After sending the two away, Xu Chuan returned to the second floor to pack his luggage.

If he followed his old habits, he would usually stay in his hometown until the end of the twelfth year before setting off.

After all, it is rare to come back several times a year, and it is a sure thing to do during the New Year to accompany your parents and relatives.

However, this year's situation is special, and he is ready to leave on the sixth or eighth day of junior high school.

Although I don't know how far HiSilicon and Huaxin have advanced chip research, the international situation and the country's desire for chips are indeed not optimistic.

When chatting with the two in the afternoon, Ding Jingguo, the leader of the development and reform, mentioned these things more than once, especially after the realization of controllable nuclear fusion technology, the relevant situation and development have become more severe.

Therefore, Xu Chuan planned to end his vacation early and go to Hisilicon and Huaxin to see the situation.

If he can solve these mathematical problems and the country can make a breakthrough in chips, he is still very happy.

After all, the development of science and technology has never been a one-person thing.

Just like a breakthrough in controlled nuclear fusion, it would not have been possible without the efforts of other researchers, engineers, and even grassroots workers such as security and construction.

In those fields that are easy to be ignored by ordinary people, there are countless people who struggle silently.

After saying goodbye to his parents, Xu Chuan took the high-speed train back to Jinling first.

Outside the villa at the foot of the Purple Mountain, under the leadership of Zheng Hai, a middle-aged man passed through the security check with some nervousness and some expectation, and entered the villa that did not look outstanding.

"Academician Xu, the people from Huaxin brought it over." In the living room, Zheng Hai reported.

Xu Chuan nodded and said with a smile: "I'm sorry for you." ”

Zheng Hai: "I'll go out first, if there's anything, you can call me at any time." ”

After saluting, Zheng Hai turned to leave. In the living room, there were only the middle-aged men arranged by Xu Chuan and Hua Xin.

Seeing the figure sitting in the living room, the middle-aged man was a little nervous, but also a little expectant.

After all, this is the legendary bull, and in terms of achievements alone, he is no different from those well-known figures in history who are admired.

Xu Chuan didn't pay much attention to it, glanced at the middle-aged man who was standing there nervously, and said with a smile: "Sit down first, has the data problem been brought?" ”

"Hello Academician Xu, my name is Mao Shun, I am the leader of the third group of Huaxin Technology R&D Department, mainly responsible for mathematical design, this is the information you requested."

Hearing this, Mao Shun was excited, walked forward quickly, took out a small box from the backpack he was carrying, took out a hard drive after entering the password, and handed it to Xu Chuan respectfully.

Xu Chuan stretched out his hand to take the hard drive, looked at it twice, and said with a smile: "Do you understand all these mathematical problems in the hard drive?" ”

Mao Shun: "I don't dare to say I understand, but most of them understand something." ”

This is also the main reason why the leader arranged him to come over this time, his level is not high, theoretically speaking, if you want to dock with this big guy, you should be a deputy general manager level figure at the most.

However, at the meeting yesterday, Mr. Ren of Warwick said that it was best to arrange a person who has some understanding of chip mathematical design.

But if you want to understand mathematics, I'm afraid that in front of the big guy in front of you, no one in the world dares to say that you know mathematics.

Throughout the history of mathematics, it is difficult to say how many people this big bull can rank. But in this era, he is undoubtedly the first person in the world of mathematics.

"Then don't waste time, come with me." Xu Chuan nodded, got up and walked towards the study.

Solving mathematical problems in chip design is a matter of no time sooner rather than later.

Since the Mao Shun in front of him knows something, let's take a look at it first, and if there are any questions, you can also let him explain it on the spot.

In the study, Xu Chuan connected the hard drive to his computer.

I didn't use a notebook, I used a small supercomputer that Zheng Hai arranged for someone to build in the villa next door.

If his notebook is still likely to be hacked because it will be connected to the Internet to query papers and other things, then this specially built small supercomputer is basically not likely to be hacked.

This kind of thing that involves the development of national chips should be paid attention to.

Clicking on the hard disk and entering the password, Xu Chuan found the problem file that Mao Shun said, and carefully flipped through it.

Although he doesn't know much about the circuit design and other fields in chips, he still understands the operation basis of chips or computers.

The foundation of the computer was inseparable from three people, Leibniz invented binary, George Boole brought Boolean algebra, and Shannon described switching circuits.

The combination of these three methods laid the foundation for the development of computers.

Binary is the algorithm change of the decimal system, which is the basis for the operation of the machine and the embodiment of the introduction of mathematics into life.

