Minions: A Dynamic and Distributed Virtual GPU Cloud Built on Idle Apple Silicon Devices
Summary
Minions presents a distributed virtual GPU cloud built from idle Apple Silicon Macs to run large-scale ML inference workloads. It highlights elastic, fault-tolerant GPU utilization across consumer hardware, addressing cost and wait-time constraints of centralized GPU farms. The work demonstrates how existing devices can form a practical GPU-inference platform, with implications for on-device and edge ML deployments.