With the continuous development of very large scale integrated circuits, transistor density has further increased rapidly. With the expansion of chip scale, the difficulty of chip design has continued to increase. Under the influence of many factors, the key aspects of chip design need to use EDA tools to improve design efficiency. , Such as logic synthesis, placement and routing, simulation verification. The evolution direction of chip design has put forward new directional requirements for the development of EDA. However, looking at the domestic market, its main processes and technology platforms are still dominated by international EDA manufacturers. Therefore, as a card-neck technology in the semiconductor industry, the development of the domestic EDA industry has become one of the focuses of the industry.
Driven by this trend, some EDA start-ups have also emerged in China. As a rookie in the local EDA industry, Xinhuazhang released the “EDA 2.0 White Paper” during the 2021 World Semiconductor Conference, hoping to provide a new idea for the development of the domestic EDA industry.
The author believes that through the “EDA 2.0 White Paper”, we can have a deeper understanding of what is changing in the EDA industry. And in the process of transition to EDA2.0, what opportunities can the change of EDA value bring to local EDA manufacturers?
What is the EDA 2.0 era
Before exploring the future development of the domestic EDA industry, we must first understand the changes that are taking place in the EDA industry and what causes this change in the EDA industry. Only by prescribing the right medicine can we fundamentally solve the difficult problems in the development of the domestic EDA industry.
In the “EDA 2.0 White Paper” published by Xinhua Zhang, they divided the development of EDA from the 1890s to today into EDA 1.0 stage and EDA 1.X stage. Among them, the EDA 1.0 stage refers to the EDA established between 1990 and 2003. During this period, integrated circuit design has been basically finalized as an IP-based module and a large-scale RTL cluster design method. After 2003, with the advancement of AI and process technology, the emergence of stacking tools based on EDA 1.0 promoted the arrival of the EDA 1.X stage.
But from EDA 1.0 to EDA 1.X this stage, that is, from 2003 to the present 20 years, although the complexity of the chip has increased by tens of thousands of times compared with the previous 20 years, the cost has increased by 100 times, and the chip process has also evolved to Nano level. But the chip design methodology has not been revolutionary.
With the advent of the Internet of Things and AI era, the market’s demand for customized chips has increased, and customized chips require that the chip design cycle and design costs be greatly optimized on the current basis, and the semiconductor industry chain from design verification to manufacturing has been improved. High demands. Therefore, the existing EDA 1.X is also facing more and more challenges. Xinhuazhang believes that in the post-Moore era, the challenges faced by EDA 1.X are mainly reflected in the following four points:
1. The design cycle is long, unable to meet the needs of rapid application innovation
2. The EDA design process lacks correlation with system-level software and hardware requirements
3. The design investment is large, the cost is large, and the project risk is large
4. Need to build a large-scale team, highly dependent on experience, hard to find talent
These challenges have promoted the upgrading of EDA, and also promoted the EDA industry from 1.X stage to 2.0 stage.
“Chip development is driven by the system. In this case, enabling system innovation is the new task of EDA.” Fu Qiang, vice president of operations at Xinhuazhang, said: “EDA 2.0 is no longer a combination of tools, but a support system + The service platform developed by chip + algorithm + software is the biggest difference between EDA 2.0 and other eras.”
Xinhuazhang believes that the customized chip jointly developed by “system + chip + algorithm + software” determines the competitive advantage of system applications, and the innovation efficiency of the chip has become the focus of attention. The next 10 years will be 10 years when society puts forward demands for faster development of chip technology. EDA tools and methodology need to be fully advanced in order to lower the technical threshold and further improve the speed of chip technology development and the efficiency of innovation.
The critical path to realize EDA 2.0
In other words, in the EDA 1.0 stage, EDA tools designed for general high-performance computing need to be oriented toward vertical industries and develop in a more refined direction. In this development process, a comprehensive advancement of EDA tools and methodology is required, and this means that EDA must start to innovate from the underlying technology.
From the perspective of the goals to be achieved by EDA 2.0, EDA 2.0 can be achieved by meeting the following four points: one is to use tools to fill the gap between software and hardware; the other is an automatic and intelligent process, not just a combination of tools; the third is that it needs to be open. The service-oriented platform provides the possibility of customization and the possibility of open source; the fourth is to shorten the cycle from chip demand to application.
Yang Ye, Director of Product and Business Planning of Xinhuazhang, said: “From the perspective of system design requirements, EDA 2.0 is a process from system down to chip manufacturing and production. From software, this process will eventually return to software. This is a key feature of EDA 2.0.”
As a result, Xinhuazhang also proposed three key paths of EDA 2.0 in its “EDA 2.0 White Paper”, including openness and standardization, automation and intelligence, platformization and serviceization:
Open and standardize. The upstream and downstream industries jointly develop open standards. Based on these open interfaces and standards, customization is demand-oriented, which facilitates the integration of process automation and AI intelligent processing.
Automation and intelligence. The goal of EDA 2.0 is to significantly reduce the proportion of manpower involved in chip architecture exploration, design, verification, layout and routing from the existing EDA 1.0 process, and to absorb past design experience and data into EDA tools to form intelligence EDA design.
