Using Search Space Reduction to Speedup Automatic Compiler Optimization Option Selection

• Main Authors: Chang,TingKoung, 張廷光
• Other Authors: Lin,naiwei
• Format: Others
• Language: zh-TW
• Published: 2011
• Online Access: http://ndltd.ncl.edu.tw/handle/84379901395649258207

碩士 === 國立中正大學 === 資訊工程研究所 === 99 === Compilers usually provide a large number of optimization options for users to optimize their programs. But most users are not familiar with compiler optimizations so that they don’t have the capability to select suitable optimization options. Genetic algorithm is a good solution to select the best optimization options set, but it’s time consuming. This study uses UROM to eliminate some options that don’t have effect for target program. Average has been a 46.62 percent decrease in search space. So we can reduce the evolution time by reduces population size. We also divide optimization options into four groups, and then greedy find the optimal solution in each group by multi-stage search. This mechanism can reduce 71.53% of search time.