A chip the size of a fingernail may be the tiniest and most ambitious project ever created by human beings—tens of billions of transistors integrated together—can be described as a miniature city.
Just as buildings in cities require a large number of prefabricated parts, in the chip industry chain, semiconductor IP is a key technical link in the design upstream, just like “prefabricated parts” in the construction industry, using a building block-like development model to help shorten chip development time and reduce R&D risks .
However, the top ten IP suppliers in the world are mainly from the UK and the US. Under the monopoly of foreign giants, can the fresh blood of domestic IP bring new vitality to market competition? How can domestic IP break through?
Recently, Zeng Keqiang, chairman and co-CEO of Xinyaohui Technology Co., Ltd., a domestic semiconductor IP start-up, said in an exclusive interview with The Paper (www.thepaper.cn) in Zhangjiang, Pudong, Shanghai that China’s semiconductor IP demand “has reached an impending explosive growth. period”. Although the domestic IP industry is still relatively weak at present, he firmly believes that now is the best time to accumulate and develop. The Chinese can develop future-oriented IP technology on their own, provided that the top talents in China must be attracted.
Zeng Keqiang was the deputy general manager of China at Synopsys, the world’s top IP design company. He has been responsible for the industrialization of chip EDA and IP market for more than 20 years in Synopsys. He keenly captured the high speed of the semiconductor IP market. Although the market share of foreign investors exceeds 90%, China urgently needs advanced IP technology and products independently developed.
IP design is not a closed-door process. Without more than ten years of mass production, running-in, and iteration with customers, it is not realistic to develop IP. Judging from the development history of global IP in the past 30 years, the rapid iteration of technology and the emergence of new demands for market applications have jointly promoted breakthroughs in the IP industry. At present, in China, chip applications are at the forefront of the world. This is an opportunity that can be seen and grasped in this special era.
June this year is the first anniversary of Xinyaohui, Zeng Keqiang led various teams to celebrate. “Everyone feels that what we are doing is in line with the times. Everyone has the same dream and the same belief. The key is to bring people together.”
The interview time was set at 10:00 a.m. When the surging journalists arrived, he had just finished a meeting. In the more than one hour interview, Zeng Keqiang was very passionate and experienced the pain of China’s lack of “core”. “At that time, everyone already had a feeling, what if one day IP was stuck in the neck?” Zeng Keqiang said that their generation of semiconductor people You have to stand up and say, “We really should do something for the industry.”
“Our team has accumulated decades of experience in the world’s top companies, so we have the responsibility and mission to create an independent research and development Chinese IP company to break through the bottleneck problem and solve problems for the country and the industry.”
“Prefabs” in chips: shortening development time, tandem chip design and manufacturing
The semiconductor IP core is the key technical link in the upstream of the system chip design industry chain. It is a proven, reusable chip module with specific functions. The complex and diverse end products, the increasing difficulty of chip design, and the continuous increase in R&D resources and costs have prompted the continuous refinement of the division of labor in the global semiconductor industry, and the IP industry has emerged as the times require.
Because it has been verified and can be reused, semiconductor IP cores are like prefabricated parts in the construction industry. At present, most chip companies in the world will use IP when designing chips. By purchasing a mature and reliable IP solution, a certain function can be realized and the chip shortening time can be shortened. Development time, reducing R&D risk, and putting up a building in no time. According to the report of IBS (International Business Strategies), a US electronics industry strategy consulting company, taking the 28nm process node as an example, the number of IPs that can be integrated in a single chip is 87. When the process node evolves to 5nm, more than 200 IPs can be integrated.
The increase in the number of IPs that can be integrated into a single chip provides a new space for more IPs to be reusable in SoCs (system-on-chips). IP has become an indispensable element in the design and development of integrated circuits, promoting the further development of the semiconductor industry and chip technology.
IBS has predicted that the global semiconductor IP market will increase from $4.6 billion in 2018 to $10.1 billion in 2027, a growth rate of 120%, and an average annual compound growth rate of 9.13%. Semiconductor IP is ubiquitous, it can be divided into processor IP, interface IP, physical IP, etc. Among them, the high-speed interface IP is large and wide, such as mobile phone charging and external data exchange, the USB interface is used. Due to the explosive growth of mobile phone applications, the data to be transmitted is getting larger and larger, and the data transmission rate of the interface IP is getting faster and faster. Otherwise, the performance of the mobile phone and the user experience will be affected.
Zeng Keqiang gave an example, “The USB interface has been used by everyone for many years. If you look closely, you can see that the shape has changed a lot. It used to be a relatively thick interface, and then it came to Type C and then to now use the lightning in Apple. belongs to the interface IP.”
“The interface IP has changed from USB2.0 to 3.0, 3.1 to the latest 4.0, and the transmission speed has been increased by 85 times.” This also means that in the past, we used to export a full HD video from a mobile phone. It needs to be measured in hours, but now Export a higher-quality 8K video in less than a minute.
For another example, computing devices need to store data and interact with the CPU, which requires the use of DDR memory interface IP. Zeng Keqiang introduced that the current games are known for their exquisite picture quality. The premise of achieving realistic picture quality is that the device can perform a large amount of data calculation and processing to complete the rendering, which requires high bandwidth and large-capacity storage support.
