A Logic Programming Approach to Predict Effective Compiler Settings for Embedded Software

摘要: This paper introduces a new logic-based method for optimising the selection of compiler ags on embedded architectures. In particular, we use Inductive Logic Programming (ILP) to learn logical rules that relate eective specic program features. Unlike earlier work, aim infer human-readable and seek develop relational rst-order approach which automatically discovers relevant features rather than relying vector predetermined attributes. To this end generated data set by measuring execution times 60 benchmarks an system development board developed ILP prototype outperforms current state-of-the-art learning in 34 benchmarks. Finally, combined strengths state art our hybrid reduced average 8% up 50% some cases.