the core competitiveness and challenges of future intelligent vehicles and chips

the core competitiveness and challenges of future intelligent vehicles and chips

Smart cars are the future direction of development, so what does the future smart car look like? How much computing power must a smart car have? The rapid rise of "Silicon Valley Iron Man" Musk and Tesla, what is the business logic and development method that applies to their disruptive innovation? How can we seize the opportunity to leverage the trillion-dollar smart car market under the biggest change in a century? In this issue, we invited the founder and CEO of Xinto Intelligence, Mr. Zhang Hongyu, to discuss the industry development and the decisive role of chip rate in the development of smart driving, as well as the challenges and opportunities in the field under the market competition, and we believe that readers can get different inspirations and gains from it.

"How much computing level is necessary for5g gnss future intelligent vehicles under the wave of "new four

The future is the era of intelligent vehicles, there is no doubt about this, but the future of intelligent vehicles to have what characteristics to lead the trend of the times? The vehicle industry feels that there are four recognized trends in the future development of cars: electrification, network connectivity, intelligence, and sharing. Among them, networking depends on high-speed wireless broadband network, and sharing depends on the sharing economy operation platform, both of which do not entirely depend on the vehicle itself, but more on the construction of the overall infrastructure; while electrification and intelligence are more important signs to consider the quality of the vehicle itself, the former is related to the new energy power system, and the latter generally refers to intelligent driving or driverless subsystems - of course, intelligence in the broader sense should never be limited to the role of driving.

So, what level of computing must a smart car be? This depends on how smart cars must function. For now, it is clear that AI-based autonomous driving capabilities (ADAS/AD) are one of the key factors driving the continued growth of computing levels. General analysis feels that the AI computation rate necessary to upgrade from Level-2 to Level-5 (refer to the National Highway Traffic Safety Administration NHTSA website for definitions of different levels of automation) for smart/unmanned driving is growing tenfold. For example, typical Level-2/3 autonomous driving functions may require AI computation levels in the tens to hundreds of TOPS (Tera-Opspersecond), while Level-4/5 autonomous driving functions may require AI computation rates in the hundreds or even thousands of TOPS. In the future, as driverless algorithms continue to evolve, the requirement for AI arithmetic power is likely to continue to grow.

In addition to autonomous driving capabilities, SmartCockpit requirements present another driver for continued growth in computing levels. Customers' continued pursuit of human-computer interaction and the quality of their "content+service" is making the role of the smart cockpit increasingly challenging for CPU and GPU computing levels. Take Qualcomm's Snapdragon SA8155 and SA8195 as an example: both have a CPU rate of 100-150KDMIPS and a GPU rate of 1-1.5TFLOPS, while the next generation Snapdragon SA8295 will have a CPU rate of 220KDMIPS. The next generation Snapdragon SA8295 will have a CPU rate of 220KDMIPS and GPU power of 3TFLOPS. Compared with Qualcomm, another mainstream supplier of smart car chips Nvidia (Nvidia) is more aggressive, its Orin-X chip CPU/GPU arithmetic rate of 220KDMIPS and 5.2TFLOPS each, the next generation of Thor-X in development will reach 600KDMIPS and 9.2TFLOPS each.

But should be seen in the eyes of the smart car managers Tesla (Tesla), Qualcomm and Nvidia's chips are not yet talking about the top of the equipment. Tesla's latest model is equipped with AMD's Ryzen CPU and Radeon GPU, each with 400KDMIPS and 11TFLOPS. AMD's family. It is expected that the new Tesla models will have even more eye-catching CPU/GPU equipment in the future.

With the future 5G popularity to bring content, service upgrades, games, meta-universe and other new applications on board and even the development of intelligent cockpit and intelligent driving combined with its intelligent travel services and the extension of the digital twin edge computing in the cloud, the integrated rate of intelligent car brain will certainly continue to climb and turn into a super computer on wheels!


Related Hot Topic

Describe NMEA GNSS.

Similar to how ASCII is the industry standard for digital computer characters in the world of computers, NMEA is now the standard data format for GPS that is supported by all GPS manufacturers. NMEA was created to enable users of equipment to mix and match hardware and software.

Is it free to utilize GNSS?

But because GNSS signals are now freely available and GNSS receivers are affordable, GNSS technology is now widely used in civil, industrial, scientific, and military fields.

Why is GNSS necessary?

Border security and maritime safety both benefit from the use of GNSS positioning and navigation. Modern ballistic guiding systems heavily rely on GNSS, which is used by European security forces for vehicle and person tracking, navigation, and personnel tracking.

How trustworthy is GNSS?

GNSS/GPS receivers typically have a vertical accuracy that is 1.7 times greater than their horizontal accuracy. A receiver with 1 m 2DRMS horizontal precision, for instance, would probably offer 2 m vertical accuracy. This estimate is based on broad observations of a variety of receivers rather than in-depth testing of any one receiver.

  • TAGS