Accelerating towards the future, Arm automotive electronics solutions pave the way for large-scale application of autonomous driving

Shu Jie, Senior Automotive Marketing Manager, Arm China

"As we all know, Arm is the IP supplier behind the chip. Currently, Arm has more than 1,500 IP licenses worldwide, and the shipments of chips based on the Arm architecture have reached 125 billion pieces. In 2017 alone, the shipments of chips based on the Arm architecture exceeded 21 billion pieces." At the beginning, Shu Jie used a set of data to highlight the huge size of the Arm ecosystem. "The authorized partners include more than 500 industry leaders, startups, chip companies and OEM manufacturers . In addition, 16 top automotive chip manufacturers have authorized Arm's IP."

In 1996, Arm officially entered the automotive field, starting with the IP of brake control chips, and has developed for more than 20 years. At present, in the IVI field, 85% of processors use the Arm architecture ; in the ADAS field, more than 65% of application processors use the Arm architecture . From traditional power system control, chassis control and body control to ADAS/IVI, control chips are using the Arm architecture. All electronic devices use a common architecture, supplemented by the strong support of the leading tool ecosystem, so that automakers and suppliers can quickly innovate hardware and software. These innovations will revolutionize safety, reduce energy consumption, and improve the driving experience.

When Arm develops IP, it needs to have a longer-term consideration for industry applications. Therefore, it is also deploying IP required by higher-level autonomous driving computing platforms to provide better chip support for autonomous driving. In the process of developing IP, Arm needs to sort out and respond to more industry needs in order to develop reasonable IP to meet market demand. Shu Jie said: "At present, Arm has collaborated with partners such as chip manufacturers, operating system suppliers, and application software developers to build a complete ecosystem based on Arm, so that it can serve Tier 1 suppliers and car manufacturers well."

The automotive industry has been talking about the need for ECU (electronic control unit, also known as "driving computer", "on-board computer", etc.) consolidation for many years. But the discussion usually focuses on physical space, hardware costs, the weight of ECUs, and the length of wiring required to connect them together. Broadly speaking, consolidation is driven by cost, flexibility, energy efficiency, and evolving network architectures. Consolidating functions into fewer ECUs simplifies the network of controllers in the vehicle and facilitates communication between different systems. Currently, there are more than 100 ECUs installed in cars - each with a processor, which realizes intelligent control through distributed sensors and processors. By 2020, there may be more than 200 sensors in a car.

Shu Jie pointed out that from the development trend of autonomous driving, most of the autonomous driving levels before 2019 were L1, L2, and at most L3. At present, some companies are making L3+ prototypes, and the first batch of L3+ mass-produced models are expected to be launched in 2025. The truly large-scale deployment of L3+ requires the computing platform to have the computing performance of a server , the power requirements of a desktop , and the heat dissipation characteristics of a notebook . It cannot be a server located in the trunk. In the autonomous driving scenario, from sensing, perception, decision-making to execution, the required computing power demand is increasing. Cars are already running on code. The Boeing 787 Dreamliner has about 14 million lines of code, while L5-level autonomous driving may require more than 1 billion lines of code.

At the same time, the various functions of self-driving cars require many control units to control them, and the control units have different computing power requirements for different places. The control computing power of the cockpit instrument needs to reach 50K DMIPS, while the computing power of the chassis control needs 15K DMIPS, and the computing power requirement for semi-autonomous driving is about 350K DMIPS. Therefore, the computing power requirements of different control units in the car are different, and the overall trend is rising. If you reach a higher level of autonomous driving, the computing power requirements will be even higher.

Therefore, for the future development of the automotive industry, Arm not only hopes to improve computing power, but also hopes to fully consider functional safety from the chip IP level to help partners or chip designers reduce time and cost for subsequent work. Arm's Safety Ready program gives customers a first-mover advantage in safety. The security package provided by Arm includes leading security features and technologies, certified software components and tools, as well as robust methodology and certification support.

In September 2018, Arm launched the first autonomous driving processor with integrated functional safety, the Cortex-A76AE , designed specifically for the automotive industry. The chip is equipped with Split-Lock technology, a disruptive safety innovation that is implemented in an application processor for the first time. AE stands for "Automotive Enhanced", and any Arm IP with the AE logo will include specific functions to meet automotive applications.

However, for autonomous driving, the amount of real-time data collected and processed is much greater than that of ADAS. Therefore, in December 2018, Arm launched another IP for high data throughput computing, the Cortex-A65AE , for the automotive electronics market, mainly optimized for 7nm. Its biggest feature is that it supports SMT multi-threading, and its performance throughput is 3.5 times higher than the previous generation. It is expected to be launched in 2020. According to Shu Jie, some customers have already purchased Arm's IP chip, mainly for applications that require high data throughput computing capabilities.

Moreover, from perception to decision-making, Arm's Cortex-A, Cortex-R and Cortex-M processor series will each be responsible for different areas in autonomous driving. Cotex-A is responsible for ADAS, part of the center console, GPS, and DCM. It is a high-performance processor that emphasizes computing. Cortex-R is an embedded product. EPS, ABS, power output, battery management and other parts in the vehicle that require high stability are all handled by Cortex-R. Airbags, GPS, EPS and other devices require the assistance of Cortex-M processors, which are small and low-power products.