The recent U.S. sanctioned ZTE Incident will be written in the history of the Chinese nation’s great rejuvenation as a milestone in China’s modernization. It has caused the nation’s in-depth reflection on the development of its core technology industry and has actively taken action.
Forty years of reform and opening up, in order to achieve rapid development, we have bypassed some of the more difficult to be 'hills' on the way forward, including engines that are the heart of the industry, and the central processor of the 'heart' of the information industry (Central Processing Unit (referred to as CPU), high-end industrial control systems and high-end instruments, etc. Now that these uncaptured 'mountain heads' are chaos at the rear, they become a huge hidden danger. These 'mountain heads' have a common feature, that is, it takes a long time to capture it. Because they are all complex systems. There are many factors that affect the quality of complex systems. They need to be developed in constant trial and error. They cannot be accomplished overnight. Just like they go upstairs, they can only climb from the stairs one step at a time. Wrong. Foreign products have experienced many rounds of trial and error in decades of application. It is a step-by-step floor building along the stairs. Since the reform and opening up, our country has found it easy to obtain results for each industrial upgrading. Products can only be made using foreign components, systems, computers, automobiles, and airplanes. For self-developed products, 'under the same conditions First use 'is equivalent to never have to, because the process of blocking the self-developed products in the application of trial and error, trial and error in the spiral development of the rise.
The difficulty of core technology products is not based on scientific principles, but on the perfection of engineering details. Take the automotive engine as an example, the science principle textbooks of the engine are clearly written. The key is how to make mistakes in the application after trial and error. It is not difficult to build several cars based on autonomous engines to perform various experiments in the laboratory and in the field. It is difficult to verify the long-term application of certain batches. Some open to the thin plateau of the Qinghai-Tibet Plateau, and some open in the winter. To the low temperature of Mohe, some open to high temperature and high humidity in Hainan Island in the summer, and there must be a certain number of batches and cycles. A car without problems does not mean that the batch does not have a problem. A year without problems does not mean There is no problem for ten years. It is necessary to find problems in the application and continuously improve based on the problems found. Sometimes multiple rounds of improvement are required to be effective because the details of the complex system are too many, even if the circuit of the injector is controlled in the engine. Boards and chips are not simple. Just as there are a hundred short boards in a barrel, you need to change five rounds each time you change 20 pieces. The water in a barrel can increase, and the quality of a product can be significantly improved.
Trial and error takes time and patience, and time is the most effective threshold for core technology products. Take a radiation-resistant CPU as an example. After mastering the radiation-resistant technology and developing an anti-irradiation CPU, first perform multiple rounds of irradiation experiments on the ground. , Including total dose experiments, single-particle experiments, anti-latch experiments, and doing these experiments need to accelerate the ability of the CPU to resist irradiation by accelerating various particles including electrons, protons, and heavy ions on the chip. Ground experiments can only be carried on for the first time in heaven instead of real application. After the experiment is passed, it can be used in the backup system. The backup system can be used as a master without any problem. The time for each step above needs to be years. Units. Godson started developing anti-irradiation CPUs in 2006. The first time it was launched in 2015, it started on small batches in 2017. It took more than a decade before and after. What social capital is willing to support the entrepreneurial team for more than ten years? Not profitable? But once you step into the threshold of radiation-resistant chip applications, others would want to come in again for many years. Therefore, time is the most important threshold for core technology products. Any one, two years or Three years will be able to do things out of the threshold is not high, the threshold is not high product profits are not high.
Is there any way to avoid the above-mentioned time-consuming trials and errors, and instead of climbing stairs, we go upstairs? Since the reform and opening up, we have made various attempts. The first stage is not as good as buying, not as good as rent; the second stage is 'Market for technology', hopes to master core technologies through the establishment of joint ventures; the third phase is the direct acquisition of foreign high-tech companies. Facts have proved that these are not fundamental solutions. The total profit of China's top 100 information industry enterprises in 2011 is 40% of American Apple. In 2012, Apple and Samsung accounted for 97% of global mobile phone profits. In 2016, Apple and Samsung accounted for 94% of global mobile phone profits. The logic behind this is simple, that is, the West is not Will really help us to develop. On the contrary, they will set various thresholds to prevent us from entering their 'territory', including this time the United States sanctioned ZTE and waved sanctions against Huawei's 'stick'. In the application of trial and error is the core The inherent law of the development of the technology industry, we can accelerate the trial and error process by increasing input and improving institutional mechanisms, but it cannot replace the trial and error process. Only insist on self-reliance. Rehabilitation, long-term adherence, China's core technology industry can 'upstairs', can not be controlled by others.
Second, the development process of autonomous CPU in trial and error
Since the 18th National Congress of the People's Republic of China, significant progress has been made in the development and application of autonomous CPUs represented by the Loongson CPU. First, CPU performance has continuously improved, surpassing the low-end products of international mainstream CPUs, and is approaching mid-to-high end. It is expected to approach the international mainstream around 2020. The performance of the CPU is 'ceiling'. The second is based on autonomous CPUs that form an industrial chain comprising nearly 1,000 companies. An autonomous and controllable information industry system is emerging. The third is a large number of areas in national security, party and government offices, energy, and transportation. Applications, the use of autonomous CPUs in areas involving national security and national economic security has become a consensus. The autonomous CPUs can achieve the above-mentioned progress, and are mainly obtained the opportunity of application and trial and error, and they have several levels of 'staircase' in trial and error.
