Compatible and controllable domestic CPU seeks to change

Recently, the discussion on how to develop domestic CPU has been very enthusiastic. The main topics involved are two aspects: First, the unified issue of domestic CPU instruction set and architecture, and the second is about the ecological environment of domestic CPU. As an important core role for the national information security in the information society and a product with a major strategic economic value, the CPU is an important area that the country must attach importance to and strongly encourage development, whether it is from the national security strategy or from the economic development needs. How to develop domestic CPUs, whether or not to unify the domestic CPU instruction set and architecture, etc., may not only be the problem of whether the domestic CPU instruction set and architecture are unified, but how to seize the current development opportunities and better realize the rapid development of the domestic CPU industry. Soaring question.

Entering a new era of crowds and hegemony

In light of China's national conditions, in the face of the more advanced stage of social information development in the future, in the face of increasingly complex international competition, should our country build its own information security and economic security on the basis of relying on foreign technology? China's next step is to transform from a manufacturing power to a strong power, and to develop high-end manufacturing. If we bypass the CPU industry with the highest technological content, huge economic benefits, and broad market application prospects in the world today, China can still be regarded as the manufacturing industry. Has a big country changed to a strong country? These do not need much reason. We know how to choose. This is also the basic position that we must adhere to in order to solve the long-standing and difficult problems of the CPU.

He Xiaoqing, deputy secretary-general of the China Software Industry Association’s Embedded Systems Branch, told the “China Electronics News” reporter that the development of domestic CPUs is imperative and should be started from three levels: First, the government and the community should support universities and research institutions in the CPU. Innovations in architecture, instruction set, bus, IP, OS, and software tool technology are encouraged, and hundreds of flowers and hundreds of schools of thought are encouraged. Support schools and scientific research institutions to carry out technology transfer through intellectual property rights, and cooperate with the industry to encourage entrepreneurship and venture capital investment. Second, in the fields of aerospace and other military applications and national information security, domestic CPUs should still be autonomous and controllable. Open source software and open standards should be used to strive to establish an application-level interoperable ecological environment. Third, in the application of market-oriented industries, domestic CPUs should move closer to the international mainstream three major CPU architectures in order to achieve consistency and compatibility. This is to maintain a level with foreign competitors. At present, ARM architecture has absolute advantages in mobile terminals; Intel architecture has obvious advantages in PCs and servers; MIPS also has some advantages in digital home and network communications, but it is being eaten by ARM. POWER architecture is only based on communications. Equipment and automotive electronics have certain advantages. In a wide range of embedded applications, the 32-bit MCU chip ARM has obvious advantages, and the rising trend is fast. The 8/16-bit architecture is different, and the actual shipment is still the largest, and domestic MCUs can focus on development in this field.

From a historical point of view, our foundation in CPU technology is too weak. In addition to the strong rule of the Win-Tel system, domestic CPUs with independent intellectual property rights may be very difficult. However, now and in the future, these disadvantages seem to be rapidly disappearing and changing, providing us with an excellent opportunity to enter. First, China has basically achieved a complete breakthrough in CPU technology. At present, there are at least 20 companies that have mastered the CPU design technology. Second, the market demand trend is changing. The Win-Tel system has begun to loosen and disintegrate. Intel is subject to ARM and many others. With the severe challenges of CPU manufacturers, the world has entered a new era of CPU hegemony. Although the main battlefield of this CPU hegemony is apparently applied in applications, it essentially uses the CPU as the core and the basic platform. Whoever has superior CPU performance will have a higher starting point than others and will be in a favorable position in the competition.

To be compatible and autonomous

The field of CPU is a rapidly developing field. If China's autonomous CPU wants to gain market recognition, compatibility or inevitable choice. However, if we continue to develop the target lock only by being compatible with international development, we will spend a lot of effort on research and development of “intellectual property rights” that have long been mature in technology and market, and we are trying to come up with a “compatible” path. A R&D route that does not have a high starting point and is in violation of the spirit of innovation is more important in compatibility and control.

According to Gu Wenjun, chief analyst of IHS iSuppli Semiconductor, the thinking of China's development of domestic CPU must be "internationally compatible, autonomous and controllable." If it is not internationally compatible, it will easily be marginalized. There will be great difficulties in the construction and development of the ecosystem. If it is not self-controllable, it will not be able to control the country’s security, product safety, and so on. In the hands of foreign companies, the significance of domestic CPUs is lost. In addition, the choice of architecture should be based on choices that cannot be marginalized or strongest, because it is difficult to control. People will not customize the architecture for Chinese companies, nor will they be “controlled” by China. After selecting this kind of structure, which is a large-scale application in the international market and not the most powerful one, it can be controlled or partially controlled by some means of capital. Then, in this framework, companies on the ecosystem are encouraged to become bigger and stronger, and domestic CPUs will be "internationally compatible and autonomously controllable."

For the compatibility of domestic CPUs and international popular CPUs, in principle, they can be implemented in software according to the “conversion table”, and there will not be too much work. Therefore, it is untenable to worry too much about the incompatibility of domestically produced CPU instructions and the incompatibility of various types of CPUs currently on the market. New applications need to define new CPU instructions and research new architectures. This is a common development opportunity. In fact, the instruction set of the CPU is continuously developed in the inheritance, presenting the characteristics of a generation of products and a generation of instruction sets. It synchronizes with Moore's Law and averages 2 to 3 years ago as a technical step. Taking Intel's CPU development as an example, it is exploring while walking. Therefore, to dynamically view and understand the development of CPU instructions and instruction sets, CPU instructions have evolved with the needs of applications and advances in technology.

Application-oriented construction of ecological environment

Faced with the actual needs of the Internet of Things, big data processing, and the new generation of the Internet in the future, the development of domestic CPUs also needs to be combined with applications.

"In today's mature chip design technology, the open ecological environment is helping Chinese companies develop on domestic CPUs, but whether domestic CPUs can become the mainstream of China's electronic design industry depends on application." He Xiaoqing pointed out, " Such as the consumer electronics industry, tablet PCs, smart appliances, embedded appliances, high-performance and low-power servers, surveillance, network security and satellite positioning navigation products, etc., have potential for development."

Gu Wenjun believes that developing domestic CPUs in the new situation must be supported by the government, led by the market, and led by enterprises. Because the difficulty of doing CPU under the new situation is still very great, there are difficulties in the ecosystem, software cooperation and the difficulty of the chip. CPU is not a product issue, but is a problem of the entire ecosystem. Therefore, without the support of the government, it is difficult for the upstream and downstream industrial chain of the ecosystem to cooperate with each other, and it is impossible to form a national chess game.

“There is also a key issue is the market leads, detailed market planning, and understanding of the development path of good products and markets.” Gu Wenjun stressed that “through the market to drive the definition of products, drive the planning of the CPU so that we can serve the market well.” In order to do a good job of CPU, the most important thing is that the enterprise is leading and the enterprise is fighting at the forefront of the market, and those companies with hardships, industrialization capabilities, and real sales performance must be able to lead the development of domestic CPU under the new situation. ”

"In addition, the government and all sectors of the community should support innovation and development of domestic CPUs that are application-oriented, traction-driven, use the SoC and the open software ecosystem, and strive for breakthroughs in the application markets. They must be market-based operations," said He Xiaoqing.