[Editor's note] This article was written by Mr. Zhiping Yu, and the first draft was published in December 2020 in "Assume a Mission and Write a New Chapter - The 40th Anniversary Album of the Institute of Microelectronics of Tsinghua University". This paper makes an affectionate review of the development process of the "Panda System", which has an important influence and status in the development history of China's EDA. Relevant alumni and colleagues have given high praise and supplemented some important information. Based on this, some supplements and revisions have been made to the first draft. The revised manuscript is now published on the public account of "Department of Micro-Nano Electronics, Tsinghua University".

Inscription

The writing of this article has a dual purpose: 1) to commemorate the 40th anniversary of the establishment of the Tsinghua Institute of Microelectronics (referred to as the Institute of Microelectronics) that we walked together; 2) to review the unforgettable history from 1986 to 1992, the first complete The whole process of the integrated circuit computer-aided design (ICCAD: Integrated Circuit Computer-Aided Design) system from project establishment, startup to implementation.

Citation

2018 is the 60th anniversary of the invention of the integrated circuit (invented by Jack Kilby of TI in the United States in 1958. He has visited the Tsinghua Institute of Microelectronics, and in 1999 I was an adjunct researcher at Stanford University with him at TI in Dallas, Texas. Had lunch at company headquarters). On several occasions, I have given a report on the title of "Chip Leveraging the Great Era". When I say "big age" I naturally mean the age of information (IT). Specifically, it refers to PCs (IBM) in the 1980s, networks in the 1990s (Cisco, Netscape: the world's first Internet browser), mobile communications in the first decade of this century (Apple/iPhone, 2007) and AI/DL (Deep Learning) and Tesla in the second decade. These are the hallmark applications (and leading companies) of the information age, all of which depend on integrated circuit chips.

So what is the pillar of chip development? My consistent view is that there are two pillars, namely the foundry (foundry, including the manufacturing part of IDM - Integrated Device Manufacturer) and EDA (Electronic Design Automation, electronic design automation) tools. And chip design is a superstructure. This memoir is devoted to discussing the development of EDA in China, especially in the early stage of Tsinghua Micro Institute.

01

A brief history of the early development of EDA

Now let's review the early history of EDA (also known as ICCAD) and the history of its establishment in China.

Talking about the early history of EDA and its foothold in China, it has to be mentioned that the Institute of Microelectronics of Tsinghua University received an ICCAD design system called Applicon in Boston, USA in 1984. Now everyone knows that there are three giants (in order of size) of EDA tool suppliers in the world: Synopsys, Cadence, Mentor. But in reality, Synopsys was only founded in 1986, and Cadence was listed in 1988 (although its two original constituent companies were created in 1983). The early EDA tool suppliers were Calma (the company that established the GDS-II data format, founded in Silicon Valley in 1964, and this data format will be discussed below), Applicon and ComputerVision (famous for graphics editing software, established in Massachusetts in 1969) ) Three giants and an ECAD company (which is the layout design verification software Dracula, including DRC/ERC, the inventor and supplier of Design/Electrical Rule Check, one of the two companies that later formed Cadence). All three graphics editing software companies were founded before 1980.

This Applicon design system is produced by the Applicon Graphics System Company (Applicon is derived from English Application) founded by four Ph.D. graduates from MIT in 1969. It is specially used for computer-aided drawing (such as the layout of integrated circuits and the wiring of printed circuit boards). ) and edit. The software system can run on a general-purpose minicomputer, equipped with a graphic terminal. The Applicon system also has a large plotter for graphic output of the layout. In 1984, Tsinghua Institute of Microelectronics also purchased a small computer PDP-11 from DEC (Digital Equipment Corporation) in Boston, USA, which was considered very advanced in China at that time. This DEC company was also founded in 1957 by several MIT (Lincoln Laboratory) researchers. DEC was one of the most influential computer companies other than IBM in the mid to late 1980s. It can be seen that the existence of famous universities in the United States played a key role in the rise of the local high-tech industry (at that time, many computer and other high-tech companies on the east coast of the United States were located in the western suburbs of Boston, which was called the "126 Highway Corridor". Silicon Valley in California is even more famous). In the early 1980s, the operating model of EDA companies in the United States was to provide both design software and computer platforms (including workstations) on which the software was run.

