The rise of nanometers

September 27th.

Dalian Exchange frequently makes moves in the international ore market and invites mining companies from many regions to list here.

And on the other side of the Bohai Bay.

Texas Semiconductor Base.

As soon as Huang Xiuyuan read the relevant report from Dalian Exchange, the phone rang.

"Hey! Xuedong, what's going on?"

On the other side of the phone, Lu Xuedong's voice was slightly excited: "Xiuyuan, it's done. The Fuxi chip design is completed. I have performed supercomputer simulation calculations, and the first draft is ready for tape-out."

"Okay, arrange the tape-out as soon as possible." Huang Xiuyuan also showed rare emotional fluctuations. After a year and a half, Suiren Department invested no less than a billion in R\u0026D funds and finally completed the first draft of the Fuxi CPU design.

"Jingjian and I are already preparing for the tape-out."

"Thank you for your hard work. Take care of your health and call me if you need anything."

"Don't worry, I'm in good health. How is the situation over there?" Lu Xuedong answered casually and began to change the topic.

Huang Xiuyuan shook his head. He had ordered Huang Weichang to keep an eye on the health of Lu Xuedong and others. If they were overworked, he would immediately force them to rest.

He summed it up: "All kinds of supporting chips and electronic components can basically be mass-produced on a large scale."

The new semiconductor system under the leadership of Suiren Department adopts the R\u0026D idea of ​​triple preparation of low, medium and high in various supporting chips, professional chips and electronic components.

For example, in terms of storage chips, Ziguang Group focuses on high-end storage chips, while Yangtze Memory focuses on mid- and low-end storage chips.

Tsinghua Unigroup's storage chips are now in difficulty, but the tape-out of Yangtze Memory's low-end storage chips at Texas Semiconductor is very good. After several months of improvements, they are basically ready for large-scale mass production.

It is this triple combination of low, medium and high that ensures that even if high-end products are unsuccessful, there will always be mid- to low-end products to replace them.

The current pattern of domestic semiconductor products is not to blindly pursue high-end products, but to solve the problem of availability.

Even if the performance is slightly worse than foreign products, as long as it can be used, the feeling will not be too obvious.

Especially for many ordinary products, especially consumer electronics and other high-end and low-end spare parts, sometimes the feeling is not too obvious during use.

After hanging up the phone, Huang Xiuyuan called Xiao Yingnan over.

"Chairman, this is Mr. Zhang's report."

Huang Xiuyuan took it and browsed it quickly. There were 16 types of supporting chips for mobile phones and computers, as well as dozens of related electronic components.

Each type of chip and electronic component has detailed real-time progress. Usually, each type of spare parts has two or three manufacturers.

The progress of those low-end spare parts has all entered the mass production stage; among the mid-range spare parts, 41% have entered mass production, 35% are in debugging and optimization, and 24% are in design and research and development.

Regarding the final high-end spare parts, the proportion that enters large-scale mass production is about 13%; the proportion of debugging and optimization is about 33%; and the remaining 54% is in design and research and development.

I quickly checked two sets of spare parts on the parts list, one with all low-end parts and one with mid- to low-end parts.

These are mainly on mobile phones.

As for computers, Fuxi's mildly complex instruction CPU is specially designed for computers.

The so-called mildly complex instruction set CPU is a research and development idea established when Huang Xiuyuan formulated the Fuxi architecture.

The complex instruction set under the Microsoft system was abandoned by Huang Xiuyuan, and instead he chose an idea between a simple instruction set and a complex instruction set.

This is the background of the birth of the mildly complex instruction set. The architecture derived from this instruction set is Fuxi-Sixty-four Hexagrams.

Fuxi-Sixty-Four Hexagrams can be pieced together using Fuxi-Bagua structure chips.

In fact, it is a multi-threaded multi-core mode.

The Fuxi chip currently being taped out is designed based on the Fuxi-Bagua architecture. At the same time, the Fuxi chip can be combined and deployed to become a chip for computers.

According to the initial design plan, the dual-core version of the Fuxi chip is a chip for computers; while the single-core version is for mobile phones.

This design direction is actually to reduce the time of developing computer chips and achieve the purpose of one core for multiple purposes.

In fact, this is also the direction of the future. As chips become more and more integrated and the number of integrated transistors increases, the boundaries between computer chips and mobile phone chips will become increasingly blurred.

In addition, the integration between CPU and GPU is also a major trend.

If the time was not too short, Long Totem would definitely design a chip integrating CPU and GPU. This kind of chip has been tried by Intel and ARM.

Among the supporting chips at this stage, the design of the GPU, which is second only to the CPU, has not yet been completed. This project is not the responsibility of Dragon Totem alone, but is a collaboration between Dragon Totem, Huawei, and Ziguang to jointly develop the GPU.

Huang Xiuyuan checked the development progress of United GPU and found that only about 85% was completed. According to the report of the project leader, the first draft design is expected to be completed by the end of this year.

This conflicts with his plan to launch Dragon Totem mobile phones and computers in March next year.

However, sometimes things about R\u0026D work cannot be forced. He picked up the phone and called Zhang Weixin.

"Hey, Wei Xin."

"Chairman, what are your orders?"

"The progress of the joint GPU is somewhat behind. Go and investigate, increase support, and transfer some researchers from the CPU project team."

"clear."

hang up the phone.

Huang Xiuyuan stood up and stretched.

At present, chip research and development has entered the sprint stage. The use of 22nm Fuxi CPU will overtake Intel's 32nm Core i7 in one fell swoop. Even the so-called 28nm from TSMC and Samsung will not be an opponent of Fuxi CPU.

What's more, Huang Xiuyuan knows TSMC and Samsung. The 28nm processes of these two companies are somewhat watery.

From the perspective of the future, although the processing technologies of TSMC and Samsung have surpassed their big brother Intel after 12 nanometers, their chip technology has the problem of secretly changing concepts.

This is why Intel has been using the mature 12nm, which can still compete with their so-called "7nm" and "5nm", because there is a lot of water in it.

This is also the fundamental reason why Huang Xiuyuan acted very calmly when he heard the news about the process breakthroughs of TSMC and Samsung.

After all, you can say whatever you want on the PPT, and outsiders can’t tell the difference between the craftsmanship and Samsung claims that it has a 28-nanometer process. Only Samsung itself knows whether it is true or false.

He called Zhang Lei over.

"Zhang Lei, please arrange your trip. I will go back to Shanmei in a few days."

"Understood." Zhang Lei nodded.

Huang Xiuyuan was the only one left in the office, and he made reasonable arrangements for Texas Semiconductor's affairs.

The work here has actually been completed under his arrangement. Although not all of it meets expectations, at least there are products that can be used.

The only thing he is worried about now is the tape-out of Fuxi CPU and the design and development of joint GPU.

Therefore, he planned to go back to Shanmei in person to inspect the actual situation of Fuxi CPU and United GPU, and to check and fill in the gaps.

Zhang Lei adjusted his schedule within a short time and returned to Shantou and Meizhou on October 4.