Silicon Foraging: Harvesting Excess Compute for Sustainable Edge Computing

The proliferation of the Internet of Things (IoT) has created a vast source of untapped computational capacity. We demonstrate that typical consumer IoT microcontrollers (MCUs) utilize only 4% of their computational capacity under normal workloads. We introduce silicon foraging - the practice of harvesting excess computational capacity from deployed IoT devices to offset demand for additional hardware.To evaluate the over-provisioning of compute, we develop CrumbMark, a benchmark for measuring computational headroom in deployed IoT devices. We demonstrate silicon foraging with an implementation of an IoT cluster that uses six smart plugs as a distributed web server while maintaining their original functionality. Our evaluation shows that this approach can effectively serve web content with negligible energy overhead, suggesting new possibilities for sustainable edge computing using existing infrastructure. This work provides designers with tools and methods to identify and utilize computational resources already embedded in homes and workplaces, potentially reducing the need for additional hardware.