New Automotive SoCs Provide a Window to ADAS TrendsFebruary 16, 2021 by Jake Hertz
Taking a survey of new automotive-facing chips can tip us off to the design trends ADAS may take in the future.
While the automotive industry has been particularly hard-hit by ongoing chip shortages, chip innovation has not slowed down. Semiconductor companies continue to pour resources into IC design in the automotive space as the demand for fully autonomous and fully electric vehicles continues to grow.
As cars become smarter, they depend more heavily on ICs. Image used courtesy of Siemens
Three companies that have recently made headlines in the world of automotive semiconductors include Microchip, Renesas, and Socionext. In this article, we’ll explore each announcement to get a better understanding of what automotive design trends are coming down the pike.
Microchip Simplifies Ethernet Audio Video Bridging
Ethernet is an old technology that has found a new purpose in today’s increasingly “smart” automobiles. Many of the information, entertainment, and ADAS system components in a car today are connected using the Ethernet protocol.
Ethernet is among the many communication protocols utilized in automotive solutions. Image used courtesy of Renesas
Microchip is looking to simplify the development of these systems with its newest automotive IC: a so-called fully-integrated solution for vehicle Ethernet audio-video bridging (AVB). The new chip, the LAN9360, is a hardware-based audio endpoint, working as a controller that interconnects audio between the devices in a car’s “infotainment” system.
Capable of fully supporting Ethernet AVB, the chip is unique in that it doesn’t require any software from third-party stacks, meaning developers can assume more control over their products and develop them quicker.
In regard to time to market, Microchip's VP of automotive Matthias Kaestner explains, “In today’s rapid-pace design environment, this out-of-the-box device gives engineers a quick start to development and allows them to avoid months of engineering work and technical risks involved in coding or engaging third-party integrators.”
Renesas’ Updated R-Car V3H SoC
Renesas made headlines this week when it announced updates to its R-Car V3H SoC for improved on-vehicle machine learning applications. Renesas says the SoC builds on the previous IP, enhancing CNN processing performance by four times compared to the previous generation.
Leveraging a proprietary CNN-specific accelerator, the SoC can provide up to 7.2 TOPS. The company claims the SoC also integrates video processing and image recognition IP while meeting many levels of ASIL compliance.
Block diagram of the R-Car-V3H. Image used courtesy of Renesas
According to the press release, the SoC is optimized for intelligent driver systems such as driver monitoring, park assist, and occupant sensing. These features are especially valuable as the ADAS market estimated to reach $62.7 billion by 2024. ADAS technology gives rise to more smart-camera applications, which leverage advances in camera hardware, machine learning, and computer vision.
Socionext Scales to 5nm
Finally, Socionext, a custom SoC company, has announced that it will be using TSMC’s 5nm process technology for its next generation of automotive SoCs. “Innovative semiconductor technology is essential for accelerating the development of next-generation smart automobiles,” notes TSMC's Dr. Cliff Hou.
It's for this reason that Socionext has tapped TSMC's new N5P family as the basis of its ADAS chips. This specific branch of TSMC's 5nm technology is said to feature reduced defect density and offers a 20% boost in performance, consumes 40% less power, and adds 80% logic density over the previous N7 technologies.
Use cases for Socionext's automotive SoCs. Screenshot used courtesy of Socionext
Socionext isn't the only company with its sights on the 5nm process node, either. Last summer, NXP also announced that it would be using TSMC's 5nm technology for its automotive computing platforms.
The Stretch for Level 5 Autonomy
From recent news, we can see that simplifying design, decreasing time to market, and improving processing capabilities, specifically for machine learning, are a few of the key priorities for automotive chipmakers as the quest for level 5 autonomy continues.