It is undoubtedly the lowest level of the computer, because whether it is simple or complex, the problem needs to be simplified into arithmetic operations and operated by logic switches.

The core of this is the use of Boolean algebraization to reduce all operations to 0 and 1 and non-operations.

For example, if you add a+b, the carry digit is a&b, and the 0 digit after addition is (a&!). b）|（b&！ a) That is, it can be used to express addition and subtraction with or, or not.

Binary, combined with Boolean algebra, is able to represent all mathematical operations.

Claude Ellwood Shannon brought "Symbol Analysis of Relays and Switching Circuits", which brought binary and Boolean operations to reality through relays, laying the foundation for today's digital circuits.

Today, no matter how the integration of chips continues to increase, how the latest CPUs, GPUs, mobile phone processors and other devices are released, and even the number of transistors has exceeded 10 billion.

But the mathematical logic that runs inside it is still the binary proposed by Leibniz, the Boolean calculation invented by Boolean, and the switching circuit described by Shannon.

It hasn't changed!

Of course, this is only the lowest foundation.

The real calculations and designs in today's computer chips are much more complex than that.

Including the information that Xu Chuan will see.

Although he doesn't know much about chip design, he understands mathematics and modeling.

So he didn't go through the math problems for the first time, but learned some information through the documentation first.

After all, if he didn't even understand the situation, he couldn't solve the problem.

It's like solving a math puzzle where you first have to understand where the core of the problem is.

Through the information brought by Mao Shun, he had a general understanding of the ideas and progress of HiSilicon and Huaxin.

Judging from the data, the architecture of the seven-nanometer chip designed by these two companies is based on the ARM architecture.

The so-called Arm architecture is a software compatibility architecture implemented by ARM for various microarchitectures to provide various combinations of power, performance and area, and is one of the three strongest architectures of the current processor (CPU).

Buying someone else's infrastructure and then developing it is a very common thing in the chip industry.

Just as the vast majority of scholars working on a mathematical problem need to step on the shoulders of those who have gone before them.

Constructing your own method from start to finish is still a very rare thing.

And HiSilicon and Huaxin are doing just that.

Through in-depth research on the ARM architecture, they continued to optimize and transform it, and established a set of their own NPU neural network architecture.

The so-called NPU (Embedded Neural Network Processor/Network Processor) is a processor that is specially used for network application packets, and it is also a type of integrated circuit.

But unlike the single function of special-purpose integrated circuits (ASICs), neural networks are more flexible in dealing with more complex problems.

Generally, software or hardware can be specially programmed according to the characteristics of network computing to realize the special purpose of the network, and many different functions can be realized on one chip to be applied to a variety of different network equipment and products.

The architecture designed by HiSilicon and Huaxin is better at processing massive multimedia data such as video and images than traditional methods by adopting a data-driven parallel computing approach.

It can be said that the blue is better than the blue.

However, at present, the construction of this architecture involves a lot of systematic mathematical problems, such as activation functions, two-dimensional data operations, decompression and other modules.

In the case of a regular CPU or GPU chip, these problems are not a problem. However, the NPU neural network architecture is different in nature from the first two, and its mathematical compatibility is completely different.

As a result, the designed chips are even inferior to 10nm or even 14nm chips in terms of performance.

At present, the difficulty of card owners HiSilicon and Huaxin is here.

Neural networks and machine learning processing are still in the early stage of the explosion of demand, and there are relatively few mathematicians in this area than traditional CPU/GPU architecture mathematicians, and there are not many foreign talents to study this area, let alone in China.

So as a last resort, Fagai found him, hoping that he could help solve the mathematical problems in this area.

After understanding the overall situation, Xu Chuan clicked on the math file and flipped through it.

Eigenvalue Calculation of Large-scale Sparse Matrix

"The overall solution of the initial value problem of a system of nonlinear ordinary differential equations"

Initial Value Problem of First-Order Nonlinear Time-Delay Differential Equations in Neural Network Architecture

Eigenvalue Parallelism of Polynomial Matrix Based on NPU

《. 》

The questions crossed in his eyes, there were not too many questions, not too few, a total of more than a dozen, most of which were based on the difficulties born under the NPU neural network architecture.

Xu Chuan first clicked on the problems one by one and looked at them, from a mathematical point of view, these problems are not difficult for him, but when it comes to the NPU neural network architecture, he doesn't understand it very well.

