Organize | Zheng Liyuan
Listing | CSDN (ID: CSDNnews)
"Only when the roots are deep can the leaves flourish, and only when the roots are strong can the branches flourish." With the rapid development of a new round of technological revolution and industrial transformation, more and more people are beginning to realize that the software industry is like a towering tree: branches and leaves are flourishing. Mao's upper-level applications require the strong development of basic software under the earth-the concept of "root technology" was born from this.
The so-called root technology is a technology that can derive and support one or more technology clusters, and can continuously provide nourishment for the entire technology tree, including operating systems, databases, middleware, and programming languages:
The year before last, we invited representatives of the nine major operating systems to gather together to discuss the difficult survival path of the operating system and debate the future development of the operating system;
Last year, the top ten database leaders gathered in Changsha to look at the future from history and focus on the present to discuss the best solutions for databases.
This year, in the face of the unpredictable international situation and the iterative changes of new technologies, "root technology" has gradually become the focus of the global software industry, and has increasingly become the main battlefield for a new round of competition for the commanding heights of science and technology. In this regard, we can't help but ask: Where is China's basic software and hardware industry currently? How should we promote the sustainable development of root technology to truly "break the game"?
At the Yuelu Dialogue of the 2022 Changsha·China 1024 Programmers Festival and the 2022 Technical Heroes 100 Meeting, CSDN specially invited the leaders of important industries that built the root community to gather together. Wu Qingbo, Director of the Domestic Basic Software Engineering Research Center of the Ministry of Education, and Tsinghua University Under the chairmanship of Chen Yu , a permanent associate professor , the discussion focused on AI computing, chips, operating systems, databases, talent training and other dimensions to discuss the path to breakthrough for China's core industries .
The dialogue guests are as follows:
Chen Haibo Chairman of OpenHarmony Technical Steering Committee
Yang Zhenkun, Chief Scientist, OceanBase
Bao Yungang, deputy director of the Institute of Computing Technology, Chinese Academy of Sciences
Zhou Binhua , CTO of Ascend Computing Business
Kong Jinzhu Senior Deputy General Manager of Qilin Software
Chen Songzheng Senior Vice President of Hunan Qilin Security
host:
Wu Qingbo, Director of Domestic Basic Software Engineering Research Center, Ministry of Education
Chen Yu, permanent associate professor at Tsinghua University
Where will China's root technology go?
Wu Qingbo: China's 5,000-year history has inherited a lot of traditional culture, but there still seems to be a certain gap between us in terms of computing power and chips. So, what is the current state of root technology in our country ?
Zhou Bin : In the field of AI, we have a lot of root technologies. In addition to basic processors, programming languages, compilers, various acceleration libraries and AI frameworks are all root technologies. Huawei has also made many contributions to AI root technology, including: various underlying computing systems, ranging from mobile phone devices to large-scale computing centers, servers and edge devices, and a series of basic software and hardware platforms; it has open sourced its self-developed AI framework MindSpore; there are also some acceleration libraries and industry application enablement suites, etc. We open source a lot of these research results, because these are the indispensable root technologies in building a prosperous AI ecosystem .
Kong Jinzhu : I feel that compared with previous years, China's "lack of core and lack of soul" has been improved to a certain extent, but the degree is still not satisfactory. Two days ago I watched a game called "Nine Yin Manual". Martial arts is divided into 24 levels: the first level is called "vulnerable", which is probably what we were like 30 years ago, and the second level is called "vulnerable". The third level is called "beginner's first practice", the third level is called "first glimpse of the door", the fourth level is called "slight success", the fifth level is called "familiar with ease", followed by "unparalleled", "outstanding among the others" and so on. I think that we have gone from being "vulnerable" twenty or thirty years ago to being "slightly successful" now, but we have not yet reached the stage of being "familiar", so we still have to work hard .
Yang Zhenkun : After decades of development, China's database has also reached the stage of "initial glimpses" . It has accumulated some basic technology, but there is still a certain gap between it and the world's first-class level. However, this also has a lot of room for development and opportunity.
