Enhancing the performance of android applications on multi-core processors by selecting parallel configurations for source codes

摘要

Paralleling source code is the most common approach to optimize the performance of Android applications on multi-core processors. However, the problem of selecting a good parallel structure and its parameter tube is still a challenge. This paper proposes the performance optimization method based on selecting the best parallel configuration including the parallel structure and corresponding parameter tube. Native C/C++ code of an Android application is analyzed to find parallel structures using OpenMP framework and parameter tubes of each parallel structure. The evaluation function is built to evaluate the performance of an Android application under parallel configuration. This function is implemented in an optimization program to select the best parallel configuration of the Android application.