Right-size your hardware

The choices made in hardware component selection can greatly influence the operational phase impact of devices and therefore must be carefully considered. The use case should dictate the required CPU/MCU, peripherals and communication modules on the device.  When choosing the processor, amount of RAM, and amount of flash storage, the designer must focus on resource optimization; over-provisioning hardware can lead to choosing a processor that has a large power draw but stays mostly idle or has an excess of memory that never gets used, increasing the overall carbon footprint of the device unnecessarily. Under-sizing hardware can lead to long execution times or a lack of extensibility necessary to insure the longevity of the device.

Your design will typically fall into one of three broad classes of devices – Microcontroller (MCU) class devices, Microprocessor (MPU) class devices, and devices that use hardware acceleration for machine learning use cases (which we refer to as "Inference class" devices).