What kind of chips are needed for the 100-billion-dollar smart cockpit market?

Driven by the upgrade of electronic consumption demand and the enablement of ICT technology in cars, cars are no longer just a mechanical travel tool, but have evolved into a smart product. In the process of this new automobile industry transformation and product form reshaping, new demands have also been placed on chips.

Therefore, film manufacturers also responded to the call. On December 14, at the "Digital Intelligent Low-Carbon Future" OktoberTech Infineon Eco-Innovation Summit, Infineon Technologies Global Senior Vice President, Greater China President, Infineon Technologies Greater China Power and Sensing System Division Director Pan Dawei introduced many new products launched by Infineon in the automotive semiconductor field, such as XENSIV™ sensors, AURIX™ MCUs, HYPERRAM™ memories, and IGBT/PMICs.

In addition, he also said that in order to cope with the current rapidly developing new energy and electric vehicle markets, Infineon's Industrial Power Control Division launched the "Short Flow" business model. The so-called "Short Flow" refers to providing customers with customized topology and optimized chip layout according to specific application scenarios. It can also add, remove or replace internal components and improve thermal performance to make the overall solution perfectly match the application needs. .

Pan Dawei, Global Senior Vice President of Infineon Technologies, President of Greater China, and Head of Power and Sensing Systems Division of Infineon Technologies Greater China

New trends in automotive grade chips

With the acceleration of electrification, networking and intelligence, automotive chips are widely used in many fields such as power systems, bodies, cockpits, chassis and safety. They are mainly divided into computing control chips (mainly functional chips MCU and main control chips SOC ), memory chips, power semiconductors (mainly IGBTs and MOSFETs), communication chips, and sensor chips (mainly CIS, MEMS, gyroscopes, etc.).

At the summit, Han Sanchu, chief expert of Changan Automobile Co., Ltd., shared the new trend of automotive chips:

First of all, the electronic and electrical architecture of automobiles has evolved from distributed to domain controller architecture, and is gradually evolving into a centralized architecture. The number and structural layout of chips have also changed accordingly. In order to meet the needs of personalized and differentiated requirements, Automotive product strategy, to achieve the goal of serialization and industrialization of smart car platforms, the standard series of low-power, low-cost control chips in the vehicle control domain will gradually become the mainstream of the industry.

Secondly, with the service-oriented end-to-end piloting and driving of complete vehicles, the continuous improvement of user experience in cockpit control integration, central computing and other technologies will promote the application of higher-performance computing chips in the new automotive field.

Data shows that the L3 level is the watershed in computing power requirements. The required AI computing power is roughly 100TOPS, the computing power required for L4 will reach 500TOPS, and the L5 computing power requirements are more stringent and are expected to exceed 1,000TOPS. Of course, this depends The degree of optimization of each party’s algorithm.

In addition, since chips are spread all over the body of new cars, next-generation cars require future chips with higher reliability, stronger performance, and stricter management systems. Compared with consumer electronics, the application environment of traditional car-grade chips is harsh. Its operating temperature ranges from minus 40 to 155 degrees, and its target working life is as long as 15 years. The functional safety and reliability certification standards for car-grade chips are strict, and the target implementation efficiency is zero. , and the iteration speed of high-performance computing chips brought about by autonomous driving is much faster than that of traditional control chips.

Multi-scenario processing capability becomes key to smart cockpit chips

The process of automobile intelligence mainly includes autonomous driving and smart cockpit. Among them, because the cockpit is the product and interface closest to the user, and is also regarded as the traffic entrance of the smart terminal of the car, it has attracted much attention in the topic of car intelligence in recent years.

According to Hong Kong-based ICV Tank, China's smart cockpit market value will grow to 107.2 billion yuan ($15.9 billion) in 2025, 2.14 times the 2020 level.

In the process of making car cabins smart, LCD instruments have begun to replace mechanical instruments. Large and multi-screen central control screens have gradually become standard equipment. HUD has been promoted at an accelerated pace. At the same time, cockpit entertainment systems have been continuously enriched, and many applications such as navigation, games, and lifestyle are gradually equipped. In vehicle-mounted systems, the interaction method is gradually turning to voice interaction.

"The current car buying crowd is getting younger and younger, pursuing convenience and embracing intelligence, which has given rise to the emergence of demand for all scenarios." said Xiong Jian, senior director of the Automotive Electronics Division of MediaTek Technology Co., Ltd.

In the process of communicating with car companies, he found that "in the past few years, users' needs have been two-band, three-band, and four-band, but the future trend is five-band, six-band, seven-band, or even eight-band, and even 8K, The demand for 120HZ ultra-high-definition screens is also increasing. At present, more than ten cameras may be needed in the future. At the same time, the corresponding SP processing capabilities must be strengthened, and the sound processing capabilities-DSP capabilities must be enhanced. The ability to have multiple scenarios becomes a very important parameter.”

In this context, how do chip companies respond? In this regard, he said that from a chip perspective, supporting multi-scenario capabilities actually requires the chip to have very, very powerful parallel processing capabilities, which requires the chip to have high computing power. The computing power of CPU, GPU, and NPU is becoming more and more powerful. High, of course the cost must be controlled, the capability must be strong, and the specifications must be high, but the cost must be controlled.

In this process, there are two challenges: one is how to truly and effectively apply chip computing power, and the other is how to ensure system security, such as how to protect users' personal privacy, important data security, and future payments. The security of digital renminbi, etc.

He revealed that MediaTek is working with partners to develop some new executable environments and trusted virtualization, which can achieve system security to the greatest extent and provide users with an easy-to-use, powerful, multi-functional and highly secure system. cockpit environment.