Automatic program segment similarity detection in targeted program performance improvement

Targeted optimization of program segments can provide an additional program speedup over the highest default op-timization level, such as-O3 in GCC. The key challenge is how to automatically search for performance sensitive pro-gram segments in a given code, to which a customized set of optimization compiler options could be applied. In this paper we propose a method for automatic detec-tion of performance sensitive program segments based on program segment similarity. First we create a proxy seg-ment template database trained over a set of random in-put programs. The compiler identifies program segments by correlating them to the pre-build proxy segment templates using the syntax structure and architecture-dependent be-havior similarity. We argue that the identified program seg-ments can be custom optimized to improve the overall pro-gram performance. The method is evaluated on the Intel XScale PXA255 platform using randomly selected benchmarks. The experi-mental results show that our method can provide additional speedups over the highest optimization level in GCC 3.3 (-O3) for an arbitrary set of applications.