Bao Yungang : Looking back on the development of the past two decades, we have made great progress, just because the entire international technology is developing, and while we are making progress, others are also making progress, so it seems that we are always catching up.
In the field of chips, my country has made many contributions along two lines in the past two decades:
The first is the high-speed rail model , which is a technical route that first introduces advanced technology, and then through digestion, absorption, and innovation, gradually becomes a Chinese technology brand, such as HiSilicon, Feiteng, etc.;
The second is the Beidou model , which is to independently build a technical system and ecology. It takes a long time, but it has also achieved a lot of success, such as Loongson, Shenwei, etc.;
In recent years, a third route has gradually emerged: the 5G model , which involves integrating open international standards, participating in international competition and cooperation, and mobilizing the enthusiasm of domestic enterprises . Currently, we are exploring this model through open source chips, which can also form a good centripetal force to promote ecological construction.
These three models are not about replacing the other. In the current situation where China's processor chip industry needs "saturation rescue", these three models are all important .
Chen Haibo : The chip industry represents a very long industrial chain of the entire electronics industry, including raw materials, instruction sets, architecture design, and related packaging and testing. Overall, the demand for chips in all industries is very strong, but the development of the chip industry still has problems of insufficient and unbalanced development . How to resolve the contradiction between the two may be what we need to focus on.
With the development of root technology, what are the obstacles to the construction of a new ecology?
Wu Qingbo: There are many root communities for domestic operating systems. Is this a problem?
Chen Songzheng : I have been engaged in operating systems for more than 20 years. I am personally optimistic about this: I believe that China’s operating systems will definitely master root technology in the future.
We believe that the root technology of the operating system is not strong, and the fundamental reason lies in dominance and productivity . Why do we feel unsafe? Because we do not have a leading role in the field of operating systems; why do we think that domestic operating systems are not easy to use? Because compared with Windows, the productivity of domestic operating systems is weak. But if you turn your attention to Linux-based server operating systems, you will find that our international standards are comparable.
As for the root community of the operating system, I think there are not many. We cannot artificially decide which one should be eliminated . This elimination should be determined by the market .
Chen Haibo : Multi-rooting is indeed a problem that needs to be considered, but what we need to consider more is how to promote effective collaboration between roots, and even include the sharing of some related capabilities , such as how the two root communities of OpenHarmony and openEuler are capable. Sharing and ecological interoperability.
Considering that there are actually many "hybrid" devices in reality, that is, combining many functions with other device functions to incubate a new type of device, this requires different operating system root communities to combine with each other to form composite capabilities.
Chen Yu: The development of domestic software and hardware is very hot, and this may also cause another problem. For example, RISC-V's current goals are very large, involving desktops, servers and even supercomputers. But if the goals are too big, does it mean that it is not focused and will affect ecological development?
Bao Yungang : I think the first thing to do is whether it is driven by real demand. Real needs do not require policy guidance. All enterprises are using RISC-V voluntarily. This is also in line with the current trend of the Internet of Things and the Internet of Everything. The Internet of Things itself will cover many fields, so it is understandable that RISC-V is popular in mid-to-low-end application scenarios .
Secondly, it is not that simple to develop RISC-V to the high end . In fact, many companies have plans, but they have not yet invested heavily. In this regard, it actually requires multiple enterprises to join forces to enhance cohesion and synergy.
Chen Yu: If a new interconnected ecosystem is to be formed, can it be understood as forming a unified standard in which everyone seeks common ground while reserving differences? Is this possible?
Kong Jinzhu : I was just thinking about what a root community is. what is root It is actually the vast number of talents who have mastered root technology, and the community is connected through talents. For example, when Kirin launched a desktop system, it also united more than a dozen domestic operating system manufacturers to build it together. I think the future must be a blue ocean of human-machine-thing interconnection and ubiquitous computing.
However, one thing needs to be made clear. At present, except for the open source communities in which major manufacturers participate, one of the main reasons why other communities in China are difficult to sustain is that there is no source of funds.
Zhou Bin : Whether it is open source OpenHarmony, openEuler or MindSpore, it is our firm attitude to dedicate the open source community to society. We are also practicing this concept, not only for the Chinese open source community, but also for the world.