I am reviewing the early history of EDA and the process of settling in China here to point out that people often hear that China's semiconductor/integrated circuits are lagging behind (compared with advanced countries) because of its late start, which is actually inaccurate. Originally, the development of various countries has precedence. In the time span of 62 years (for example, 73 years since the invention of the transistor at the end of 1947), a few years can be regarded as starting at the same time.

02

The purchase of the American Applicon design system laid the foundation for the development of the CAD room of the Microelectronics Institute

The CAD discipline of Tsinghua Institute of Microelectronics officially started with this Applicon design system and PDP-11 minicomputer. Such a very advanced ICCAD design system would cost considerable foreign exchange. Professor Yang Zhilian was seconded from Tsinghua University for one year (September 4, 1982 to September 1, 1983) by the then State Science and Technology Commission (High-tech Bureau). After the secondment ended and returned to school, he began to prepare the CAD room of the Micro Institute. This introduction fund was approved by Hu Zhaosen, director of the High-tech Bureau of the State Science and Technology Commission, and then reported to the leaders of the Science and Technology Commission for special approval.

From July 21 to September 7, 1984, Professor Yang Zhilian led three teachers (Wang Zeyu, Zhang Jin and Wang Hui from the Department of Radio) to Boston, USA to receive the system and receive relevant training. Because I was studying for a doctorate in the Department of Electrical Engineering of Stanford University (jointly dispatched by the Ministry of Education and Tsinghua University in October 1979), and it was the summer of the school, I also flew from the western United States to Boston to participate in the reception work for a period of time. This system has been widely used in the Micro-Institute. The layout design and verification work of various integrated circuit research and development in the institute is carried out on this system. It wasn't until the introduction of the Daisy system (another ICCAD company in Silicon Valley that was on par with Mentor and Valid in the 1980s, collectively jokingly called DMV – an acronym for DMV) that ended its historical mission. Numerous graduate students also conduct their own research projects at PDP-11.

Before 1984, the CAD research of Tsinghua Micro-Institute had begun to take shape. For example, the circuit simulation program SPICE was used to analyze the circuit of the memory and microprocessor developed by the institute. After 1984, the research content mainly includes semiconductor device simulation (my research topic at Stanford is this direction), such as the development of two-dimensional and three-dimensional device simulators, and the research on design methodologies, such as the layout of the standard cell method. Wiring study. In 1990, he successfully developed the semiconductor device model parameter extraction program MODPEX, and then cooperated with Symmetry Company of Silicon Valley in the United States and sold this software to foreign countries.

03

The Development Trend of ICCAD Tools in my country in the Early 1980s

The independent development of China's ICCAD design tools has actually been carried out for a long time, but the scale is small, the layout is scattered, and it is mainly concentrated in universities and research institutions. The Department of Computer Science of Tsinghua University has carried out research in this area for a long time. In the early 1970s, it established the CAD Research Office (Professor Hong Xianlong is in charge), engaged in CAD theory, algorithm and system research, and developed a layout editing system. Other universities in Beijing and Shanghai also conduct research on the layout of integrated circuits. In addition, Wuxi Huajing Company has also developed an integrated circuit layout design system. All these have laid the foundation for the subsequent development of the national key project panda design system.