Loongson's first class 'staircase' is during the 'Twelfth Five-Year' period. It focuses on single applications based on embedded operating systems and simple applications based on Linux general operating systems. The autonomous CPU starts with the operating system and is resolved during the integration process. Hundreds of technical problems, including inadequate CPU maturity, lack of various board support packages (BSPs) and graphics packages for embedded operating system VxWorks, no available Java virtual machines on Linux, Browser, office software and database, etc. Through the establishment of quality system and service system, and obtained the support of industrial chain partners, the above problems were solved initially, and the independent basic hardware and software reached 'essentially available'. In 2015, the Godson industrialization main body was Godson. Branch company achieves profit and loss balance through market sales, and on this basis realizes rapid growth of profits. The main applications of this stage are all 'small chimneys', each application is different in hardware and software, requires specialized technical services, and bulk Not so much. This kind of application is like salty and alkaline land. It can be seriously planted and can be hungry. It is deeply monopolized by foreign monopoly groups in the IT industry. Under the circumstances, Godson can only plant 'saline' as a base for further development. In the first round of trial and error, we also found that CPU performance is insufficient (only one-tenth of the mainstream products in the market), and some software features (such as online Flash video Lack of play, inadequate input/output device adaptation, etc. Overall, the main bottlenecks in the first level of 'staircase' are the CPU, the operating system and its integration department.
Loongson's second-stage 'staircase' is the first two years of 'Thirteen Five'. It is mainly for complex fixed applications including party and government offices and industry business systems. Each application scenario has thousands of computers, involving operating systems, and databases. , Middleware, Browsers, Office Software, Flash Video, Geographic Information Systems, etc. Although complex, but with boundaries. In this process, the CPU is upgraded to the second generation, performance is greatly improved to 4-5 times that of the first generation products. The operating system is upgraded to 64-bit, and the stability has been greatly improved. The performance of the Java virtual machine and browser has been improved to 2-3 times that of the first-generation products. The online Flash video playback and other issues that restrict desktop applications have been resolved. / Output devices such as Gao Paiyi, scanners, printers, ID card readers and other companies actively join the independent industry chain, and input/output devices are becoming more and more abundant. CPUs, operating systems, databases, entire machines, system integration vendors, etc. Collaborate to solve the problems found in the user's pilot process and form a 'application pilot, identify problems, solve problems and improve the platform, test the virtuous circle in the pilot. The application has reached the ' The level of user resistance to autonomous infrastructure hardware and software gradually disappeared. Newly discovered problems include: For complex external network applications, lack of instant messaging software such as WeChat and QQ; For large-scale applications, poor scalability Database becomes a performance bottleneck; solutions for emerging IT architectures such as cloud computing also need to strengthen R&D, etc. Overall, climbing to the second level of 'staircase' has greatly eased the bottleneck of basic software and hardware, and the main contradiction started from the CPU and operating system. The joint department turned to the combination of operating system and application.
Godson climbs the third level 'stairs' in the latter half of the 'Thirteen Five' period. The application oriented has full-service, all-terrain features. The autonomous base hardware and software will mature after the third round of trial and error, with comprehensive promotion. Conditions. Godson will also improve the basic hardware and software in the third stage of the 'staircase' process. First, the CPU's general processing performance will increase by 1-2 times, approaching the 'ceiling' of the international mainstream CPU. By optimizing the processor Microstructure improves pipeline efficiency, and increases the frequency through physical design. It is expected that the performance of Loongson CPU will approach the level of mainstream CPUs of the market by 2020. The second is to improve the independent basic hardware and software specifications, and to achieve stable system architecture and convergence of the technology platform. In the middle, the operating system needs to be suitable for different motherboards and upgraded CPUs, and in the Wintel (Intel Corporation's X86 processor and Microsoft's Windows operating system) system can achieve binary compatibility of different motherboards and CPU operating systems Behind it is a unified system architecture of Wintel system, including instruction system, address space layout, interrupt system, multi-core interconnection architecture, IO interface. Fan et al. Require CPU, BIOS, bridge, and operating systems to complete. In contrast, the ARM architecture is fragmented because it only regulates the instruction system and the on-chip bus. Loongson is improving based on the Loongson CPU and bridge. The unified system architecture of the chip greatly reduces the workload of hardware and software adaptation, and ultimately realizes the binary compatibility of the operating system between different motherboards and CPUs. Third, the user experience is the center, and the independent basic hardware and software development system is combed and optimized. The intensive system allows the user experience to be substantially improved to achieve a 'good for use' level. After this, the main bottlenecks in the autonomous base software and hardware will be transferred from the combination of the CPU and the operating system to the combined parts of the operating system and applications.
