"How to build more transistors on a limited area is not only the design direction of the CPU, but also the direction of the memory. There is no need to talk about this. Let's mainly talk about the structure of mosfet."

When they entered the technical discussion, there were a few more people in the conference room. Su Yuanshan grabbed the manuscript paper and drew a mosfet structure diagram like a graffiti.

"Wait, I need to say it." Li Mingliu interrupted Su Yuanshan: "Mr. Shan, don't say it nicely. We all know that building a building on a piece of land can accommodate more rooms than building a bungalow. But chip design is not architectural design. Architectural design only needs to consider structural mechanics, materials and mechanics. As long as you don't engage in strange-shaped buildings, construction is generally fine."

"The difficulty of chip design lies in the implementation of the process. I can draw a multi-layer structure right now. But if you want to carve it out, this is the difficulty."

Su Yuanshan was amused when he heard this: "This kind of craftsmanship is handed over to the wafer fab."

Li Mingliu obviously didn't believe it: "It's really that simple? I think the wafer fab will hit people."

"Okay, it's not that simple." Su Yuanshan smiled and put down his pen, looking at the crowd in the office.

These are the top people of this era, and have been in related industries for so long. If you say that they would never have thought of developing the structure to multiple levels, it would be really funny. However, as Li Mingliu said, the structure from single-story to multiple levels, "from bungalows to buildings", involves not only design, but most importantly craftsmanship.

Just like building a house or building a bungalow, you can do it by calling a few or two masons at will. But if you want to build a skyscraper, you don’t have a professional construction management team and no solution to various problems. Even if you call a dozen or ten thousand masons, you can’t build it.

To create a new architecture, we must create new processes to cooperate with the implementation.

From channel to pile sinking, to p, n doping, etching and cleaning, ion implantation, oxidation and evaporation, vapor phase precipitation... After the entire process, there are at least dozens of steps.

With this set of processes, the technology may not need to be innovated, but the steps need to be redesigned. Not to mention, we have to face the difficult problems such as cell unit leakage and poor gate polysilicon adhesion.

Playing with engineering is not like playing with materials, just like playing with European Emperor. New crafts are all based on money, and time and loneliness can only be piled up.

...

After a few seconds of silence, Su Yuanshan slowly said.

"Design and craftsmanship are like two legs. Whether it is the birth of a new architecture or the breakthrough in the process, it will drive the entire industry forward."

Su Yuanshan pursed his lips and carefully considered the language: "But from the emergence of large-scale integrated circuits to the present, the development of semiconductors is mainly enjoying the dividends brought by process breakthroughs. Everyone can look at the process nodes and then calculate the number of transistors to know."

"But the process has its limits after all. In the foreseeable future, breakthroughs in the process will definitely become difficult. At that time, the improvement of architecture is the only way to save Moore's Law."

Memories flashed in front of Su Yuanshan. It was when the light source of the lithography machine stopped at duv193nm that the idea of ​​finfet (finfield effect transistor) appeared, the idea of ​​gaa (grid structure) also appeared, and the more radical EUV solution also appeared.

"Now we are not just behind the process. If we can't lead others in architecture... Then, how can we fight with others in the future?"

Su Yuanshan said and looked at everyone: "Everyone doesn't want to rely on VCD to support the business unit, right?"

This sentence made everyone feel indescribable.

Yuanxin’s main sources of income were two before, one was EDA and the other was VCD.

Then Su Yuanshan sold 20% of Xinghai's shares and acquired Cyrix, built Yuanxin Technology Park with the money he brought back. Then he sold more than half of EDA shares, took back more than 2 billion US dollars in cash, started building the Shanghai Stock Exchange Science and Technology Park and Xinde International, and also built the Silicon Valley R&D Center.

As for microcontrollers, yxcpu, architecture licensing, and various commercial chips... these are actually small money.

Especially for microcontrollers, in order to force the 91 microcontroller, Yuanxin's microcontroller license price is very low. Although it successfully entered the European market, the money earned is only enough to smooth out the expenses.

Therefore, in just two years, Yuanxin successfully turned the microcontroller to a bargain price with its own strength. The cost of many domestic electronic products, ranging from TV remote control panels to radios and recorders, has been reduced by dozens of times.

For this reason, Science and Technology Daily also published a special column article, which made unconcealed praise to Yuanxin, which was also a political gain.

Therefore, the current salary of Yuanxin employees is almost entirely due to the VCD business department. Last year, Yuanxin’s factory and R&D center were more than 10,000 people, and the salary of more than 500 million yuan came from the contribution of the VCD business department.

Even if Li Mingliu has been hailed by the outside world as the father of VCD, he has to admit that VCD... does not have much technical content.

It is really...

Very embarrassing.

...

After the meeting, Su Yuanshan stuffed the draft paper with his own thoughts on the table into the shredder.

"I have stayed here these days. Let's sort out my ideas with everyone and finally determine the direction. Once the direction is determined, let's bow our heads and fight to the death."

As he said that, Su Yuanshan looked at Jiang Tao again: "Senior Brother Jiang, the controller involves a lot of technologies and is also very complicated, but as long as you understand it, it is no problem to design any chip."

Jiang Tao nodded: "If you follow the direction you said, I think that the processor of the main control chip may have to use a dedicated processor like yx architecture to be sufficient."

Su Yuanshan said: "Well, that's the inevitable direction. Large-scale data throughput requires a high-speed processor. But now, the microcontroller is barely enough."

After dinner, Su Yuanshan returned to the hotel.

Originally, Jiang Tao planned to borrow a laptop from him, but even an international hotel like Hilton, which had not opened the Internet at this time, so he had to give up.

"Isn't it okay today?" Su Yuanshan saw Zhou Xiaohui take out his cell phone from his briefcase. When the technical discussion was held this afternoon, Zhou Xiaohui took his cell phone away from the conference room to avoid disturbing them.

"It's not a big deal, but the SAR said that the email office has completed the internal test."

"Oh?" Su Yuanshan became interested and sat upright: "What do you say from Senior Brother Ding?"

"I said I planned to go to the Science and Technology Park tomorrow and use Yuanxin's server to set up an internal email server, so that everyone can help me test it. If there is no problem, I can launch the market." Zhou Xiaohui paused and said, "I asked him to contact Mr. Xi directly."

"Yeah." Su Yuanshan nodded, and suddenly remembered something funny and asked, "Have he named the email office system?"

"It is said to be called 163." Zhou Xiaohui blinked and said, "Old Ding said that he had been inquiring for a long time before he found out that the civil dial-up network account that the State Administration of Telecommunications is setting up is 163. He plans to take advantage of it."

Zhou Xiaohui spread her hands: "He said the original words."

"Ha ha ha ha!"

Su Yuanshan laughed and picked up the buzzing phone.
Chapter completed!