Power and Temperature Prediction for Computer System Power Optimization
Koji Nishi, Shota Takadai, Takamichi Kanedai · Linköping electronic conference proceedings · 2020
This paper investigates desktop computer system power optimization with Modelica.Firstly, microprocessor power equation is modeled with voltage and temperature dependency.Secondly, steady state heat transfer path is modeled as a variable thermal resistance with fan power change and is connected to microprocessor power model to discuss optimum working point of fan which minimizes power consumption of the power rail for the microprocessor on the target computer system under CPU-intensive workload condition.After that, transient heat transfer path modeling is explored as a Cauer thermal network creation.