Arm on Tuesday announced the latest addition to its portfolio of "Automotive Enhanced" chips. The Arm Cortex-A65AE is a high-bandwidth, low-latency processor designed to safely handle the high-throughput requirements of automotive sensor data.
As cars achieve higher levels of autonomy, there will be a significant increase in the number of sensors collecting data, including cameras, radar and LiDAR. "This is going to drive a hockey stick increase in throughput requirements... to process this data," Lakshmi Mandyam, Arm's automotive VP said to reporters.
Meanwhile, inside the car, there will be an evolution of cockpits. Features like augmented reality displays, alerts and improved maps will offer more data displayed to the driver in real time. To deliver that data to drivers, "you need safe information processing right down to the hardware level," Mandyam said.
Also: The 10 skills you need to land a job working on autonomous vehicles TechRepublic
The Arm Cortex-A65AE delivers this via its multithreading capability combined with Arm's "split lock" technology. With split-lock capabilities, the processor can be locked for safety, delivering two cores running the same instructions. Alternatively, the processor can be split for higher multi-core performance.
The new chip follows the release earlier this year of the Cortex-A76AE, the first of Arm's Automotive Enhanced IP. Arm also earlier this year unveiled its "Safety Ready" program to provide automotive OEMs with the tools needed to more easily integrate safety mechanisms into vehicles.
Mandyam explained how the Cortex-A76AE would work with the new Cortex-A65AE:
"When we think about the whole chain of processing in ADAS [advanced driver assistance systems] or autonomous workloads, you need to first gather data from sensors, then perceive what it's trying to tell you, and plan a course of action," she explained.
The Cortex-A65AE is suited for gathering data from sensors, thanks to its high throughput. Meanwhile, an OEM could use a combination of both chips for the second step -- perceiving. The Cortex-A76AE is well-suited for the third step.
"In my regular conversations with leading OEMs and Tier Ones, it's clear that a broad range of compute is required to meet the needs of tomorrow's vehicles, and one size will not fit all when it comes to the compute powering these vehicles," Mandyam said.
Previous and related coverage:
Hyundai and Kia car owners will able to use Google's Android Auto to send texts via voice and use Kakao's navigation services.
Tesla vehicles made in China will reportedly only be sold to local consumers.