| OCR Text |
Show Although the most powerful optimization eliminated function calls and two innermost loops, the drop in execution time was only 25%. This suggests that a dynamic program optimization for imperative programs using only optimization techniques employed by compilers has a limited effect. Other approaches, an example was memoing, must be combined with these techniques to achieve significant reductions in the execution time. Dynamic program optimization might be effective in languages that perform complex runtime operations. An important class of such languages are object oriented languages. Most existing attempts mix static program analysis and efficient runtime support to increase the execution speed of object oriented languages. Dynamic optimization could use information from the execution stream obtained dynamically during execution. Since most of this information is unavailable to static program analysis or too expensive to obtain by runtime support, dynamic optimization would significantly extend the reach of optimization methods. 62 |
