Detailed explanation of the disadvantages of RISC-V on cars

As the penetration rate of electrification, networking and intelligence in the automotive industry continues to increase, the importance of automobiles in the planning of semiconductor manufacturers has increased year by year. In the context of sluggish consumer electronics, in addition to traditional automotive chip manufacturers, more manufacturers are now targeting automobiles as their largest target market, among which domestic MCU manufacturers have taken the most obvious actions.

In terms of automotive MCUs, data from market analysis agencies have also given manufacturers confidence. What has attracted more attention is that IC Insights has pointed out the high growth of the automotive MCU market in two consecutive related reports. In the 2021 report, IC Insights pointed out that automotive-grade MCU sales are expected to surge 23% in 2021 to US$7.6 billion, a record high. Subsequently, automotive-grade MCU sales will increase by 14% in 2022 and 16% in 2023; in the latest report in 2022, IC Insights once again mentioned that global MCU sales will increase by 10% in 2022 to a record high of US$21.5 billion, among which the growth of automotive MCUs will exceed that of most other terminal markets.

However, our topic today is not about automotive MCUs, but about what opportunities and challenges RISC-V, which has initially proven itself in the MCU market, will face in the automotive market. This article is actually a follow-up to my previous article, "RISC-V has great potential for automotive use, but also great difficulties", which mainly analyzed the potential of RISC-V in the automotive field, but did not focus on analyzing the advantages and disadvantages from an industrial level, and did not systematically look at the current status of the industry. These are the main points of this article.

RISC-V targets the automotive sector

At present, in the RISC-V field, both at the core level and the chip level, there have been trends of entering the automotive field.

In terms of core, Xinlai Technology CEO Peng Jianying previously stated at the RISC-V China Summit that Xinlai Technology is preparing to launch automotive-grade RISC-V cores when sharing the company's product route advancement. In the relevant reports on Xinlai Technology's new round of financing, it was also mentioned that Xinlai Technology's CPU IP with Security and Functional Safety features has entered the customer integration verification and standard certification stage in the orderly product landing.

It has also been reported that Xinlai Technology has become the founding unit of the "Hubei Automotive-Grade Chip Industry Technology Innovation Consortium". This innovation consortium is mainly committed to promoting breakthroughs, transformation, and application of automotive chip results, and helping the automotive industry to develop in a high-quality and stable manner.

Andes Technology has currently launched the automotive functional safety IP core - N25F-SE, and according to the plan, it will complete ISO 26262 ASIL-B certification in the first half of this year.

Previously, domestic MCU manufacturer APT and Alibaba's Pingtou Ge further reached in-depth cooperation, and the two parties plan to launch six RISC-V chip series products in the next year, mainly targeting the automotive field. If the two parties work together to create automotive-grade MCUs, then Pingtou Ge is likely to supply automotive-grade RISC-V cores to APT, and Pingtou Ge's R series itself is aimed at industrial control vehicle scenarios.

As for international manufacturers, Codasip, which provides leading RISC-V processor IP and advanced processor design tools, also regards the automotive field as an important direction. Jamie Broome, vice president of Codasip's automotive business, said that in the automotive field, Codasip is investing a lot of energy and money to ensure that developers are confident in the quality of RISC-V processor IP, because this is crucial to automobiles.

In terms of MCU product dynamics, many manufacturers have already released their own automotive-grade RISC-V products.

On December 17, 2021, at the first "Dishui Lake China RISC-V Industry Forum", Wang Zhenshan, vice president of Lingsi Microelectronics, introduced the company's automotive-grade wireless MCU series products based on RISC-V architecture and BLE5.0/1 specifications - the LE503x series. According to him, this series of products is mainly used in body control, instant messaging, in-vehicle entertainment and other aspects.

Founded in 2022, Silicon Labs is a chip design company focusing on automotive-grade MCUs and MPUs. Its products cover a full range of automotive-grade MCUs/MPUs and domain controller SoC chips, including body control, motor control, chassis control, instrumentation, vehicle networks, and smart cockpits. Recently, the company has just completed an angel round of investment of tens of millions of yuan.