Going back further, the ICCAD conference held in Yangshuo, Guilin in the autumn of 1978 can be regarded as the starting point of Chinese EDA research. After that, the first-level system called integrated circuit CAD system (referring to the layout plotter and engraving machine controlled by the computer) and the second-level system (referring to the layout graphics editing system and DRC, ERC, SvL, etc.) were successively developed in China. Schematic vs. Layout tool and layout data output function. The development of this system was in charge of Professor Hong Xianlong of the Department of Computer Science of Tsinghua University, started in 1984, and won the second prize of the National Science and Technology Progress Award). However, due to the lack of top-level design and careful layout, the database formats adopted by each company are not compatible with international standards, and are basically a loose combination of some point tools, which cannot meet the complete design requirements of large-scale integrated circuits. In this context, the Ministry of Electronics Industry (then called the Four Machines Department) gradually came up with the idea of organizing national forces to develop a complete ICCAD system. In the early stage, Professor Wang Yangyuan of Peking University (later an academician of the Chinese Academy of Sciences), Professor Yang Zhilian of Tsinghua University, and Wu Zhengming of the Integrated Circuit Group of the Office of the Large-scale Integrated Circuit and Computer Leading Group (referred to as "Daban") were the main driving forces. Wu Zhengming has organized experts to discuss how to develop my country's CAD technology for many times. Next, Zheng Minzheng, the director of integrated circuits in the Ministry of Electronics Industry (Zheng returned to the Electronics Department from the "big office", he is also a Tsinghua alumnus) made a lot of preliminary preparations.

04

The small meeting in the post and telecommunications sanatorium became the opportunity to start the panda system

Since its establishment in 1980, Professor Yang Zhilian has been the deputy director in charge of scientific research in Tsinghua Institute of Microelectronics, assisting Academician Li Zhijian. After completing the introduction of the Applicon system, he proposed the development of an ICCAD system, which he called a three-level system at the time, to meet the needs of ever-increasing integrated circuit designs.

I finished my post-doctoral research work at Stanford University in the United States, and in February 1986, I returned to Tsinghua University to report to become a full-time teacher (I was a graduate student of Tsinghua University when I was sent to Stanford for my doctorate in October 1979). After receiving my Ph.D. in Electrical Engineering from Stanford University in June 1985, I worked as a postdoctoral fellow at Stanford University at the University of Bologna in Italy (the oldest university in the world, with a history of more than 900 years) and SGS in Milan (later with France The Thomson company merged into the current STMicroelectronics company) engaged in the research of semiconductor device and circuit simulation for half a year.

Shortly after I returned to teach at Tsinghua University, Professor Yang Zhilian contacted Professor Wang Yangyuan of Peking University to discuss whether to discuss in a small area and how to suggest that the country should set up a key research project for integrated circuit CAD (at that time, it was the planning time of the Seventh Five-Year Plan). In this way, in April 1986 (the specific date is no longer remembered), a meeting was held in the Sanatorium of the Ministry of Posts and Telecommunications (now known as Beijing Posts and Telecommunications Conference Center) on the south side of Summer Palace Road outside the west gate of Tsinghua University. A four-person meeting. The meeting lasted for two days. Although the meeting place was very close to Peking University and Tsinghua University, we still lived in the nursing home of the Ministry of Posts and Telecommunications, so that we could concentrate on coming up with the first draft of the plan. At that small meeting, a proposal for the development of a brand-new integrated circuit CAD design system was formed to the state. Considering the continuity of history, we named this new ICCAD system as the third-level system to indicate that it was carried out on the basis of the accumulation of the first-level/second-level system.

05

The establishment, initiation and implementation of the three-level system (1987-1992, a total of five years)

After the small meeting in the post and telecommunications sanatorium, the opinions of the meeting on the development of a three-level integrated circuit computer-aided design system were submitted to Zheng Minzheng, the director of the integrated circuit of the Fourth Machinery Department. During the same period, Li Tieying served as the head of the Fourth Machinery Department. He nominated Professor Wang Yangyuan from Peking University to be seconded to the Fourth Machinery Department as the deputy director of the Microelectronics Bureau. In this way, the proposal to start the three-level system is on the fast track. A related development is that in the same year (July 1986), the Fourth Ministry of Machinery established the Beijing Integrated Circuit Design Center in Jiuxianqiao, Beijing (electronic industry cluster), which provided the material basis for the start-up of the tertiary system.

