Temperature Modeling and Control on Multi-Core System-on-Chip

As semiconductor technology continues its marching toward the deep sub-micron domain, soaring power consumption and rising temperature have become major concerns for modern embedded systems design. A series of numerical and analytical system-level power and thermal modeling methodologies have been developed for power and temperature analysis on different system scales and architectures. In this work, we study stable state power and temperature modeling using ZYNQ SoC embedded architecture. First, we compare the power models' accuracy with and without leakage-temperature dependency. Then, we study the single-core and multi-core temperature modeling in the thermal stable state. At last, we validate the theoretical models using deep neural network applications.