Arm has recently announced a new addition ‘Arm Cortex-A65AE’ into its ‘Automotive Enhanced’ chips portfolio. It is a high-bandwidth, low-latency processor specially manufactured to handle the high-throughput needs of automotive sensor data safely.
While speaking to media, Lakshmi Mandyam, the automotive VP of Arm said that with the cars attaining higher levels of autonomy, the number of sensors meant to collect data, like radar, cameras, and LiDAR could witness immense growth. It is basically designed to optimize the processing of the multiple streams of sensor data received from next-gen vehicles.
The features, including augmented reality displays, optimized maps, and alerts are going to ensure that more data gets displayed to the driver in real time. Mandyam added that for the apt data delivery to drivers, safe information processing is required especially at the hardware level. The new addition Arm Cortex-A65AE makes this happen through its multithreading capability linked to the ‘split lock’ technology of the arm. With split-lock competences, the processor can be locked. This will deliver multiple cores running similar instructions. Otherwise, the processor can be separated for enhanced multi-core performance.
The new chip has followed the release of Cortex-A76AE, which was disclosed previously this year. It was the first in the Automotive Enhanced IP by Arm. Previously, the company also launched its ‘Safety Ready’ program to ensure the automotive OEMs possess the tools required to integrate safety mechanisms seamlessly into vehicles.
While explaining the working of Cortex-A76AE with the new Cortex-A65AE, Mandyam added that when you consider the entire chain of processing in ADAS or autonomous workloads, the step of collecting data from the sensors comes first, then you have to perceive the message it’s trying to give you, and finally come up with a course of action.
The Cortex-A65AE is a best pick for collecting data from sensors, owing to its high throughput. Whereas, an OEM may rely upon combining both chips for perceiving.
The Cortex-A76AE is apt for the third and the final step.
Mandyam mentioned that a wide range of calculations are required to cater to the needs of tomorrow’s vehicles, and one size isn’t going to fit all, in terms of compute powering such vehicles.