academy of Sciences was still arguing, the Modu Supercomputing Center was still considering, and the National Parallel Computer Engineering Technology Center was disdainful, Huang Yongqin, director of the 56th Institute, brought someone to the door.

Well, what I brought is Academician Chen Zuoning, who was just rated as an academician of the Academy of Engineering last year.

Seeing Lin Zheng, Huang Yongqin's first words were: "Mr. Lin, are you really ready to be the Chinese's own central processing unit?" ”

"yes......"

"Why don't you come to our 56?"

"Huh?" Looking at Huang Yongqin, who came to Xingshi to ask about the crime, Lin Zheng carefully touched his head, and always felt that there was something wrong with this plot: "This ...... Loongson research group ......"

"Can Loongson's MIPS architecture compare to our alpha architecture?" Huang Yongqin, who came to Xingshi to ask the guilt, waited for his eyes, with a posture of "why did you Lin Zheng give us 5 yuan to your wife": "The performance of the alpha architecture is much superior to that of x86, you don't know, right?" ”

At the beginning of the alpha architecture, Lin Zheng really knew how powerful the alpha was, and in Lin Zheng's opinion, this product was simply a bug-level existence: at the beginning of the design, the alpha had already taken into account the needs of 25 years later, and reserved a 64-core many-core architecture. In other words, if necessary, the processor with the Alpha architecture can completely replace the Xeonphi processor of Inter, which is far ahead of the MIPS architecture.

"To know is to know." Lin Zheng asked cautiously, "But...... Does this have anything to do with all of us at 56? ”

"Of course it matters," Academician Chen Zuoning interjected: "Does Mr. Lin know about the Shenwei-i supercomputer?" ”

"Kamui-i?" Lin Zheng nodded: "Impression, what?" Is it produced by our 56? ”

Lin Zheng really has some impressions of the Shenwei-i supercomputer, which seems to have been successfully developed in 1999, but the official information processing of this supercomputer is very low-key. The specific performance is really not clear to Lin Zheng.

"Of course," Academician Chen was quite proud: "It was successfully developed by our 56 Institute in 1999, and the highest floating-point operation speed is 384 billion times per second. ”

Lin Zheng was taken aback, he didn't know much about the performance of Shenwei-i, but he still had some understanding of the basic performance of several supercomputers that came out around 1999, in terms of the most familiar floating-point operation speed, the fastest floating-point operation speed of the Dawning-2000II supercomputer in 1999 was 111.7 billion times per second, and the floating-point operation speed of the Dawning-3000 in 2000 was 403.2 billion times per second. 56 institutes were able to produce 384 billion supercomputers per second in 1999?

I couldn't help but ask: "Could it be that this supercomputer is the alpha processor used?" ”

"That's right, and we've acquired the full intellectual property of alpha21164 by 56 firms. Mr. Lin, you may not know that the architecture of the alpha processor is designed to be very advanced, fully considering the needs of future computing, and a CPU can be expanded to up to ......"

"It can be expanded to 64 cores, and the public core architecture is reserved, I know this," Lin Zheng interrupted Academician Chen's words, turned his head to look at Huang Yongqin and said: "Director Huang, our 56 Institute is a supercomputing R&D institution affiliated to the General Staff Department, and you and we jointly develop CPUs." This one...... Classified ......"

"Of course, we came to you with the permission of the General Staff, "You are very stupid, you didn't even think of such a simple truth?" The innocent little eyes looked at Lin Zheng: "Haven't you Lianchuang Technology and the military always had a cooperative relationship?" The leaders of the General Staff Department have said that you are trustworthy. ”

It's still straightforward to be a soldier. Lin Zheng sighed in his heart: "So." 56 is a CPU that is ready to work with us to develop the alpha architecture, right? ”

"That's right." Chen Zuoning and Huang Yongqin nodded at the same time.

Looking at the two academicians, Chen Zuoning and Huang Yongqin, who were looking at him expectantly, Lin Zheng frowned.

There are many advantages to Alpha, the biggest of which is that the Alpha processor is the fastest of all the reduced instruction set processors, and it is the only one that continues to be supported by WindowsNT. Neither the ARM architecture, the MIPS architecture, or the PowerPC architecture can be compared - of course, if alpha is carried forward in the hands of the Chinese, it is not ruled out that Microsoft will exclude alpha.

"I'm not very familiar with alpha-based CPUs, but as far as I know, alpha processors were designed for the server market?" After thinking about it, Lin Zheng asked: "As long as it is used for servers, it can't be cheap, don't hide it, the reason why Lianchuang Technology is ready to cooperate with the Chinese Academy of Sciences is because the price of the MIPS architecture Loongson can be very cheap." ”

"Who says alpha architecture CPUs can't be made cheap?" Without waiting for Huang Yongqin to speak, Academician Chen stopped doing it first: "The alpha21164 sold by DEC is not cheap, according to our calculations, under the premise of the same performance, if the alpha central processing unit can reach the scale of ARM, the price can be only 80% of the price of ARM processor...... But can ARM easily scale to 64 cores? ”

The core of ARM is difficult to expand, not to mention expanding to 64 cores, and it takes a lot of effort to expand to 16 cores, because ARM did not consider enough when making the initial architecture design. The Alpha architecture was designed with a minimum life cycle of 25 years, but what about the ARM architecture? The design goal is 15 years.

