Reducing Energy Consumption and Decentralizing Computing through Heat Redistribution

A byproduct of computations performed by semiconductor devices is heat. Many compute intensive functions are currently performed in the cloud by hardware placed in data centers, where heat must be removed at additional energy cost. Granite Mountain Technologies is exploring an energy conserving model and technology for decentralizing computation from data centers to areas where heat is required. This compute model can provide vast energy savings benefits, while also redistributing monetary gains from large organizations capable of running data centers to individuals. We report on a decentralized computing appliance implemented with relative timed integrated circuits in the 12nm technology node. The appliance can accept calculations at a variable rate which allows heat created from computations to achieve the desired temperature of a room. The benefits of applying this approach to residential heating is evaluated.