But there is currently a source problem that needs to be solved urgently: at this stage, we have very high requirements for open source and community, but without a good, organized community support, people will have insufficient motivation and resources to do this voluntarily. Therefore, we need to call for contributions to the open source community, and strive to make more contributions to the open source community from the system, including enterprises, governments, and various communities, and support the development of the open source community in a more benign direction .
Focusing on the contemporary era, how to promote talent cultivation in root technology?
Wu Qingbo: Now that we have talked about human issues, and it coincides with the 1024 Programmer's Day, teachers, would you like to talk about your experience or suggestions in cultivating root technical talents?
Yang Zhenkun : In terms of quantity, China's open source talents account for a large proportion of the world, but in fact the value we play is out of proportion to the number of people. For example, many domestic databases are still based on MySQL or PostgreSQL. This means that although many people are in this field, their mastery of root technology is still very limited . In this regard, I also want to ask a question: How can the efforts of these people be transformed into real root technology?
Bao Yungang : We conducted a survey in 2017 and counted the distribution of first authors of ISCA, the top international conference in the field of architecture, from 2008 to 2017. It was found that, in terms of nationality, the United States accounted for 25%, and China also accounted for about 20%, that is, there is not a big gap in the number of talents between China and the United States. But after graduation, 85% of people work in the United States, and China only accounts for 4%, which also leads to China's lack of talents in the field of chip architecture.
Essentially, in recent years, the overall market for chip talents in the national industry has not increased. As a result, many start-up companies can only recruit people from other places by raising wages. The cost of the entire industry has increased very sharply. In the long run, this is not good for the development of an industry. Not favorable. To this end, we have launched the "One Chip for Life" program, hoping to cultivate chip design talents on a large scale and with high quality, allowing students to be exposed to the chip design process earlier, and helping to shorten the cycle of talents from the training stage to being put into the front line of scientific research and industry.
Chen Haibo : I think one of the key shortcomings for operating systems is that talent has been very scarce in the past two decades. I think there are three good ways to improve this dilemma:
The first is education , which effectively combines education with root technology and open source, and combines technological frontiers with classic theories in educational practice;
The second is research . It is very important to think about how to guide the academic community to the research and development of root technologies, conduct original research and lead the development of the industry;
The third is the relationship between open source and operating system research . It is necessary to continue to promote academic open source technology and connect it with the mainstream open source community.
Overall, I think talent training, research innovation and open source community building can be combined very well, and we hope to continue to promote this in the future.
Jiang Tao (Founder & Chairman of CSDN, Founding Partner of Geek Bang Venture Capital) : The research and development of root technology requires investment and love. I would like to ask if you understand the current status of young people? What kind of attitude do they have towards studying technology itself?
Chen Songzheng : Overall, there is still a very lack of computer talents. Many of the people our company recruits are actually "transcoded" people, some from electronic engineering, and even some from liberal arts. This is the trend of market demand, because it is easy to find a job in computers and the salary is relatively high.
Kong Jinzhu : I feel that China’s computer level has improved significantly in the past two decades. At that time, there were only a dozen or twenty people in our research institute. Because there was no computer environment and conditions, many of the students we taught could not even write a Hello World by the time they graduated. The young people now being recruited by the company all have very solid foundations and strong ability to think outside the box. Twenty years ago, most of the students I taught were not as good at programming as me, but now, I feel that 90% of the company's R&D personnel can surpass me.
Zhou Bin : Huawei has a saying, "Take root down to the roots and poke up to the sky." In the surrounding environment that I understand, young people now have abundant resources and good qualifications. They are also very diligent and can calm down and "get in".
Yang Zhenkun : The database field is relatively more professional. At present, our OceanBase team is mainly trained from fresh graduates, and most of the people we recruit now are 20 to 25 years old. Love lasts for a long time, and many young people are more enthusiastic than we imagined, and do better than we did back then.
