Abstract
Many techniques have previously been proposed for using low-level CPU Performance Monitoring Counters in power estimation models. In this paper, we present some common myths of these techniques, and their potential impact. Such myths include: (1) sampling rate can be ignored; (2) thermal effects are neutral; and (3) memory events correlate well with power. We aim to raise the awareness of these interesting issues, which existing power modeling techniques usually do not address. Our discussions provide some guidance to avoid these myths and their effects through detailed specification of software and hardware configurations.