There is a large gap in FPGA talent, and the domestic FPGA industry is facing challenges!

小柳叶

There is a large gap in FPGA talent, and the domestic FPGA industry is facing challenges! [Copy link]

 

Alibaba's semiconductor company "Pingtou Ge" has released the Xuantie 9 10 chip and will open it to developers. Developers around the world can download the FPGA code of the processor for free and quickly carry out chip prototyping and architecture innovation. With the rapid arrival of the traffic era, the amount of data and information has exploded, and the way of processing data can only keep up with the pace of technological development. Of course, no technology can solve all problems comprehensively. From CPU, GPU , to FPGA, ASIC, to future neuromorphic computing, quantum computing, etc. , Intel is comprehensively laying out future end-to-end computing innovations to fully unlock the value of data. Among them, FPGA has flexible, efficient and reprogrammable characteristics , which can achieve customized performance, customized power consumption, high throughput and low batch latency, so it is more and more widely used in artificial intelligence, cloud services, data centers, 5G, autonomous driving and visual processing.

The problem that comes with this rapid development is the current situation of severe shortage of talent development. At present, the global FPGA market size has reached 5 to 6 billion US dollars, and by the end of 2022 the scale will exceed 10 billion US dollars. According to the 2018 "China Integrated Circuit Industry Talent White Paper" , the current talent reserve in China's integrated circuit industry has reached 400,000 people, but there are less than a few thousand professional FPGA talents, which is only one-tenth of the foreign talent market . This serious FPGA talent gap in China not only hinders the development of enterprises , but also has a serious impact on the domestic FPGA ecosystem . Therefore , it is urgent to cultivate available FPGA talents and build a mature and stable FPGA ecosystem .

The FPGA market is facing the problem of no output and no training for FPGA talents . It takes three to four years for colleges and universities to train talents, and the number is limited, the output is insufficient, and the project capacity is not strong. At present, the training of FPGA talents in China obviously cannot keep up with the development needs of FPGA. Most of the existing talents choose to use C language, C++ or Java, etc. The teaching staff and the number and quality of students in FPGA programming language are seriously insufficient, and the lack of project practice links has led to a disconnect with social needs .

It is very difficult for FPGA talents to get started and further their studies, and it is difficult for college graduates to get in touch with high-tech. At present, colleges and universities have not yet formulated a systematic training system for FPGA talents , and the courses are general and fragmented . Therefore , it is difficult for novices to get started and it is even more difficult for practitioners to improve . FPGA itself has complex hardware and language problems , and development tools are updated quickly ; even FPGA engineers who have been in the industry for many years often find it difficult to get in touch with the latest FPGA technology. As a discipline that combines intelligent technology and complex algorithms , FPGA has encountered the practical problems of no platform, no practical operation, and no projects in the learning process . Naturally, it is difficult to achieve rapid entry of talents and further improvement of technology, thus restricting the possibility of high-end employment.

It is precisely because of the high entry difficulty of the FPGA industry that it is difficult for China to establish a leading FPGA company . It is difficult for colleges and universities to cultivate talents. If the subsequent work of enterprises can keep up, this situation can be alleviated. However, domestic FPGA companies are generally in their infancy, with many problems such as "backward technology level, lack of talents, lack of core patent technology, and low level of industrialization". Graduates enter the company, but because they have no project practical experience, it is difficult for them to follow the company to engage in the most cutting-edge FPGA research work. This has become a cycle. The less experience you have, the more difficult it is to grow and become an excellent FPGA talent through work.

At present, the training cycle for an FPGA chip R&D engineer is at least three years , and the training cycle for an FPGA software algorithm engineer is at least five years . If a graduate wants to become a qualified FPGA engineer, he needs to go through a series of project operations, from "training and learning basic FPGA knowledge, to taking on R&D tasks with corporate mentors, and then to taking on R&D tasks alone . " Such a long talent training cycle has become a problem that hinders the growth of domestic FPGA talents.

Overall, the FPGA talent problem has greatly exposed the shortcomings of the rapid development of domestic FPGA. In terms of solving the talent problem, it is still necessary for the government, enterprises, and universities to work together to establish a domestic FPGA ecosystem, thereby promoting a qualitative leap in China's FPGA.

飞鸿浩劫

Poor ecology

小柳叶

Feihong Haojie published on 2019-8-27 12:54 The ecology is not good

It will gradually get better

yang8555u

The main problem is salary. Without salary, it's all nonsense.

小柳叶

yang8555u posted on 2019-8-28 20:00 The main problem is the salary. Without salary, it’s all nonsense.

The salary in the FPGA industry starts at 8k, depending on your work ability

yang8555u

Xiao Liuye posted on 2019-8-29 08:38 The salary in the FPGA industry starts at 8k, depending on your work ability

Internet wages start higher

小柳叶

yang8555u posted on 2019-8-29 23:53 The starting salary for Internet jobs is higher

8k is just the starting point, there is a lot of room for advancement, ability and income are directly proportional

generalcircuits

FPGA is slow to learn and relatively difficult to get started with, but the salary is okay.

yang8555u

Xiao Liuye posted on 2019-8-30 09:11 8k is just the starting point, there is a lot of room for advancement, ability and income are proportional

But the overall salary is indeed not as good as that of the Internet, and sometimes it is even much worse. This is a fact.

小柳叶

generalcircuits posted on 2019-8-30 10:46 FPGA is slow to learn and relatively difficult to get started. The salary is okay, right?

Yes

小柳叶

yang8555u posted on 2019-8-30 19:28 But the overall salary is indeed not as good as that of the Internet, and some are even much worse. This is a fact.

There is no comparison. Those who know the Internet may not know FPGA, and those who know FPGA may not know the Internet.

lizhengjun

小柳叶

lizhengjun published on 2019-9-1 16:50

chenbingjy

I want to learn FPGA.

小柳叶

There is a large gap in FPGA talent, and the domestic FPGA industry is facing challenges!

hotsauce1861

The salary gap is too big and the learning curve is steep, which may be the reason why there are relatively few people working on FPGA.