The identification of redundant computation is in general undecidable and prevents the obtainment of an ideal case as reference to the measurement of the remaining unexploited potential for redundancy removal and to the evaluation of code optimization effectiveness. This paper presents a methodology for optimization effectiveness analysis by observing the complete dynamic stream of executed instructions and memory references in the whole program execution, and by applying an extended value numbering algorithm to the execution trace. This method reduces the interprocedural analysis to the analysis of a large basic block and detects redundant memory and arithmetic operations that are visible only at the run-time. This way, the work extends the load-reuse analysis and provides both a more accurate approximation of the upper bound of exploitable optimization in the program and a reference point to evaluate optimization effectiveness. The results of applying this method to representative benchmark (SPECInt 2006) executables created with each compiler optimization level in GNU C/C++ Compiler are reported. The programs are run with a full-system simulator based on Power ISA 64-bit (version 2.06), and the whole application execution trace is collected. The proposed analysis reveals a significant amount of remaining unexploited redundancies even with the highest optimization level available. Sources of inefficiency and implications on exploring dynamic optimizations are discussed.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.