Comparative study of the impact of processor architecture on compiler tuning

Abstract

Deciding the nearly optimal optimization options and selecting the right values for compiler parameter set is a combinatorial problem. In order to obtain the maximal performance, sophisticated tuning strategies were employed by many researchers. Impact of tuning and thereby quality of the Compiler generated code often depend on many factors like the optimization infrastructure of the compiler and its maturity level, optimization objective, application source and also the target processor architecture. In order to understand the impact of processor architecture, we have conducted an empirical study on X86 and ARM platforms and compared the results. We have employed Genetic Algorithm based tuning techniques on SPEC benchmark programs for both the compiler optimization option selection and also the parameter tuning problems independently as well as together. Results demonstrate that there is a significant impact of processor architecture on compiler tuning.