arxiveess.SYcs.OS2026-07-14
SynapticOS: An Inference-First Runtime Architecture for Neural Processing Units on Resource-Constrained Microcontrollers
Microcontrollers with on-die neural processing units (NPUs) have become mainstream, but the system software hosting them has not: production combinations of Zephyr or FreeRTOS with TensorFlow Lite Micro treat AI inference as an application-layer library, leaving memory fragmentat…