He observed that undergraduates and postgraduates majoring in computer science are generally reluctant to learn a more basic computer system structure, but are more concerned with computer applications.
'The talent pool in the university's chip industry is worrying. Almost everyone is doing something about computer applications rather than basic things. ' Li Yudong understands the objective differences in talent between the computer basic field and the application field, 'the employment rate of colleges and universities, Students look at the market's expectations, but he believes that there is a big difference between salary and total talent.
According to the white paper of China's integrated circuit industry (2016-2017) (hereinafter referred to as the white paper) released by the Ministry of Industry and Information Technology of the Ministry of Software and Integrated Circuits (CSIP) in May 2017, the total number of integrated circuit employees in China is currently less than 300,000. People, but according to the total output value, it needs 700,000 people, and the total amount of personnel training is seriously insufficient.
How to fill the 400,000 chip talent gap? The answer to the question is not only hidden in the education sector.
Chip talent gap 400,000 'Top-heavy'
"Essentially it is all about teaching students how to use computers, not how to teach students how to build computers. Just as a car major teaches a bunch of drivers." Talking about the topic of chip talent, researcher at Institute of Computing Technology, Chinese Academy of Sciences, 'Lonse' Hu Weiwu, the person in charge of the processor, has a lot to say. In his view, China's chip industry is extremely unbalanced in talent cultivation. Most of the talents are concentrated in the technical application level. However, there are too few talents in the underlying systems such as algorithms and chips.
He readily gave a practical example: The vast majority of Internet companies are programming in Java, with corresponding talent pools of hundreds of thousands or even millions, but researching Java Virtual Machines (simulating various computers on real computers) There are very few talented people in the function of abstraction computer. 'I didn't even have 10 people when I started a business in 2010', but today the country may still not have more than 100.
'College education not only teaches the talents of using computers, but also teaches an architecture, an operating system, and these teaching systems should be developed.' Hu Weiwu called for computer professionals to strengthen the training of basic talents.
To this question, Li Guojie, an academician of the Chinese Academy of Engineering and honorary chairman of the Chinese Computer Society, has summed up four words - 'Top-heavy'. He mentioned that the reserve of talents and training are relatively weak. It is because China's chip semiconductor industry is still compared with the international top level. There is a clear difference in a key factor.
According to the white paper, from 1999 to 2016, the average annual growth rate of integrated circuit design in China was 44.91%, which is booming. However, the independent innovation capability of China's integrated circuit industry is weak, and the key core technology has high external dependence and lack of talents. prominent.
According to the "National Integrated Circuit Industry Promotion Program," the industrial scale will expand by more than five times by 2030, and the demand for talent will increase exponentially. The supply of industrial talent does not match the growth rate of industrial development, relying on the development of ICs (integrated circuits). Talent cannot meet the requirements of industrial development.
According to the White Paper, the annual salary of R&D professionals in integrated circuit companies is nearly 300,000 yuan, and the production and manufacturing professionals are nearly 200,000 yuan. The average annual salary of graduate students with undergraduate degrees in chip design is nearly 150,000 yuan, and the doctoral degree is nearly 300,000 yuan. The survey shows that 80% of companies pay their salary once a year, and the percentage of salary increase is between 5% and 10%.
However, this salary level is significantly lower than that of Internet companies’ hot posts, especially those of big data, artificial intelligence, etc. Recruitment information of internet talent recruitment websites such as Lagounet shows that they graduated from computer majors, and Artificial intelligence talents with 4 to 5 years of work experience can get a maximum monthly salary of 40,000 yuan. Considering that many Internet companies will be paid for more than 12 months, the final annual salary may exceed 500,000 yuan.
Obvious salary differences have led to a number of chip R&D talents who have accumulated deep in the infrastructure field and have begun to transition to the Internet application field. Qin Lin (a pseudonym) is one of them. According to him, the chip R&D institutions in the first-tier cities in the north The salary of the company's work is often inferior to the salary that first-line Internet companies can provide, and Internet giants such as Ali and Baidu have also started to invest more in core technology research and development.
The talent assessment system in the IT field has changed.
The problem of being 'top-heavy' has made Li Yudong very worried. He said that if this situation continues to develop, there will be no talent available for computer research in the next three to five years. 'Solving this problem requires systematic building of governance.' Zhiben. '
Li Guojie analyzed that the reason why there is a top-heavy problem in the cultivation of domestic IT talents is because the underlying technologies such as chips have a relatively high threshold, and only top universities such as '985' have been trained; on the other hand, they have also been trained by domestic talents. There are still some problems with the institutional mechanism.
He said that currently the assessment of computer talents by domestic universities and scientific research institutions mostly uses published papers as the main evaluation criteria. Compared with big data, artificial intelligence, and other fields of application, the chip researches these types of papers in areas where trial and error costs are very high. Or the difficulty of making original invention patents is significantly higher. Therefore, there are fewer opportunities to be selected for training programs such as the National Outstanding Youth Fund.
Zhang Yongfeng, an associate professor at Dalian Neusoft Institute of Information, bluntly stated that the talent assessment system in the IT field should be changed. “We can't put all the majors on the same page, especially the assessment of the national professional titles, performance appraisal.” He believes that in the chip field, China is already lagging behind Europe and the United States. The country will not be able to produce a lot of achievements in a short period of time and make a lot of money. If the assessment of chip talent is just like other fields at the moment, only looking at academic papers and only performance theory, then there may be a lot of people leaving this field. Better-paying work.
He believes that higher education in China should strengthen the cultivation of engineers' culture. 'Everybody is not paying much attention to engineering in order to publish papers, and does not pay much attention to engineering. It feels that engineering is a relatively low-grade thing. In the field of chip R&D and production, engineers decide the chip. The key factor of whether design innovation can be landed. 'Zhang Yongfeng suggested that with reference to Europe and the United States' mature experience, establish a national unified learning and practice platform with integrated circuit design and manufacturing as the theme, provide integrated circuit design EDA tools, process libraries, and even experiment Platform. National IC related students can apply to use the resources on this platform. This will improve the quality of personnel training from the basics, and reduce the duplication of resources between various universities. Like some mature integrated circuit technology, Share on the platform and let students learn.