With the advent of the era of electrification, the car will no longer be limited to transportation, but will become a digital platform that can be moved, just like the smartphone. Obviously, car companies are working in this direction. To put it bluntly, software will define the car. Since the future car is like a mobile phone, then the mobile platform has Qualcomm 8Gen1, MediaTek 9000, and what computing platform does the vehicle software rely on to work.
From the current scale, the intelligent driving system for mass-produced passenger cars is at the L3 level and below, and the intelligent driving technology is still being continuously upgraded. Considering that the actual service life of a car is at least ten years. The upper limit of the computing power of the in-vehicle computing platform determines the maximum amount of software service upgrades that can be carried during the life cycle of the vehicle. Therefore, the use of current computing power alone will not cover the technology upgrade requirements in the next decade.
In order to facilitate the understanding of our mobile phones, we still take the example of mobile phones, the chip upgrade of smart phones is often a generation, but the software level upgrade will be dozens of times a year. Then we compare to the use cycle of more than ten years on the car, and the main engine factory has to preset several times the current computing power when it leaves the factory. At present, most of the chip computing power is in the tops of the ten figures, and in order to meet the future, the main engine factory should increase the intelligent driving computing power to the level of 500-1000Tops.
In recent years, the advent of high-computing chips has accelerated the upgrade speed of smart cars, such as Nvidia's Xavier (30TOPS), DRIVE Orin (254TOPS), AND DRIVE Atlan (1000TOPS). According to industry data, the computing power to achieve L3 automatic driving needs to reach 20-30TOPS, and to L4 level it needs more than 200TOPS.
Computing power aside, in-vehicle computing platforms also need to be smarter. For example, intelligent expansion to meet diverse needs. The computing power is scalable, the function is configurable, and the experience can be upgraded to meet the needs of different users and different models, and at the same time, it has the ability to continuously upgrade after-sales service.
Intelligent design with high reliability. Vehicle-grade design and development, ASIL D high functional safety, to meet the high heat dissipation, high shock resistance, high protection and other reliability requirements. Intelligently integrated, software-defined cars. Aurora is an open platform designed for SOA to meet the needs of different software applications for rapid integration and deployment.
In addition, the future computing platform must also have central computing power. Integrating the core functional domains of intelligent cockpit, intelligent driving, and networking services, it realizes the leap from "domain control" to "central computing", which is in line with the evolution direction of the post-Moore era of semiconductors.