The DDR interface is constantly iterating to meet this demand. The current most advanced DDR interfaces are DDR5 and LPDDR5. Adopting the most advanced DDR interface has become one of the important means for Electronic terminal to differentiate products and markets.
As the personal computer industry moves towards the mobile phone industry, electronic terminal products are more complex and diverse, and the complexity of chip design has increased significantly. Zeng Keqiang said that IP design standards are also developing in the direction of larger bandwidth, larger data volume, higher transmission rate, and lower power consumption, so research and development is more difficult.
For the chip ecosystem, IP is also an important part of the supply chain. As it is located in the upstream of the industrial chain, it plays an important role in the innovation of the entire industrial chain and can also drive downstream development.
The value of IP companies is to pre-design common functions in chips, and make them highly reliable and compatible, so that chip companies can reuse them directly without taking R&D risks and saving development time.
A mature IP must be fully verified to avoid risks to chip manufacturers. You must know that the chip industry is an industry with almost no room for fault tolerance. The cost of tape-out of an advanced process chip is hundreds of millions of dollars.
On the other hand, IP and chip manufacturing (chip foundry industry) share weal and woe. Zeng Keqiang said that each chip manufacturing company has its own unique chip process, and the names of the process and process are similar to the public, such as the 14nm process or the 7nm process, but the technology paths are flourishing, and the process parameters at the nano-level are also very different. IP research and development needs to cooperate with each chip manufacturing company’s unique process to design products, so as to ensure the “freedom of IP choice” for downstream chip companies.
“So chip manufacturing companies also regard IP manufacturers as one of their most important ecological partners,” Zeng Keqiang told The Paper (www.thepaper.cn), “China’s semiconductor industry is striving to catch up, and it is the current standard to benchmark international advanced technology. The goal, Xin Yaohui is also one of them.”
Foreign market share exceeds 90%, what is the difficulty of semiconductor IP research and development?
Semiconductor IP market players can be roughly divided into two categories: one is semiconductor IP suppliers that are bundled with EDA tools, such as Synopsys in the United States; the other is suppliers that only provide semiconductor IP in professional fields, such as British ARM, Imagination. These companies were all established in the 1980s and 1990s.
And today the most widely used semiconductor IP also mainly comes from them. According to a survey by IPnest, a research organization in the IP field, in the market share distribution of global semiconductor IP suppliers’ sales revenue from 2019 to 2020, the top two are ARM and Synopsys. Among them, ARM’s market share in 2020 is 41%, and Synopsys is 19.2%. These two companies alone account for 60% of the global semiconductor IP market.
“Domestic IP is basically in a state of being stuck.” A chip practitioner told The Paper (www.thepaper, cn), “Even the most powerful chip design companies in China have to rely on foreign parties. Not to mention other chip companies.”
The foreign market share of semiconductor IP exceeds 90%, and China urgently needs advanced IP technology and products independently developed. However, China’s semiconductor IP has not successfully broken through so far. Zeng Keqiang said that the main challenges come from three aspects. The first is the rapid upgrade of various interface IP standardization protocols, which determines that IP research and development is not a one-off thing, but requires long-term investment.
Secondly, IP research and development needs to continuously track the process evolution of each chip manufacturing company, and must have the ability to quickly transplant the process.
Finally, after integrating protocol standards into IP and customizing each process, it is necessary to work with downstream chip design companies to conduct chip verification and mass production to ensure the marketization of IP products.
“The local IP standard itself is slower than foreign countries, and the technology has not accumulated.” The above-mentioned chip practitioners said that 14 and 12 nanometer IP is currently a scarce technology in China. “The advanced process IP that is actually used by customers in the market and has been tested by the market is almost blank.” The reality is that the development of new applications such as high-performance computing, 5G, and smart cars requires advanced process IP to support. .
“IP is not easy to do, and it is even more difficult to do well.” Zeng Keqiang said frankly that semiconductor IP is not a scientific invention, but a typical industrial technology, an industrial project with an interdisciplinary and long industrial chain. There is no shortcut to the development of industrial engineering. It requires technology accumulation, product iteration, industrial support, and ecological chain coordination. Without any link, there is no way to progress.
Taking product iteration as an example, the optical design of an IP product takes at least one year. After the design is completed, the chip must be produced, tested, and returned to the end user for trial. From small batch trials to large batch trials, this process takes at least two years. “This is still an ordinary chip. If you encounter a complex chip used in a high-performance computing server, it will not take a year to design it.” The above-mentioned chip practitioners said that various problems may occur after this generation of products is produced. Some of them can be solved with software by applying a patch, and some must be modified by hardware, that is, engineering change order (ECO, Engineering Change Order).
“ECO still means that you can find where the problem is. It takes half a year to do an ECO. After this change, and then go to production, it will be almost 9 months after returning.” The chip practitioner said that the frequently encountered problem is, The hardware modification barely passed, but the customer hopes that the next generation of products will no longer have this problem. “To do better, we have to accompany the customer to make another generation, and this generation has passed two or three years. When two or three years pass, it may be A new standard has come out.”
