SpotLight: A Hotspot-Greedy, Light-Weighted, and Automated Thermal Modeling Framework for Early Smartphone Design

This paper introduces SpotLight, a light-weighted, automated framework that employs a hotspot-greedy approach for rapid thermal analysis of smartphones during the pre-silicon stage. Additionally, we utilize superposed models and a specially designed least penalty problem to enhance accuracy. Experimental results demonstrate that SpotLight can precisely simulate system-level hotspot temperatures in smartphones, achieving an average runtime of 0.56 ms.