Microcontroller delivers horsepower for AI applications

Built on a dual-core microcontroller and incorporating an ultra-low-power deep neural network accelerator, the MAX78000 delivers the computational horsepower necessary for high-performance artificial intelligence (AI) applications.

It is ideal for applications including machine vision, facial recognition, object detection and classification, time-series data processing, and audio processing.

The device integrates two microcontroller cores for system control: an Arm Cortex-M4 processor and a 32-bit RISC-V coprocessor.

The RISC-V processor integrates special features to support fast and low-power data loading to the neural network accelerator.

Once configured and loaded with data, the MAX78000's 442-Kbyte-weight convolutional neural network (CNN) accelerator runs AI inferences at less than 1/100th of the power and over 100 times faster than microcontroller solutions.

The MAX78000 processor offers highly efficient power management, maximizing operating time for battery-powered Internet of Things (IoT) devices.

With dynamic voltage scaling, the processor minimizes active core power consumption, while selectable SRAM retention is available in low-power modes.

The processor's combination of low-power performance and highly efficient computing reduces latency by a factor of 100 and enables the execution of AI inferences at the IoT edge.

It is supported by the MAX78000 evaluation kit.

In addition to a MAX78000 processor, the evaluation kit features a digital microphone, gyroscope/accelerometer, parallel camera module support, and a 3.5-inch touch-enabled color TFT display.

Also available from Mouser is the MAX78000FTHR, a development board in the Adafruit Feather form factor, designed for rapid prototyping of smaller system