Shulou(Shulou.com)11/24 Report--

Recently, in order to promote the development and ecological construction of domestic automotive intelligent computing chip technology, Amou Technology (China) Co., Ltd. (hereinafter referred to as "Amou Technology") joined hands with a number of automotive semiconductor industry partners to hold an offline technology salon with the theme of "Vehicular Intelligent Computing" in Shenzhen, and issued a major "Automotive Intelligent Computing Chip White Paper (2023 Edition)" (hereinafter referred to as "White Paper"). The "White Paper" is compiled by Amou Technology in conjunction with Horizon, Core Technology, Core Technology and Zhi Xiehui. From the perspectives of opportunities and challenges of in-vehicle intelligent computing, software-defined vehicles, on-board heterogeneous computing chips, hardware and software co-design, etc., this paper systematically expounds the development status and future trend of the hardware and software platform for vehicle intelligent computing. Combined with the software and hardware solutions and application cases of Amou technology and automotive chip manufacturers in the field of intelligent driving module, it provides an important reference for domestic upstream and downstream industrial chain enterprises to grasp the market opportunity and form the core competition barrier. Zhao Yongchao, head of Amou Technology Intelligent things Union and Automotive Business Line, interpreted the white paper at the event and shared his observations and thoughts on the automotive chip industry.

Figure 1: the launching ceremony of "2023 vehicle Intelligent Computing Chip White Paper"

In recent years, with the rapid development of artificial intelligence, mobile Internet, big data, cloud computing, 5G and other information technology, automobile has gradually evolved from a single means of transportation to a "super mobile intelligent terminal" integrating many high and new technologies. Software and algorithms have become the focus of differentiation competition among major automobile manufacturers. At the same time, as a system engineering across different technical fields, auxiliary driving, autopilot and intelligent cockpit are facing many landing challenges, including complex sensors and control units, real-time processing of massive environment-aware data, incompleteness of AI algorithm model, algorithm evolution cycle faster than chip design cycle, lack of high computing power and low power chip solutions, high difficulty in functional security technology, and so on.

In order to solve this series of technical difficulties and better meet the multiple needs of consumers for automobile intelligence, software, algorithms and electronic and electrical architecture have gradually become the competitive points of enterprise technological innovation. at the same time, it also puts forward higher requirements for the performance and safety of automotive chips. According to the "white paper", software-defined vehicles have become the mainstream trend in the industry, which can not only provide more complex safety protection functions such as safety patches and intelligent driving optimization, but also have a higher level of system autonomy and a higher degree of software compatibility, enabling the automotive industry to achieve new revenue growth through continuous OTA (download over the air) software upgrades while reducing service costs.

In addition, the on-board architecture is also constantly innovating, and the computing hardware unit is showing a trend of centralization and standardization. At present, a large number of mass-produced vehicles adopt the traditional distributed or domain-centralized electronic and electrical architecture, which is expected to evolve towards a central integrated electronic and electrical architecture and further integrate domain controllers in the future. high-performance heterogeneous computing chips that support different workloads are expected to become the core hardware basis for software-defined automotive functions. It is worth mentioning that, different from the demand for intelligent computing chips in general edge computing scenarios, vehicle intelligent computing chips carry most of the key core computing tasks. for example, massive environmental perception data modeling, target object recognition, planning decision control and a large number of deep neural network reasoning as well as complex logic and mathematical operations, and this kind of computing has high real-time requirements.

Based on its accurate insight into the automotive chip market, Amou Technology combines the local innovative self-developed IP with the world's leading Arm IP to create a high-performance fusion computing chip IP platform for smart vehicles. On the basis of general computing IP such as Arm CPU and GPU, heterogeneous integration of self-developed IP such as "Zhouyi" NPU, "Star" CPU, "Shanhai" SPU, "Linglong" ISP and "Linglong" VPU. Through coordination and complementarity, each computing unit forms a complete platform-level solution, which effectively takes into account both versatility and specificity, meets the functional safety requirements of the vehicle specification level in line with international standards, and accelerates the research and development cycle of on-board chips in all directions. help the local automotive semiconductor industry to achieve an intelligent leap forward.

Zhao Yongchao said: "under the current wave of the 'new four modernizations' of the automotive industry, the domestic automotive chip industry has also ushered in a new opportunity for development. In the on-board intelligent computing scenarios such as assisted driving, autopilot and intelligent cockpit, Amou Technology launched a basic hardware platform capable of carrying core computing-the high-performance fusion computing chip IP platform, which provides low-level technical support for the continuous improvement of the performance of automotive chip products. In addition, it also benefits from the in-depth development of cutting-edge technologies such as software-defined automobile, cloud native, and heterogeneous computing, which effectively promotes the adaptive coordination between basic hardware platforms, software and algorithms, and fully tap the potential of computing power. so that the vehicle chip in practical applications to play a more excellent performance advantages. In the future, Amou Technology will continue to bring core competitive product solutions to the domestic automotive chip industry, and work with industry partners to promote the innovative development of on-board intelligent computing. "

Photo 2: Zhao Yongchao, head of Amou Technology Intelligent things Union and Automotive Business Line, delivered a speech.

In addition to the above technology trend analysis, the "white paper" also shares the typical application cases of the leading manufacturers in the industry, including Horizon Expedition 5, Core engine Technology "Fuji No.1", Xichi Technology V9p and other new generation intelligent driving module chips equipped with Amou Technology IP products, as well as Zhixiehui's vehicle and intelligent driving data solutions based on the integrated vehicle cloud data base, from the point of view of technology landing application. It further confirms the advancement and innovation of the concept of software-hardware collaborative optimization technology, provides application paradigm and value reference for the upstream and downstream partners of the domestic intelligent automobile industry chain, and wins the opportunity of "core" of on-board intelligent computing.

Figure 3: "White Paper on vehicle Intelligent Computing Chip (version 2023)"

* download the White Paper at this link: https://www.armchina.com/ResourceLibrary

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.