The three-level system was officially established in November 1986 (it was just half a year after the four-person seminar of Peking University/Tsinghua University in the Post and Telecommunications Sanatorium in April, and the efficiency can be said to be quite high), then on December 3, 1986, the National ICCAD Expert Committee was established. Professor Wang Yangyuan (who was also the head of the Microelectronics Bureau of the Fourth Ministry of Machinery at the time) was the chairman, and Professor Yang Zhilian and Professor Hong Xianlong of Tsinghua University were both members of the expert committee. The three-level system was launched in early 1987, relying on the Beijing IC Design Center, and the overall design team was responsible for implementation. The head of the overall design team is Professor Hong Xianlong of Tsinghua University (I am also a member of the overall design team). Immediately organize domestic experts to discuss and determine the system framework, mobilize domestic development teams, and conduct technical training. Considering the high starting point of the three-level system and the goal of an independent and complete integrated circuit CAD system in line with international standards, the Fourth Machine Department decided to hire Dr. Lian Yongjun from the United States as the chief designer (he was in Austin, Texas, United States at that time. Working in a computer company, he is quite familiar with the EDA system, graduated from Taiwan University with a bachelor's degree and obtained a doctorate in the United States). Dr. Lian took office in March 1988, and immediately established a new chief design team, and identified four deputy chief designers, namely Wang Zhenghua (Beijing Design Center) and three teachers from Tsinghua University: Hong Xianlong and Liu Xiling from the Department of Computer, Microelectronics by Yu Zhiping.

The research time of the three-level system was from the beginning of 1987 to the middle of 1992, which lasted for 5 years. During this period, Dr. Lian Yongjun proposed to change the three-level system to the panda system in 1990. In 1993, the National Science and Technology Commission awarded the Panda System National Science and Technology Progress Award. In the certificate issued, the system was officially named "Panda Integrated Circuit CAD System", which is the first complete and independent integrated circuit computer-aided system in China. The complete process of design system formation. It has laid a solid foundation for the development of China's EDA industry and even the entire integrated circuit industry in the future. The title of this article is "Dawn of the East", which is a sincere sustenance and expectation for China's integrated circuit industry.

The following will introduce the three-level/panda system from the technical level, the composition of the research team, and the organizational structure.

06

The data format EDIF, a solid foundation for the tertiary system

The reason why the third-level system has the confidence to act as an independent and complete ICCAD system is inseparable from the international standard EDIF (Electronic Design Interchange Format) that appeared shortly after the public "design data exchange format". The circuit level of integrated circuit design (logic gate level/circuit netlist and schematic) must be represented by a unified data format that can be exchanged between design systems developed by different companies. EDIF is a design data standard proposed in 1983. It is in the public domain, so it is free. It is a neutral data format that has nothing to do with individual manufacturers. Dr. Lian seized this opportunity and proposed to build a three-level system on this data format to ensure the integrity and autonomous controllability of the system. This decision is obviously very correct, and it has laid a good foundation for the establishment of China's own ICCAD system (due to technical reasons, the current panda system no longer uses the EDIF format, and the process will not be expanded here).

07

The composition of the attacking team

The development team of the three-level system is mainly composed of colleges and universities, relying on the Beijing IC Design Center (they also have their own technical team). According to reports, there were nearly 120 participants from 16 units at most, and there were 10 schools: Peking University, Tsinghua University (Department of Computer and Microelectronics), Beijing Institute of Technology (now Beijing Institute of Technology), Beijing Institute of Posts and Telecommunications (now Beijing University of Posts and Telecommunications), Beijing University of Technology, Fudan University, Shanghai Jiaotong University, Zhejiang University, Hangzhou Institute of Electronic Technology (now Hangzhou Dianzi University, referred to as Hangdian) and Harbin Institute of Technology; there are 4 scientific research units: Beijing Institute of Automation, Ministry of Mechatronics, Software Laboratory of Institute of Computer, Chinese Academy of Sciences, Microelectronics Center (now Institute of Microelectronics, Chinese Academy of Sciences) and Shanghai Institute of Metallurgy (now Shanghai Institute of Microsystems, Chinese Academy of Sciences); 2 Industrial Research Units: Central Research Institute of China Huajing Electronics Group and Beijing Integrated Circuit Design Center.

