Case-Based Reasoning have been used for a long times to solve several problems. The first Case-Based Reasoning used to find good compiler optimization sets, for an unseen program, proposed several strategies to tune the system. However, this work did not indicate the best parametrization. In addition, it evaluated the proposed approach using only kernels. Our paper revisit this work, in order to present an detail analysis of an Case-Based Reasoning system, applied in the context of compilers. In adition, we propose new strategies to tune the system. Experiments indicate that Case-Based Reasoning is a good choice to find compiler optimization sets that outperform a well-engineered compiler optimization level. Our Case-Based Reasoning approach achieves an average performance of 4.84% and 7.59% for cBench and SPEC CPU2006, respectively. In addition, experiments also indicate that Case-Based Reasoning outperforms the approach proposed by Purini and Jain, namely Best10.

No abstract

The Optimization Selection Problem is widely known in computer science for its complexity and importance. Several approaches based on machine learning and iterative compilation have been proposed to mitigate this problem. Although these approaches provide several advantages, they have disadvantages that can hinder the performance. This paper proposes a hybrid approach that combines the best of machine learning and iterative compilation. Several experiments were performed using different strategies, metrics and hardware platforms. A thorough analysis of the results reveals that the hybrid approach is a considerable improvement over machine learning and iterative compilation. In addition, the hybrid approach outperforms the best compiler optimization level of LLVM.

No abstract

No abstract