"Traditional PCs are built around CPUs, a CPU, a motherboard, plus memory and graphics cards, everyone knows. This architecture is not scalable, forcing everyone to change it every three to five years, and this kind of replacement is a business need, not a real need for everyone. ”

Xia Ming started talking about the clover architecture.

"The Phantom's clover architecture, called Trefoil, has been proven on the Galika and Alfalfa engines. So in March, we're going to take it a step further and expand this architecture to the entire PC architecture. At that time, everyone's computer will be completely different from the current one, Clover Core will build a virtual pool, CPU, memory, graphics card and other functional cards, in this pool will be integrated into the overall resources of computing and storage, which is also the technical idea of data center and cloud computing. But we relied on the principle of quadruple virtual machines and brought it to personal computers. ”

"In the future, when you buy a computer, you can first buy a clover core plus the most basic hardware, and if there are higher requirements, add CPU, memory and display module. Yes, the CPU of the future will only be a module. ”

"The first generation of Clover cores will allow each computer to have up to four CPUs, four display modules, and different models. If that's not enough, you can take the form of multiple clover core afterburners. If the Clover Core has a new generation, it can also integrate the old Clover Core, so if it is just a general application, the Clover architecture can make one of your computers last a long time. ”

"Cicada's hardware department is negotiating with accessories and peripheral manufacturers about the industry standard of the new architecture, the interface may not change, but the inside of the chassis will change greatly, and the biggest change is that there are no cables. The connection of CPU, memory, graphics card and other modules is like building blocks, making it easier to install a computer. ”

Xia Ming gave a rough explanation of the Clover architecture, which is an extension of the hardware of the Galica and Alfalfa engines, which is only a small branch in his technology roadmap, and it is only a transition.

However, for the computer industry, the introduction of the clover architecture is a major earthquake.

PC sales have been declining, and sales are still at 300 million units per year. Once the clover architecture is accepted by everyone, laptops will definitely be closer, and the annual sales of laptops are also about 200 million, and the two add up to a big change.

These two industries are actually very eager for change, and if Clover is really powerful, CPU manufacturers, component manufacturers, and machine manufacturers will welcome it. Clover's architecture is highly scalable, allowing manufacturers to make more configuration schemes. As for the slow update cycle, if there is an application drive, users will definitely buy more hardware.

Therefore, although this topic has nothing to do with "Phantom Galaxy", the audience is still watching patiently, and the barrage is also intensively brushing and discussing, and most people welcome this "buy on demand" method. Of course, there are also many otaku who lament that the cicadas are too bad and want to take away the opportunity for them to install computers on their sisters' papers.

The launch of the Clover architecture in March was not Xia Ming's temporary decision, he had already done the relevant technology, and left it to the Cicada General Hardware Division to implement. Hu Shanhai, director of the business department, is in charge. What you have to do is to run errands and move your mouth, and first find some component manufacturers to come out with a testing machine.

It's a bit of a rush to officially release it in March, but it's just a prototype that can run the game, and it can be done this month, so there is no problem with the progress at all.

As for whether it can bring a change to the computer industry, it depends on the follow-up promotion, especially the promotion of applications, after all, everyone buys computers not only to play games, Clover also needs to pull Microsoft and other manufacturers in.

Xia Ming thinks that if Microsoft is sensible enough, it should follow up, and he can't wait, because Google is eyeing it.

If Microsoft's Winows doesn't want to join, Google's Android will be happy to attack the desktop market with Clover. In addition to large-scale games and many professional applications, you have to rely on Microsoft, general applications, such as office or something, Google directly uses the Android version to change the resolution, adjust the details such as materials, and it will soon be ported.

So, if you change the perspective to the operating system, Clover will really set off a revolution.

For this reason, Hu Shanhai's whole person was completely burned during this time, and he was busy day and night. In Hu Shanhai's view, Cicada will rule the desktop world of computers through the clover architecture, even if it can only collect licensing fees, it will receive softness, and he has set a target of $100 million this year, net profit.

Xia Ming didn't want to tell Hu Shanhai that clover was actually just a small action, and the money he earned was only pocket money.

Next up is the big move and the big money: the neuronal chip.

He had a detailed discussion with Xiaotang and Lin Cheng in this direction, and he also had a specific framework, but the progress in productization was definitely not fast.

Luo Xiang finally fulfilled the duties of a reporter at this time, she asked: "I think many people still don't understand what the neuron chip is about, and what kind of relationship does it have with artificial intelligence?" ”

This is the popular science that led to Xia Ming, after all, among the millions of viewers, there are not many people who really understand this.

Xia Ming stopped rhythmic, and said seriously: "Regarding the meaning of neuronal chips, I want to say that this is a natural and logical direction, just like we people must have brains before thinking." ”

"If we want to do artificial intelligence, we not only need to imitate the thinking mode of human beings, but also need to have a corresponding thinking carrier, that is, the physical structure of the brain, so neuronal chips are the only way for artificial intelligence. It's a simple, observable and controllable model that simulates a brain that until now is only a drop in the bucket. ”

"We've made a lot of progress in the software of artificial intelligence, with theoretical models of artificial neural networks, a lot of algorithms on the models, and applications like Alpha Dog and Beta Dog. But in terms of hardware, we still rely on transistors and von Neumann machines, and the efficiency and power consumption are completely inadequate. ”

"Research in this direction has been going on for many years, and IBM, HP, Google, and even in China, have put forward different solutions. There are memristors, graphene, and traditional transistors, but there are no mature solutions. ”

By this time, even many people on the other side of the ocean were watching the video and complaining that there was no decent simultaneous subtitle group translation in their country. Intel's Yaari frowned, it was late at night, and he was still listening to a stammering translation from a Chinese employee.

