Embedded ML Inference for Cortex-M Systems
A new class of machine learning (ML) processor, called a microNPU, specifically designed to accelerate ML inference in area-constrained embedded and IoT devices. The Ethos-U55 combined with the AI-capable Cortex-M55 processor provides a 480x uplift in ML performance over existing Cortex-M based systems.
Multiple configurations allow designers to rapidly target a wide variety of AI applications with up to 480x increase in performance.
Ethos-U55 delivers up to 90% energy reduction in about 0.1mm2 for AI applications in cost-sensitive and energy-constrained devices.
A unified toolchain for Ethos-U55 and Cortex-M simplifies developer use and creation of AI applications.
Provides native support for the most common ML network operations, including CNN and RNN, while allowing for future ML innovations.
The combination of world-class hardware IP, easy-to-use tools, open-source software, and a leading ecosystem means the Ethos-U55 microNPU is transforming the future of small embedded and IoT devices. The Ethos-U55 microNPU is a component of a wider solution that includes the following key components to enable the next generation of AI devices:
Discover how the Ethos-U55 microNPU can deliver significant ML in deeply embedded systems.
The Cortex-M processor family is Arm's smallest and lowest power suite of CPUs, providing area and energy efficiency for demanding industrial applications. Cortex-M based processors are at the heart of the sensor hub, delivering advanced signal-processing capabilities to support smart manufacturing.
Mbed OS is an open-source, embedded operating system that includes all the necessary features for the development of IoT connected products, including standards-based security and connectivity stacks, an RTOS, and drivers for sensors and I/O devices.
Cortex Microcontroller Software Interface Standard – Efficient Neural Network Implementation (CMSIS-NN) is a collection of efficient neural network kernels developed to maximize the performance and minimize the memory footprint of neural networks on Cortex-M processor cores.