"Well, if I think about it, I still need to do an evaluation of the 56 Institute and the alpha architecture." After thinking for a while, Lin Zheng finally made a decision: "But there is one thing I want to make clear in advance, in the short term...... At least within 5 years...... The alpha processor will not be in the desktop PC space, we will make this processor a processor for smartphones. ”

"No problem, I'll be able to answer that question right now. 56 Institute is willing to work with Lianchuang Technology to build Alpha into an excellent mobile phone processor, and we fully cooperate with the work of Lianchuang Technology! Huang Yongqin happily agreed, he could take the initiative to come to Lin Zheng, naturally he did his homework, knowing that after Lianchuang Technology cooperated with the Loongson research group, Loongson's target market was the central processing unit of smart phones.

For the 56 institutes, they are very satisfied with being able to sell the produced CPUs, and when they heard that the Institute of Computing Technology of the Chinese Academy of Sciences was arguing about whether Loongson is used for desktop personal computers or mobile phones, the eyes of all the 56 institutes were straight: it would be good if the produced CPUs could be sold to make money and support themselves. Can you still pick and choose? Do you think you're Inter or AMD?

Aren't you guys even good via?

With Lianchuang Technology's reimbursement capacity, you are still not satisfied?

The experts of the 56 institutes can only sigh that they are subordinate institutions of the Chinese Academy of Sciences, which are different from our poor military scientific research institutions.

Lin Zheng nodded, the attitude of the 56 Institute made Lin Zheng very comfortable, at least much stronger than the Chinese Academy of Sciences: "Another point, if you want to cooperate, the two sides are all-round cooperation, we not only want to cooperate in the development of alpha processors, but also cooperate in the development of supercomputers, you can also agree to this condition?" ”

"Collaborative development of supercomputing?" Sure enough, I heard Lin Zheng's request. Huang Yongqin and Chen Zuoning hesitated for a moment, and the two glanced at each other: "This ......"

"This is the premise of cooperation, if this article can't be agreed, don't talk about cooperation," Lin Zheng interrupted the two. "Of course, as a support for this project, Lianchuang Technology can now order an alpha-based supercomputer with a CPU entirely produced by us. ”

"You're going to buy a supercomputer?" Huang Yongqin's eyes widened in surprise, and his expression was full of incredulity when he looked at Lin Zheng: Is there any mistake, it's a supercomputer! Can you not say it so lightly, it's tens of millions of dollars, okay?

"That's right. We will book one, and the specific performance requirements will be submitted later," Lin Zheng nodded affirmatively, and looked at Huang Yongqin with a surprised expression: "It's just a supercomputer, you don't need to be so surprised, right?" There are many large enterprises and conglomerates in foreign countries that have their own supercomputers. ”

Seem...... Is this also the truth? Huang Yongqin touched his head, those big foreign companies not only have their own supercomputers, but some companies even have their own satellites.

But at the same time, he felt particularly awkward in his heart: so far, none of the more than a dozen supercomputers in China belong to enterprises, let alone private enterprises, even if it is two barrels of oil and two major mobile communication companies, they are just leasing a part of the computing resources of supercomputing.

"But what do you want a supercomputer for?" Academician Chen Zuoning, who came with Huang Yongqin, couldn't help but ask curiously: "As far as I know, you Lianchuang Technology already has three IBM mainframes, and these three mainframes can also meet your computing requirements, right?" ”

Supercomputers are all strategic things, mainly used to undertake major scientific research, national defense cutting-edge technology and large-scale computing tasks and data processing tasks in the field of national economy. Is this necessary? Having a mainframe is enough for your computing tasks, right?

What's more, it's still three.

When he said this, Academician Chen's heart was full of envy: the entire Jiangnan Institute does not have a mainframe, if the Jiangnan Institute also has its own mainframe, how good would it be?

Lin Zheng explained with a smile: "You are right, there are mainframes that can basically meet our computing needs now, but then our group's demand for computing power is increasing, and the three major business segments involved in the research and development of CPUs, such as extremely large-scale circuit design and simulation, new energy and industrial robots, have an increasing demand for computing power, and it is time to purchase supercomputers...... I'm afraid in a few years, we'll have to add another supercomputer. ”

One supercomputer isn't enough, another one has to be added? Do you own a computer room? Academician Chen was speechless when he heard it, but why was he so envious in his heart?

Huang Yongqin was already trembling with excitement!

He had already begun to roar in his heart: a supercomputer! A supercomputer! A supercomputer!

Because of the lack of accumulation in the field of supercomputing, 56 institutes have produced only one supercomputer so far, which is a shame for a supercomputing R&D institution that is interested in wrestling with the Chinese Academy of Sciences in the field of supercomputing.

If a supercomputer were pre-sold at the beginning of the commercialization of the alpha processor, it would be too big to overstate!

If it weren't for the fact that the 56 Institute is a scientific research institution under the General Staff Department, Huang Yongqin would almost have agreed to Lin Zheng's conditions now, but it is not too late, staring at the red eyes, Huang Yongqin gritted his teeth and asked Lin Zheng: "Mr. Lin, what you said is true?" ”

"Do you think I'm missing the amount of money I paid for supercomputing?" Boss Lin asked rhetorically, his tone was a heroic mess. (To be continued.) )