If he wants to study it, he has to learn the NPU neural network architecture first, understand the general idea, and then do it.

But that's not all, there are two or three math problems he should be able to solve by now.

No matter how complex the mathematical problems involved in chip design are, the same is the difficulty for him.

No matter how complex the calculation method is, no matter how complex the model architecture and values, it is impossible to be more complex than the high-temperature and high-pressure plasma turbulence model in controlled nuclear fusion, and it is impossible to compare it with the seven millennial problems.

In his opinion, having solved the two millennial problems, these problems may be a bit exaggerated in terms of elementary school mathematics, but at most they are the level of college students.

Anyway, when he saw these problems, he already had ideas and methods to solve them in his mind.

After spending more than half an hour roughly going through the documents that Mao Shun brought over, Xu Chuan sat up straight, stretched his waist, let out a long sigh of relief, leaned back in his chair and thought.

On the opposite side, Mao Shun, who had been sitting in the study for a long time with his butt next to half of the chair, suddenly cast an expectant gaze.

But before he could speak, Xu Chuan, who was opposite, took out the manuscript paper and pen from the drawer, lowered his head again and began to get busy.

Seeing this, the words that had already reached his throat were forcibly swallowed by him again.

On the other side, Xu Chuan ignored Mao Shun, who was still sitting in the study, and the black signature pen in his hand quickly listed the title on the manuscript paper.

【Stability Analysis of Stochastic Delay Differential Equations Driven by Fractional Brownian Motion】

"Some notation and function spaces are defined as (Ω,F,P) as a complete probability space with a non-descending σ algebraic family {Ft}t≥0 satisfies the usual condition that {Ft}t≥o is right continuous and Fo contains all P-zero sets, for all t<0 to make FL=Fo. symbol |·|Indicates "

"Then the time fractional Sobolev space can be defined as its norm as: Hs0(D)={U∈Hs(d),supp uD}"

“.”

【High-precision calculation of eigenvalues and singular values of Vandermonde matrices】

"Give first."

“.”

In the study, Xu Chuan was busy solving problems.

Sitting at the desk opposite him, although he didn't know what the big guy was studying, Mao Shun didn't dare to disturb him, so he could only sit silently and wait.

Time passed little by little, and in front of the desk, Xu Chuan had forgotten that there was another person in the room.

After writing down the last line of the equation, he looked at the manuscript paper on his desk and touched his chin.

Three questions, almost.

After cleaning up the manuscript paper on the table, Xu Chuan checked the questions answered according to his own process.

Although the questions of this level are not difficult for him, rigor is essential.

After reviewing the answers on the manuscript paper, he nodded in satisfaction.

Okay, theoretically it shouldn't be a problem.

On the other side, Mao Shun, who finally waited for Xu Chuan to stop calculating, thought about it, and after hesitating for a while, he still mustered up the courage to ask: "Academician Xu, what are you doing?" ”

Xu Chuan sorted out the manuscript paper on the table and said, "I just took a cursory look, and most of the questions in it are related to your NPU neural network architecture, and I can't give you an accurate answer for the time being." ”

"But I've solved the purely mathematical problems for you."

On the other side, hearing this answer, Mao Shun was immediately stunned, and stammered: "What do you mean that you have already helped us solve it?" ”

Xu Chuan smiled, stood up and walked in front of him with the manuscript paper, handed the manuscript paper to him, and said with a smile: "Literally, these manuscript papers are the process and the answer, although I haven't had time to sort out and print them out, but I think it's about the same, you can just bring it back later." ”

"As for the other problems, I just took a look, and most of them are based on the difficulties born under the NPU neural network architecture, and I may need to study them."

Mao Shun took the manuscript paper with a confused face, flipped through it, and the white text was already full of various mathematical equations.

Looking at the stack of manuscript paper in his hand, he couldn't help but swallow.

It seems that it shouldn't have been two hours since he walked into this villa, right?

Glancing at the time on my watch, it was two thirty-four p.m.

Clearly hearing the sound of himself swallowing saliva, Mao Shun recalled that if he remembered correctly, he should have arrived on time at one o'clock.

Aside from the security check, communication, and the time for this academician Xu to read the information, it should only be about an hour, right?

It's really hard for him to describe the shock.

As the leader of the team responsible for mathematical design, he has also been in contact with many mathematicians in China in the past year, including academicians of the Academy of Sciences.

But the problem that has not been solved in the past year has been solved in an hour in this hand?

Even if only one is solved, isn't that a bit too exaggerated.

(End of chapter)