India, Oct. 15 -- Apple researchers have developed a system that can utilise idle Apple Macs powered by M-series silicon to become a virtual GPU cloud for machine learning inference. The system, called Minions, might be particularly relevant for organisations that can utilise macOS devices across their inventory to create a massive and shared pool of graphics processing units (GPUs) for artificial intelligence (AI) compute. Researchers note that the biggest impact from a system like Minions is that GPU queue times for new requests fell to almost zero, compared to as much as 10 hours in an architecture with centralised GPUs.

In a paper titled 'Minions: A Dynamic and Distributed Virtual GPU Cloud Built on Idle Apple Silicon Devices,' the r...