In times of peace, there is no war, and the technological frontier is full of smoke.
Today, we return to the era of China's independent research and development of computer chips: Loongson manufactures "Loongson".
When we first touched the computer when we were very young, did you ever wonder as much as I did, why such a small black box could do so many things, how it worked, how was it created?
In the 1950s, the same question arose in New China.
In the 1950s, 10 years after the birth of the world's first computer, Hua Luogeng, then president of the Chinese Academy of Mathematical Sciences, formed the country's first electronic computing research team, with only 3 team members: Min Naida, Xia Peisu, and Wang Chuanying.
They start with sand and explore the mysteries of computers from the very bottom of logic.
In March 1952, Min Naida wrote the first "Computer Research Ideas and Plans" in mainland China;
By 1956, Xia Peisu had completed the design of the first computer computer and controller, wrote the prototype lecture notes of China's first electronic computer, and the era of China's independent manufacturing of computers began.
After decades of research, the mainland was able to manufacture specialized chips in the 1980s, but they were mainly used for scientific computing and military defense, lacking the general ability to run systems and watch video games, that is, the core technology in the field of computers, the development and manufacture of central processing unit CPUs, is still firmly in the hands of Britain, the United States, britain and the United States.
China's high-tech research and development plan, commonly known as the "863 Plan", is also committed to overcoming technical problems such as ID card chips.
In order to speed up the development of computer science and technology and reverse the situation of being controlled by others, Zhou Yulin, Xia Peisu and Jin Yilian, well-known frontier scientists who participated in the Xiangshan Science Conference in 1998, submitted their opinions to the central government on China's need to develop its own CPU.
In 1999, Li Guojie served as the chairman of the Chinese Computer Society, and Xia Peisu volunteered as his teacher, and delivered a speech advocating that the country begin to deploy independently developed CPUs.
In 2001, Li Guojie's student Hu Weiwu returned to Tsinghua University to give an admissions lecture, accidentally saw himself and the same door welding 8086, mixed feelings, heavy sense of mission, in his early 30s, Hu Weiwu was full of spirit, and finally made up his mind to pull up the team of self-developed CPUs, and said to his brother Li Guojie: "Don't come out, look!" ”
In May 2001, Hu Weiwu's team received $1 million in computer research funding to begin research on the FPGA prototype system; three months later, Hu Weiwu's team received good news: the FPGA prototype system was completed.
With the success of the FPGA prototype system, the possibility of completing the CPU's independent research and development is even greater. Hu Weiwu and the entire Chinese computer industry, in the era of foreign product technology as the mainstream, see the hope of taking the road of self-research.
The Chinese Academy of Sciences and the Chinese Computer Society each took the opportunity to invest $5 million each to support Hu Weiwu's research team, but in the case of almost all of the total blockade abroad from chip manufacturing materials and equipment to core technologies, few people believe that Hu Weiwu's team can really do this.
CPU, this strange English abbreviation, is it really that hard? What role does it play in the computer?
To be honest, it's really hard, it's really hard!
If computers are likened to people, then the CPU is undoubtedly the intelligent and powerful brain responsible for executing commands throughout the system, manipulating numbers and logic, storing and transmitting data, and so on.
First of all, the process of making CPU chips is extremely difficult on a physical level, and if we look closely at the surface of the CPU, we can see a maze of line nets, which in miniature of the world are high-rise buildings, which together create a micro-city.
The building material of the city is transistors, and the buildings are isolated from each other like a river channel, called the gate. The length of the gate is usually expressed in nanometers, such as 21 nanometers or 7 nanometers. When the gate is charged, electrons flow into the transistor building and the full logic switch, rather than water passing through the channel.
Typically, a simple microprocessor contains billions of transistors.
Second, just like building a city, before the actual start of construction, the architect draws architectural drawings, this process of conception, known as "design", the core of the CPU chip design is microarchitecture design.
When buying a computer, the core of the quad- or eight-core chip introduced by the salesman is the core of the basic unit that makes up the CPU. There are no grades between core units, and students with the same level of intelligence can be seen working together in a way that is called "microarchitecture".
Microarchitectures are designed to influence the highest frequency, capacity, and energy consumption levels that the core can achieve, so it is impossible for a company without independent microarchitecture R&D capabilities to have true CPU R&D capabilities.
Simply put, we can compare the development of the CPU to a city, where microarchitecture design determines the planning and layout of the city, the design of drainage systems, the planning of traffic routes, and civil works, the engineering capabilities considered, the building materials used, and how to cram as many buildings, designs, and processes into the CPU without sacrificing the width of the road.
