Prediction Models for Performance, Power, and Energy Efficiency of Software Executed on Heterogeneous Hardware

Heterogeneous environments are becoming commonplace so it is increasingly important to understand how and where we could execute a given algorithm the most efficiently. In this paper we propose a methodology that uses both static source code metrics and dynamic execution time, power and energy measurements to build configuration prediction models. These models are trained on special benchmarks that have both sequential and parallel implementations and can be executed on various computing elements, e.g., on CPUs or GPUs. After they are built, however, they can be applied to a new system using only the system's static metrics which are much more easily computable than any dynamic measurement. We found that we could predict the optimal execution configuration fairly accurately using static information alone.