Arm Ltd. claims that chips based on its designs power 85 percent of vehicle infotainment units and 65 percent of driver-assistance systems. Today, the company is moving to grow its presence in the auto market even further with an eye to self-driving vehicles.

Arm has pulled back the curtains on a new chip series specifically engineered with autonomous driving in mind. The first offering in the lineup is the Cortex-A76AE, a central processing unit based on a seven-nanometer process.

The smaller the individual building blocks of a chip are, the better its performance, which gives the new CPU an edge over 10-nanometer and other older designs. Improving transistor density also provides the added benefit of lowering electricity requirements. Power efficiency is a hallmark of Arm chips and a key element of its competitive strategy, a focus that the company has prioritized with the Cortex-A76AE.

The company claims the CPU will enable partners to develop autonomous vehicle systems with a power envelope of as little as a few dozen watts. Existing offerings’ electricity consumption is usually measured in kilowatts, two orders of magnitude higher.

Another key feature of the Cortex-A76AE is a capability Arm calls Split-Lock. The technology takes advantage of the fact that computing systems designed to support autonomous driving, even the limited forms of it that are available today, often include multiple chips. Split-Lock can link two processors with one another and have them carry out sensitive tasks together, which mitigates the risk from technical issues such as a CPU failure.

The Cortex-A76AE can alternatively each operate independently another when running less sensitive workloads. Arm sees the chip’s ability to balance reliability with the requirements of non-critical applications as one of the main draws.

“AE processors will go through a rigorous functional safety process and include safety innovations like dual-core lock-step to ensure redundancy in our CPU cores,” Rene Haas, president of Arm’s Intellectual Property Group, wrote in a blog post. “And as Arm develops more purpose-built automotive compute IP such as ML processors and GPUs, you can be certain the same safety-first principle will guide those designs.”

Haas’ post suggests that Arm might create a vehicle version of Project Trillium, a series of machine learning chips that it unveiled earlier this year. The company has said that the processors in the family will prioritize power-efficiency much like the Cortex-A76AE.

According to Arm, vehicles equipped with chips based on the latter CPU should start hitting the road in 2020. The company is launching a Safety Ready program that will give customers access to software testing tools, best practices and other resources for implementing the processor.  

Photo: Arm