In addition to Apt and Lingsi Microelectronics, domestic manufacturers that have launched RISC-V architecture-based MCUs include Qiying Tailun, Pioneer Semiconductor, China Mobile Xinsheng, Philips, Ruisi Semiconductor, TimeQing Technology, China Science and Technology Wuqi, Tailing Microelectronics and other companies. They are all potential stocks for RISC-V to enter the automotive market through MCU.

Judging from the domestic market situation, RISC-V is ready to enter the automotive market.

Advantages of RISC-V in cars

Previously, in the article "RISC-V has great potential in cars, but also great difficulties", the author has already pointed out that RISC-V and software-defined cars are very compatible. So, let's take this point of view a step further and look at the advantages of RISC-V under the concept of software-defined cars.

Software-defined car is a concept proposed by Wang Jin, senior vice president of Baidu and general manager of autonomous driving division. The core idea is that software technology with artificial intelligence as the core will determine the future of cars. Looking at Baidu's approach, they have extracted cars from the traditional framework. Instead of differentiating and defining them based on traditional core areas such as power and chassis, they have added a layer of intelligent functions to the upper layer to control the hardware below. These hard disks include traditional automotive hardware and various electronic components under the "three transformations" trend. These intelligent functions are actually various software algorithms that realize functions such as high-definition maps, assisted driving, and smart cockpits.

Under this re-established architecture, the variables start from the upper-level algorithms. Every time an algorithm function is added, the hardware carrier platform below needs to make corresponding changes and adjustments. If we remove these concepts of automotive grade and consider the closed nature of the automotive field, then the software-defined car is actually an AIoT with the car as the core, which is exactly what RISC-V is good at. Its openness, modularity and scalability allow it to better adapt to fragmented needs. From this perspective, the ARM architecture cannot compare with RISC-V in terms of innovation and iteration.

Another obvious advantage is the time advantage. We all say that Chinese manufacturers are an important force in the development of RISC-V, and domestic MCU manufacturers are currently upgrading from consumer-grade to industrial-grade and automotive-grade. Although some manufacturers have launched related products based on the ARM architecture a long time ago, it is obvious that these chips cannot compete with international manufacturers, so the shipment volume is very small, especially in the original equipment market. The time when most domestic MCU manufacturers entered the automotive market is not much different from that of RISC-V, which gives RISC-V more opportunities and provides an option to overtake in the lane. RISC-V and domestic MCU manufacturers have an opportunity to achieve mutual success.

Disadvantages of RISC-V

As mentioned above, time is an advantage of RISC-V. If the foothold is changed, time will also be a disadvantage of RISC-V in automotive applications. For example, the article "RISC-V has great potential for automotive applications, but also great difficulties" has mentioned that the automotive industry has formed a core circle of BIG-level manufacturers. Tier 1 and vehicle manufacturers will give priority to the products of these manufacturers for stability reasons. As we have also mentioned, these manufacturers currently have no plans to deploy automotive-grade RISC-V. The time disadvantage is almost the same for domestic MCU manufacturers and RISC-V, which also makes them more likely to unite.

From the characteristics of the automotive industry, the defects of RISC-V itself will be magnified in this field. Relatively speaking, the automotive industry does not like "new things" very much. In addition, more than one practitioner has mentioned that RISC-V still has a lot of room for improvement in terms of architecture, which will be considered a potential risk in the automotive field and is not allowed.

At the same time, we mentioned above that the future is software-defined cars, and software support is precisely the shortcoming of RISC-V, which actually returns to the issue of ecology. RISC-V's open source is at the instruction set level, but without a software ecosystem, chips based on RISC-V are useless in any field. Previously, industry insiders said that RISC-V-related companies cannot always want to free ride instead of actively investing in the construction of the ecosystem. Well, in the automotive field, it is actually the same. Who will build the upper-level software suitable for RISC-V, and who can make the flexibility of RISC-V be most fully released in the automotive field. At this stage, there are no answers to these questions.

postscript

Through the analysis of the two articles, we can see that RISC-V has potential in the automotive field, which is in line with the development trend of software-defined cars in the future automotive industry. And from the perspective of time nodes, RISC-V and domestic MCU manufacturers, as latecomers, can naturally combine together to achieve overtaking in architecture and market share. From the core and product levels, it can be seen that domestic manufacturers are actively promoting RISC-V to enter the automotive market.