Systematic search within an optimisation space based on Unified Transformation Framework
Shun Long, Grigori Fursin · International Journal of Computational Science and Engineering · 2009
Modern compilers have limited ability to exploit the performance improvement potential of complex transformation compositions due to their ad-hoc nature. Unified Transformation Framework (Kelly and Pugh, 1993a) provides a unified representation of different iteration reordering transformations. It results in a large and complex space to explore. This paper presents a heuristic search algorithm capable of efficiently locating good optimisations within this space. Preliminary experimental results on Java show that it can achieve an average speedup of 1.14 and 1.10 on two different platforms, and more than 75% of the maximum performance available can be obtained within 20 evaluations.