This paper reports the first-phase results of an investigation on the effect of caching on heuristic search. The heuristic focuses on solving the combinatorial optimization problem of selecting an optimal subset of features for processor workload characterization. Caching is implemented with two global optimum heuristic search algorithms: genetic algorithm, and probabilistic reasoning. The first phase of the research includes examination of the locality of reference that is inherent in these heuristic procedures. Strong spatial as well as strong temporal locality of reference are identified. The finding of the first phase will be used to device a set of experiments assessing the effect of caching on the runtime and solution quality of the heuristics.
展开▼