In the classroom of Yanhua University, the teacher said to the students with great emotion: "I hope you can be like Xia Ming, and don't give up your efforts just because you didn't go to Yanhua Ziguang to study." ”

"But he's a journalism major!"

A naughty student, like An Daoyuan, said this, and the classroom burst into laughter.

The teacher said imposingly: "You can be self-taught like Xia Ming, all the professional courses are suspended, and the school will also send you a graduation certificate!" ”

"Could it be that Xia Ming's professional classes are all hung up!?"

"What about the dignity of the university?"

"Teacher, I want to give up even more when you say that!"

The students were even more in an uproar, Xia Ming really hung up the subject and got his graduation certificate? That's not how specialization is done, is it?

The teacher said disdainfully: "Can't you even understand the parable? When did Xia Ming fail the course? In the past four years, many of his exams have been ......"

"Full marks?"

"Pass."

"Wow, that's amazing!"

In this era, the passing grade is also a top student, and it is higher than the top student with a full score.

The hiccup passed quickly, and everyone's attention turned back to the interview video.

Luo Xiang continued to be a good introducer: "So by June, what kind of technical route will the neuronal chip launched by Illusion take?" ”

Xia Ming replied: "Illusion hopes to let everyone enjoy the convenience brought by artificial intelligence as soon as possible, so the neuronal chip released in June is based on traditional transistors, which will use a new generation of 5nm manufacturing process, which is also the last generation of traditional silicon-based transistor manufacturing process." ”

"I know that companies like IBM and Samsung are also working on more advanced derived silicon-based transistor technology. However, I think that after our solution is released, the silicon-based transistor technology will also come to an end, so our neuronal chip is called Terminator, which is the Terminator. ”

"Uh...... Why? Luo Xiang asked stupidly, he should be pretending to be stupid at this time.

Xia Ming smiled: "Because complex tasks will be handed over to neuronal chips, those simple tasks, 5nm silicon chips are enough to be competent." When neuronal chips are further developed, they will directly replace traditional silicon chips. ”

Many professionals watching the video nodded subconsciously, Xia Ming's sentence seems concise, but it actually makes the current situation, turning point and future trend of chip technology clear.

Xia Ming continued: "In fact, the difficulty of neuronal chips is not in how to do hardware, just like the human brain, the physical structure is the same, but there will be different levels of intelligence. The instruction set and development language on the chip are the difficulties of the neuron chip. ”

"There used to be a saying that the human brain is less than 10% developed, and if you increase this level of development, you can make people smarter, and in novels and movies, this statement has also become a science fiction idea, and when you develop to 100%, you can soar like Lucy in "Super Body"."

Although this statement is wrong from an academic point of view, it is very appropriate when applied to neuronal chips. Even if we had built a neuronal chip with 10 billion synapses, close to the structure of the human brain, we didn't know how to program the artificial brain. The AI models and algorithms we have now are great enough to allow this artificial brain to perform one-ten-thousandth of its capabilities. ”

Xia Ming's explanation is very vivid, and it is completely a basic lesson in artificial intelligence for millions of viewers.

"Therefore, when Mirage releases the solution, it will open source the basic instruction set, quadruple assembly, and development language of the solution."

Then his words became the most important news in the eyes of IT professionals.

Luo Xiang was dazed by the news sent by the editor-in-chief and even the editor-in-chief on her mobile phone, and she couldn't sort out her thoughts, so she repeated in a daze: "You mean that Illusion wants to open source the entire neuronal chip technology?" ”

Xia Ming nodded seriously, and said in his heart that if we don't open source, our plan can't be pushed.

Luo Xiang got the opportunity to buffer, and quickly chose one of the questions sent by the editor-in-chief: "The prospect you mentioned is really exciting, but even Google and IBM have not done it for so many years, how did you do it?" ”

Xia Ming laughed a little silly, and it was only after everyone listened to his answer that they felt as if he was laughing at the question too stupid.

"I didn't do it personally, it was our institute that did it. Don't forget, we're studying artificial intelligence, and we're studying strong artificial intelligence from the perspective of bionics, and we've created a simulated brain that resembles a human brain, and we've been studying how to communicate with this brain and teach it how to think. ”

"It's like...... How to put it, mankind first had the atomic bomb, and then nuclear energy to generate electricity. What we are working on is like an atomic bomb, and now we are taking some of the results and using them on weak artificial intelligence. ”

These words are fools, Qianyan is indeed a simulated brain, but without R3AI, he has no way to communicate with Qianyan, let alone combine the technologies of quantum computing shells with the current technology system.

Obviously, no one knew this, even Yaari on the other side of the ocean believed it.

In a villa in Longhai, Xia Ming's interview video was shown on the large-screen TV, and a new female star who has been quite active on TV recently said with emotion: "I kind of understand, why everyone always respects and fears scientists, our world is getting faster and faster, and the driving force behind it is people like him." This Xia Ming is very much like a big boss behind the scenes. ”

The female star cautiously asked the man who hugged her: "Childe, I heard that you have dealt with him?" ”

Xing Tianmin's eyes were quite complicated when he looked at the TV, he shook his wine glass, and said leisurely: "He is my friend, and we have a good relationship." ”