British semiconductor firm ARM, ahead of its impending $40 billion acquisition by US chip giant, NVidia has announced new safety capable computing solutions aimed at a variety of autonomous systems, including vehicles and industrial machinery. The list includes Cortex-A78AE CPU, Mali G78AE GPU and the Mali C71AE ISP.
ARM said its new chips would “accelerate autonomous decision-making with safety capability across automotive and industrial applications” in a press release.
ARM describes its new chip as “highest performance CPU with safety”. ARM Cortex A78AE comes powered with the company’s new hybrid technology which allows the cores to run independently in either split mode or in lock mode. The split mode has a low safety requirement, while lock mode has higher security. Note that lock mode offers higher security but that slows down performance. ARM claims that even after this, this chip offers a 30% increase in performance compared to its predecessor.
ARM described Mali G78AE GPU as its “first safety capable GPU.” It comes with four fully independent partitions for workload separation for safety use cases. The company claims it would enable GPU resources to be utilized for safety enabled human-machine interfaces or for the heterogeneous computing needed in autonomous systems.
Just like the Cortex CPU and Mali GPU, the Mali C71AE image signal processor comes with a focus on safety. The company claims that the chip offers the flexibility needed to support both human and machine vision applications. Some of such machine vision applications are production line monitoring and ADAS camera systems.
Mali C71AE comes with enhanced safety features. It is an ASIL B / SIL2 safety standards complaint. This image signal processor offers support for as many as four real-time cameras or 16 buffered cameras. It delivers 1.2 gigapixels per second output.