The use of different strategies of search space reduction in mitigation of optimization selection problem

Compiler optimizations are transformations that are applied on the source code to improve its performance. Many times its a complex task choose which optimizations set must be used, so, usually are chosen optimization levels given by the compiler. However, this optimization levels not always are good enough to all programs. Thus, is needed to search for sets to improve specific programs. Currently the Best10 algorithm one of the best algorithms to mitigate the optimizations selection problem. This algorithm require one reduced search space to infer which are the optimization sets that must be applied during the compilation of programs. This work present the impact of the use of different search space creation strategies used by the Best10 algorithm. The results shows that sophisticated strategies do not always provide the best results.