The Ethos-U55 NPU, is the industry’s first microNPU (Neural Processing Unit) for Cortex-M, and has been designed to deliver a combined 480x leap in ML performance to microcontrollers. The IP and supporting unified toolchain will enable AI hardware and software developers to innovate more, as a result of increased levels of on-device ML processing for billions of small, power-constrained IoT and embedded devices.
“Enabling AI everywhere requires device makers and developers to deliver machine learning locally on billions, and ultimately trillions of devices,” explained Dipti Vachani, senior vice president and general manager, Automotive and IoT Line of Business, Arm. “With these additions to our AI platform, no device is left behind as on-device ML on the tiniest devices will be the new normal, unleashing the potential of AI securely across a vast range of life-changing applications.”
With the roll-out of the IoT, AI advancements and 5G, more on-device intelligence will mean that smaller, cost-sensitive devices can be made smarter and more capable while benefiting from greater privacy and reliability due to less reliance on the cloud or internet. By delivering this intelligence on microcontrollers designed securely from the ground up, Arm is looking to reduce silicon and development costs and to speed up time to market for product manufacturers.
Arm partners are currently shipping more than 50 billion chips based on Cortex-M into a vast range of customer applications. With the Cortex-M55, Arm is now offering its most AI-capable Cortex-M processor to date and the first based on the Armv8.1-M architecture with Arm Helium vector processing technology for significantly enhanced, energy-efficient DSP and ML performance. The Cortex-M55 delivers up to a 15x uplift in ML performance and a 5x uplift in DSP performance, with greater efficiency, compared to previous Cortex-M generations.
Additionally, Arm Custom Instructions will be available to extend processor capabilities for specific workload optimisation, a new feature for Cortex-M processors.
