Machine Learning-based Compiler Optimization

08/02/2012 - 6:00pm
FL-1001, Small Seminar Room
Arash Ashari


Selecting a good set of compiler optimization flags is a challenging task. Specifically, due to the fast change and rapid improvement in microprocessor technology, an automatic compiler optimization that exploits the aspects of new multi core processors is missing. Recently researchers have applied machine-learning algorithms on both static and dynamic profiling features of computer programs including hardware performance counters to achieve semi-optimal transformations on different benchmarks. cTuning is one of the existing tools, which is an open source framework that profiles static features of computer programs and uses machine learning algorithms to find a semi-optimal set of compiler optimization flags. In this Internship, we extend cTuning by adding new benchmarks for training and we test it on few weather forecasting benchmarks at NCAR. We perform our experiment on Janus high performance computer system and finally report our result based on iterative compilation, default optimization flag/s and machine learning flags.