Our Tsinghua Institute of Microelectronics has successively participated in the three-level system research teachers including Shen Ming, Xia Chun, Wang Zhihua (then still belonged to the Department of Radio) and me. Students He Xinping, Zhao Weijian, Shao Weizhi and others, and the others could not remember their specific names. The Department of Computer Science of Tsinghua University has teachers such as Wu Qiming and Bian Jiinian, as well as a considerable number of doctoral and master students in the department.

One of the major achievements of the three-level system is that it has cultivated a large number of outstanding talents in the EDA field, and they are still playing a leading role in the front line. Liu Yu, a graduate of Ruwei Institute (participated in the third-level system research as a member of Beijing Design Center), co-founded Apache Design Solutions in Silicon Valley of the United States (acquired by the world-renowned EDA company Ansys in 2011), and developed IC technology. Consumption analysis software is recognized in the industry as a killer ICCAD tool comparable to Mentor's layout verification software Calibre, and enjoys a high reputation.

08

project breakdown group

Dr. Lian has worked in large companies in the United States, so his management modernization level is very high. The entire three-level system, according to the recollection of our colleagues in the micro institute, is divided into the following research groups (not necessarily all), and here is a record for the reference of those who are interested:

DB (data base, database): Teacher Liu Xiling from the Department of Computer Science, Tsinghua University is responsible for

LE (layout editor, layout editor): Teacher Xia Chun from Tsinghua Microelectronics Institute participated, this is a major topic

SE (schematic editor, schematic editor)

SIM (simulation, circuit simulation): Yu Zhiping is in charge of the Institute of Microelectronics, and Wang Zhihua, a teacher from the Radio Department of Tsinghua University, participated. The students include Zhao Weijian (Tsinghua Institute of Microelectronics), Guo Miaoquan (Zhejiang University), Shi Chuanjin (Fudan) and Gu Shihua (Hangdian), etc.

UI (user interface): Shanghai Jiaotong University students participated

Testing: responsible for Shanghai Jiaotong University

GDS-II layout data output

Design Verification (including DRC/ERC)

09

The follow-up development and influence of the panda system

The first phase of the three-tier system lasted for more than five years, and was marked by the departure of Dr. Lian Yongjun in 1992. We can not forget his contribution to China's EDA cause, and hereby express our admiration.

After the Panda system was completed in 1992, it was taken over by Beijing Integrated Circuit Design Company and continued to develop and promote its application. Although the difficult process during this period is similar to the overall development of China's integrated circuit industry, it has undoubtedly become the mainstay of the domestic EDA industry with its impressive performance over the years, establishing a leading position. What is particularly gratifying is that its SPICE circuit simulation program ALPS has ranked first among the same type of tools in the world, and has been widely recognized, including Huawei's HiSilicon and the United States' nVIDIA using this program a lot. This is an excellent example of a domestic EDA tool freed from foreign restrictions.

Postscript: The layout editing software of the three-level system entered the international market in the late 1990s

As the end of this article, I will cite a successful case in which I have been involved in bringing Panda's software to the US market. The most competitive software in the Panda system at that time was the layout editing part. In 1996, when I was a researcher in the Department of Electrical Engineering (EE) at Stanford University, I worked with Dr. Hu Chengmin from Silicon Valley and Liu Weiping, general manager of BGI (who was the main force sent by Fudan to participate in the three-level system research, and later stayed in the Beijing Design Center. He is in charge of Panda Systems and is now the chairman of Huada Jiutian) together to create an EDA company named Stanza. Tsinghua Microelectronics Institute was developed by Mr. Shen Ming who came to the United States to participate in the development. This company, which is mainly based on layout editing tools, quickly opened up to serve Silicon Valley IC design companies, because the product has its own characteristics, attracted the attention of Synopsys, and was acquired by it in 1999. There is certainly no shortage of regrets from the current point of view. But it gave us a lot of confidence at that time: China's EDA tools can definitely enter the forefront of the world, and now we have seen the rising sun.

Disclaimer: This article is reproduced from "Software Market Think Tank" to support the protection of intellectual property rights. Please indicate the original source and author of the reprint. If there is any infringement, please contact us to delete it.