After the above three rounds of trial and error, the independent basic hardware and software represented by the CPU and operating system are from immature to mature, the independent basic hardware and software industry chain from combinational divergence to portfolio convergence, and the application system based on independent basic hardware and software is basically available. To be usable, easy to use, lay a solid foundation for building independent technical systems independent of Wintel and ARM+Android.
Third, the development of core technology industry must adhere to self-reliance
In the process of 'climbing stairs' in the core technology industry, we must promote the spirit of seeking truth from facts and the spirit of Yugong to move the mountains. Not only must we raise our sleeves to get dry, but we must also persist in doing so. The development and improvement of complex systems often takes thirty years. The "Yunji" aircraft was dismantled in 1986 and the C919 first flew in 2017. Many of these modules were still owned by others. After the collapse of the former Soviet Union, the Ulysanovsk aircraft carrier under construction was dismantled in 1992 and newly built in Russia. The aircraft carrier plans to enter service in 2030. Time has become the most important factor that restricts the development of China's core technology industry. Improvement is the necessary process of complex system innovation in continuous trial and error. Good institutional mechanisms and more funds can accelerate trial and error. Iterative, but it can not replace the trial and error iterative process. In the accumulation of core technology capabilities, only step by step to catch up, do not imagine 'curve overtaking'. Curved road can not exceed the car, bend overtaking is to overturn. Impetuous will let Our industrial development is in the wrong zone, but it is developing more slowly. Industrial development has its own laws. If we do not follow the rules, investing more money will also be affected by the law. Penalties. The use of self-developed products may not be satisfactory at the beginning, and there will be a more problematic phase. However, as long as it is used consistently, there will be 2-3 rounds of continuous improvement, which will lead to fewer problems. Stage, mature, even exceed foreign systems.
In the process of 'climbing stairs' in the core technology industry, it is impossible to maintain strategic strength and cannot hold unrealistic illusions against Western countries. In the modern history of China, we have three experiences of learning from foreign teachers, each time After the bully had suffered a loss, the teacher took the road of self-reliance. The first time was the Westernization Movement. In the late Qing Dynasty, the Commissioner of the Chinese Customs was the Englishman. It was the most open period in Chinese history, but it was bullied by Western powers. Chairman Mao led the Communist Party to overthrow the three mountains and liberate the Chinese people. The second time was after the founding of the country, it learned from the Soviet Union and fell on one side. As a result, the Soviet Union wanted to build a joint fleet on the Chinese soil, and wanted to build a long-wave radio station. Chairman Mao refused. The Soviet Union again withdrew its experts and forced the debt. Chairman Mao led the Chinese people themselves to engage in 'two bullets and one star' and established an autonomous and controllable industrial system. The third time was reform and opening up. We learned Western management concepts and market economy. At first, I thought it was good. Now the teacher has begun to bully the students. Our only way is self-reliance.
We must beware of the Western sugar monopoly group's “sugar cannonball” in the process of 'climbing stairs' in our core technology industry. Blocking technology when we do not have the technology, and preventing technological development through technological cooperation after we master the technology is a foreign monopoly group. The usual trick is that you can't make independent research and development when you commit foreign blockade technology. Once you open up technology abroad, you stop the historical error of independent research and development. There have been experts dubiously saying that if the United States embargoes China for five years and let it go for five years, what will we do? Can not develop. Recently, with the continuous development of independent CPU research and development and application promotion in China, some of us want to get rid of the dependent CPUs and operating systems through independent research and development, and have us build new ones through technical cooperation or joint ventures. Reliance on this. The ZTE event should allow us to completely abandon the illusion and insist on independent research and development.
Although the ZTE event has a great pressure on China's electronic information industry in the short term, the long-term advantage outweighs the disadvantages. The ZTE event once again confirmed the assertion that General Secretary Xi Jinping 'cannot hold any illusions in the introduction of high technology', and will allow us to completely abandon the illusion. , based on independent research and development. We independently developed the BeiDou satellite because Western countries refused to participate in the European Galileo positioning system. We also developed the 'Tiangong' series space station because the US-led international space station refused to participate in China. Western countries have long implemented military components for China. In the embargo, China's military components can be independently developed, and many components of the electronic system have achieved 100% autonomy. For 'Made in China 2025', early sanctions are better than later ones. Now that sanctions are available, we can Adjust the technical route, use self-developed components, and give the opportunity of trial and error of the autonomous components represented by the CPU in the application. If our industrial control system is fully automated and intelligently sanctioned by 2025, it will not be beneficial. , Now sanctions, is beneficial to 'Made in China 2025'.
Fourth, conclusion
The core technology is the country's most important device. The development of core technology industries is the country's needs, the needs of the times, and we have preliminary conditions, and there is no reason to doubt that we can do this thing. As long as we overcome the irritable mood, and promote the style of seeking truth from facts and The spirit of Yu Gongyi is to maintain a clear-headed mind before accomplishments, and to resolutely win the confidence of victory in the face of difficulties, be down-to-earth, and be proactive, and we will steadily reach our own goals with our unyielding efforts.
Author: Chief Scientist Weiwu Godson CPU Godson, president of Branch This switched "Chinese Science" magazine text the following address: http://engine.scichina.com/doi/10.1360/N112018-00111