Software energy estimation based on statistical characterization of intermediate compilation code

Abstract

Early estimation of embedded software power consumption is a critical issue that can determine the quality and, sometimes, the feasibility of a system. Architecture-specific, cycle-accurate simulators are valuable tools for fine-tuning performance of critical sections of the application but are often too slow for the simulation of entire systems. This paper proposes a fast and statistically accurate methodology to evaluate the energy performance of embedded software and describes the associated toolchain. The methodology is based on a static characterization of the target instruction set to allow estimation on an equivalent, target-independent intermediate code representation.