Intel's Tick-Tock strategy is based on these two aspects, taking turns optimizing chips based on process and design each year.
Transistors, gates, microarchitectures, designs, processes, are difficult to pass, let alone manufactured, and Hu Weiwu has no way back.
On August 10, 2002, the phrase "If you can't do it, you can see it" rang in your ears, and Hu Weiwu's team pushed the chip called "Dragon Core One" forward and blocked yo-yo's mouth.
Although the performance of Loongson 1 is only as good as that of the Pentium 586 of the last century, China's history of relying on foreign CPUs is over.
For a time, the industry cheered, and a wave of independent development of CPU chips was set off in China, and the Institute of Computing of the Chinese Academy of Sciences, the University of National Defense Science and Technology, the Jiangnan Institute of Computing, Jiangbei and Jiangnan all spent a lot of money.
In 2003, the "863 plan includes the research and development of THE CPU, and the "national medium and long-term plan 01" focuses on the development of core electronic devices, high-end general-purpose chips and basic software products, that is, the "nuclear high-base" plan.
Gone are the days when HuWeiwu's Institute of Computing Technology of the Chinese Academy of Sciences received tens of millions of dollars in research grants to develop the $1 million FPGA and $10 million Loongson One.
Also in 2003, the team of Professor Chen Jin, dean of the School of Microelectronics at Shanghai Jiao Tong University, announced the "self-developed" "Hanxin No. 1" chip, which is only half the size of a fingernail cap, integrates 2.5 million devices, uses a 0.18-micron advanced semiconductor process design, and can perform 200 million operations per second.
At the press conference, the appraisal experts agreed that Hanxin No. 1 has reached the international advanced level, everyone believes that the milestone in the history of China's chip development has arrived, and Chen Jin's team and Hanxin No. 1 have also obtained hundreds of millions of scientific research funds, and have gained fame and fortune under Hanxin.
However, the good times did not last long, on January 17, 2006, the fake post of "Hanxin No. 1" was exposed and circulated in the Mizuki Tsinghua BBS of Tsinghua University, and a month later, the official investigation team exposed the cruel facts: Hanxin No. 1, pseudo-truth!
After purchasing the Motorola Fisker56800 chip in the United States, Chen Jin returned home, wiped off the Motorola pattern on the surface of the chip with sandpaper, and then found a foundry to print the words "Hanxin No. 1" and "Hanxin No. 1".
The big lie poured like cold water on the country's vibrant computer industry and was hit hard.
At the same time, Intel and AMD, the world's leaders in semiconductors, have also begun to produce multi-core chips with increasingly mature technology, which has greatly improved processor performance.
As mentioned earlier, the more kernels without a specific order, the greater the ability to handle computational logic tasks at the same time, often referred to as human.
But in the face of a power imbalance, a country's weightlifter may have been more powerful in his prime years than a group of children who had just started elementary school, but the obvious fact is that the local researchers at the time were dominated by academic ideas, which is difficult to understand.
"During the Eleventh Five-Year Plan period, the mainland relaxed the single-core performance and vigorously developed multi-core, and the problem was soon highlighted.
2006 is a high autumn year, but the ordinary processor Loongson 2E performance is 1-2 times the mainstream x86 processor, the next few years, under the guidance of the wrong thinking, the general processing performance of domestic and foreign chips compared with the iteration of 3b chips, there has been an order of magnitude improvement, but the performance has only increased by 50%.
With the exposure of the Hanxin international counterfeiting incident and the unsatisfactory performance of Loongson International, the domestic dissatisfaction with the development of independent research and development processors continues to heat up, and more and more people think, why independent research and development, we have money, others have products, is not it an efficient way of development?
Soon, as soon as Longxin put down the self-taught Jiangshan and was gradually eaten away by this sound, Hu Weiwu began to reflect, which step is wrong, and what is the root cause of this problem?
After a long period of thinking and hard work, Hu Weiwu decided to quit his gold job at the Institute of Computing Technology of the Chinese Academy of Sciences and lead the team into the enterprise.
In 2010, with the funding of the Chinese Academy of Sciences and the Beijing Municipal Government, Hu Weiwu founded Longxin Zhongke Technology Co., Ltd., and the development of Loongson has since been exposed to the cusp of the market, and is about to usher in the transformation and baptism from the inside out.