Therefore, IP research and development requires long-term accumulation. Zeng Keqiang also said, “Without more than ten years of mass production, running-in, and iteration with customers, IP research and development is not realistic at all.” Such a leap cannot be solved by spending money.
The best time for accumulation: How to break through the domestic IP?
The global semiconductor IP market continues to grow. The road is long and the road is coming. Where are the opportunities and the way out for domestic IP?
Judging from the development history of the past 30 years, the rapid iteration of technology and the emergence of new market applications have jointly promoted breakthroughs in the IP industry. Zeng Keqiang said that the vigorous digital development of the whole society has made China the world’s largest chip application market, and the combination of digitalization and various industries has also pushed China’s chip application to the forefront of the world.
Chip technology is undoubtedly the underlying technical support for China’s development of the digital economy and industrial digitization. Today, more and more system companies that only used applications in the past also plan to realize product differentiation from the underlying hardware and start to lay out chips.
Zeng Keqiang once served as the deputy general manager of Synopsys China, and has been responsible for the industrialization of EDA and IP in the market for more than 20 years at Synopsys. He keenly captured that China’s semiconductor IP demand “has reached a period of explosive growth”.
“China’s system companies used to focus on making applications, but today they are all entering the market to make chips, including BAT and rising stars such as ByteDance. These companies are very good at defining chip requirements to support the implementation of applications, which has become a future market trend. ” Zeng Keqiang said.
“In addition, many electronic terminal companies have also joined in, the representative ones are Gree, an electrical appliance manufacturer, Xiaopeng Motors, which makes cars, and Xiaomi and OPPO, which make mobile phones. I believe that IP can create new cores for these Power provides tremendous value, but it needs to be underpinned by a new IP design methodology.”
Zeng Keqiang used a camera as an analogy: In the past, most of the photos were taken with SLR cameras, but now basically only enthusiasts use them. Ordinary users can take photos of the same quality as the previous SLR cameras with a mobile phone camera. “Users don’t need to care about the aperture. , the shutter, the mobile phone are all processed, and even the post-production is done.”
The same is true for IP. Zeng Keqiang said that the future IP will not only design small and sophisticated interface IPs, but also make IP subsystems, making it easier for system companies and chip companies to quickly create large-scale systems by integrating IP. Chips to meet the diversified needs of the application market in the digital age.
At the same time, the methodology of designing IP also needs to be innovated, so as to meet the actual needs of new core-making forces. The comprehensive technological innovation in the future is completely different from the previous methodological learning.”
The application market that is at the forefront of the world is an opportunity that can be seen and grasped in this special era, and it is also the only path for domestic IP to change lanes and overtake. Domestic IP companies have not yet “grown in place”, and the market demand has exploded. “The times call us to start a business and set up a Chinese IP company with independent property rights.” In June 2020, Xinyaohui was established in Zhuhai, with a total financing of 1 billion yuan in less than a year, gathering more than 200 domestic and international top talents, 80 % have more than 15 years of work experience in the chip industry.
The chip industry has a clear division of labor and cooperation, and IP should be developed based on the process of chip design companies. At the beginning of its establishment, Xinyaohui developed 14-nanometer and 12-nanometer IP, which is the most advanced technology that can be mass-produced in China. These batches of advanced process IPs have been introduced to the market one after another and have received good responses. “We are also working with domestic foundries to develop IPs for more advanced processes.”
Although the development of the domestic IP industry is still relatively weak at present, Zeng Keqiang firmly believes that now is the best time to accumulate and develop. The Chinese can make IP by themselves, provided that the top talents in China must be attracted.
With more than 30 years of industry experience, Anwar Award (Anwar), who once led Synopsys’ Global IP Solutions Division and served as Global Vice President at Intel, and Yu Chengbin, IEEE Fellow, University of Macau and University of Lisbon, have successively Join Xin Yaohui as global president and co-CEO respectively.
Yu Chengbin is the co-founder of the State Key Laboratory of Integrated Circuits at the University of Macau and the co-founder of Chipidea Macau Microelectronics. After Chipidea was acquired by Synopsys in 2009, he served as IP R&D Director, leading Synopsys’ mixed-signal IP R&D design team in Asia Pacific and Greater China to design and develop advanced interface IP.
According to industry insiders, the University of Macau is world-class in academic research on analog and digital-analog hybrid circuits, with a large number of talents, which is closely related to IP design. The ISSCC (International Solid-State Circuits Conference), known as the “Chip Olympics”, has collected 23 papers from academic and industrial institutions in China (including Hong Kong and Macau) in 2020, including 15 from mainland China, 6 from Macau and 2 from Hong Kong. The 6 papers included in Macau are all from the University of Macau.
“China’s semiconductors have developed for many years and have trained a group of experienced R&D and technical personnel, which provides the necessary soil for the development of IP technology and products.” Zeng Keqiang said, “Our generation of semiconductor people is fully confident to seize this century. It is a once-in-a-lifetime opportunity in the era of digital transformation, but we must use a new technological path to change lanes and overtake and make future-oriented semiconductor technology.”
The Links: NL6448BC18-01B TPS40131RHBR