Platform and service. Create a new EDA service platform EDaaS based on cloud native software architecture, make deep use of cloud elastic performance, and provide users with nearly unlimited computing flexibility and a more optimized usage model.
It should be noted that upgrading to EDA 2.0 does not mean that the results of the EDA 1.X phase are completely abandoned.
“Some experience standards and some design rules in the EDA 1.X stage will definitely continue to exist in the field of chip design for a long time. It is impossible for chip design to bypass these physical rules and design requirements,” Yang Ye said: “But these things will More distilled from existing tools and experience, into the interior of EDA 2.0, EDA 2.0 will turn experience into a model or algorithm, improve the automation and intelligence of the overall process, reduce the barriers to use and the height to people rely.”
But for EDA manufacturers, there are still huge challenges in applying new technologies to EDA tools. Yang Ye said: “The internals of EDA 2.0 tools will become more complicated. EDA companies have to complete intelligent tasks in EDA tools. But for users, EDA tools just become simpler and more usable, becoming a kind of More inclusive technology.”
In addition, what we need to focus on is the critical path of openness and standards mentioned in the EDA 2.0 white paper. According to the author’s understanding, the open EDA mentioned here does not simply refer to open source EDA. Xinhua Zhang believes that the opening and standardization of the EDA industry in the future will not only be determined by EDA manufacturers or standardization organizations, but should be defined by the EDA ecosystem in the upstream of the industry chain and the downstream industry-the entire industry from system manufacturers, chip manufacturers to EDA manufacturers Ecology to jointly develop open standards. Based on these open interfaces and standards, EDA vendors, users, and third parties can all be demand-oriented to customize, facilitating the integration of process automation and AI intelligent processing.
Behind the path of openness and standardization, there are opportunities for the development of the domestic EDA industry.
As we all know, the lack of full-process EDA tools is one of the pain points for the development of domestic EDA companies. It is precisely because domestic manufacturers do not have tools for the whole process, it means that the development of domestic EDA tools requires cooperation between local manufacturers, and this is also the birth of openness and standards.
Chen Lan, director of the EDA Center of the Chinese Academy of Sciences, also stated in the EDA 2.0 white paper press conference: “Opening EDA can meet new technologies such as Chiplet and AI. EDA 2.0 has no historical burden. Therefore, new technologies can be quickly introduced into EDA tools. , And allow customers to quickly verify. For China, open EDA is a product with traction technology, especially when there are only a few tools, open EDA tools can help companies quickly establish system competitiveness.”
The core value of EDA enterprise in the EDA 2.0 era
While the industry is upgrading to EDA 2.0, the core value of EDA companies has also changed.
Yang Ye said that the key to industrial software lies in an in-depth understanding of vertical fields and putting this understanding into the overall framework of the system. Therefore, EDA tools are no longer worthy of use as a single point tool. The combination of EDA with various platforms and the cloud creates an intelligent and automatic process, which will play a role far beyond the software itself.
Fu Qiang said: “The true core competitiveness of EDA companies in the future is to embrace the needs of system companies.”
At the same time, due to the increased demand for customized systems, this will also prompt the entire industry to iterate, rather than just relying on the iterative tools of the EDA enterprise itself. From this perspective, the iteration of the EDA industry will definitely accelerate, and after the acceleration of this demand to the iteration of tools, it will inevitably accelerate the design and manufacturing of chips. This has also expanded the EDA market space. For domestic EDA manufacturers, this has brought more opportunities for the industry to change lanes and overtake.
He said: “When we embrace such a new scientific research paradigm and new industrial changes, our new enterprise will have more development prospects.”
Xinhua Chapter promotes the development of EDA 2.0
While doing a good job in the development of real products, Xinhua Zhang also drew a very feasible blueprint for the development of the domestic EDA industry. At the same time, Xinhua Zhang also formulated a clear time point for this blueprint. Xinhua Zhang believes that the first year of EDA 2.0 will happen in 2026.
In order to promote the EDA 2.0 era, Xinhuazhang has also done a lot of work.
Fu Qiang said: “Technological development and innovation require precipitation and preparation. In addition, a new technological ecology needs to be cultivated. Therefore, based on our 20 to 30 years of experience in this industry, this process will take about 5 years.” And this new ecology requires consensus and collaboration between the upstream and downstream of the industrial chain.
Based on EDA 2.0, it will no longer be a combination of tools, but an autonomous and intelligent process and a service-oriented and customizable platform to enable more efficient and agile system innovation. Xinhua Zhang pioneered the chip design platform service model-EDaaS (Electronic Design as a Service). And based on this, we started the era of EDA 2.0.
According to Xinhua Zhang’s plan, EDaaS can provide professional consulting or design services, and can also provide customized services based on cloud platforms and open data. The goal of EDA 2.0 is to allow system engineers and software engineers to also participate in chip design, solve the problems of difficult design, fewer talents, long design cycles, and high design costs, and use intelligent tools and service-oriented platforms to shorten The cycle from chip demand to application innovation meets the diversified needs of system applications in the digital world for chips and empowers technological progress.