Bao Yungang : In fact, in the process of making open source chips, young people aged 20 to 25 play a particularly important role in it. The average age of the students who participated in the "One Core for Life" program in the early stages was 21 years old. After finishing the "One Core for Life" program, they participated in the Xiangshan Project (a domestically initiated open source high-performance RISC-V processor core project) and have now become a team backbone. When the first version of Xiangshan was produced, the average age of the team was only 23 years old. Even if we include a few of us who are over 40 years old, the average age of the team is only about 25 years old. So my experience is that young people now have great potential, and it is important to give them platforms and challenges to show .
Chen Haibo : First of all, we recently held a software Hackathon competition, and the first place in the preliminary round was a team composed of new employees. Secondly, when young people join our operating system R&D team, I find that their speed in getting started and their level of performance are very high . Now computer-related majors are also very popular in colleges and universities, so we are full of confidence in the future.
Chen Yu : Indeed, computer science has become a major with higher admission scores at Tsinghua University, and many students will transfer to computer science. But from the perspective of basic courses, our research on root technology may still be lacking . Important courses such as software engineering, compilation, operating systems, and CPU composition principles are relatively difficult and energy-consuming courses in computer majors. This makes students more willing to learn some easier courses . I think this It is a challenge we need to face.
In addition, we also need to further strengthen guidance. Although there are now a series of competitions such as various open source software competitions and system capability competitions to promote the cultivation of open source software talents, the main driving force at present is academicians and teachers. We hope that higher-level units such as the Ministry of Education will provide better support. In my opinion, students are like a blank sheet of paper when they first enter school, with great potential for development. If we have some good guidance, this will bring a lot to the cultivation of basic software and hardware and root technology in our country. help.
"Deep roots can produce luxuriant leaves, and long-term roots can lead to long-term development"
Liu Ren (Founder of DoNews) : All teachers are engaged in the computer industry. Will you teach your children programming?
Chen Songzheng : My child is in college, majoring in computer science. I taught him Java in high school.
Kong Jinzhu : I may be too focused on programming that I don’t have time to teach my children.
Zhou Bin : My child is still young and can learn whatever he wants. I won’t force him.
Yang Zhenkun : It depends on the child’s own wishes, hobbies and talents, and the one that suits him or her is the best. But I think programmer is a very good profession. If I had my life again, I would still choose the career of programmer .
Bao Yungang : In fact, during the epidemic, our Computing Institute happened to organize the children of several colleagues to learn programming together. They were very serious and wrote notes. Finally, they published a book called "Six Children Learn Programming".
Chen Haibo : My child is only in the second grade of elementary school. I taught him programming using Scratch (a simple programming tool for teenagers designed and developed by MIT) when he was very young, but I am not very good at educating children. , so he also likes to let him take programming interest classes. No matter what industry he engages in in the future, it is his own choice. But anyway, adding computational thinking to education is very important .
Chen Yu : My child is now in high school. He started learning Scratch in elementary school, as well as Python and C. However, he doesn’t want to learn it now. He may have been taught a little bit. I quite agree with what Teacher Haibo said. From a long-term development perspective, computational thinking is indispensable.
Wu Qingbo: Finally, could you please summarize your experience of this root technology forum in one sentence ?
Chen Songzheng : Our basic software, especially the operating system and CPU, are constantly advancing towards root technology layer by layer. I believe that China will one day become the leader in the development of root technology.
Kong Jinzhu : Only when the roots are deep can the leaves be luxuriant, and when the roots are far away can the history be long-lasting.
Zhou Bin : I hope that through the new era of computing power, we will create a new era and a new system.
Yang Zhenkun : Database is an emerging industry with a vast world and great potential.
Bao Yungang : It takes ten years to cultivate trees and a hundred years to cultivate people. Root technology is building the roots of our country’s information industry.
Chen Haibo : Take root to build the foundation and forge ahead to win the future.
Chen Yu : Let’s meet again next year and build a better future together!
— Recommended reading —
☞腾讯会议部分功能开始收费；马斯克称自己将担任推特首席执行官；微软明年2月将永久禁用IE 11|极客头条
☞NineData，要做程序员必备的数据库产品
☞腾讯云发布三款云原生新品 持续加码云原生布局