A Systematic Approach to Model-Guided Empirical Search for Memory

Abstract. The goal of this work is a systematic approach to compiler optimization for simultaneously optimizing across multiple levels of the memory hierarchy. Our approach combines compiler models and heuristics with guided empirical search to take advantage of their complementary strengths. The models and heuristics limit the search to a small number of candidate implementations, and the empirical results provide accurate feedback information to the compiler. In previous work, we propose a compiler algorithm for deriving a set of parameterized solutions, followed by a model-guided empirical search to determine the best integer parameter values and select the best overall solution. This paper focuses on formalizing the process of deriving parameter values, which is a multi-variable optimization problem, and considers the role of AI search techniques in deriving a systematic framework for the search.