Partial redundancy elimination with predication techniques

Partial redundancy elimination (PRE) techniques play an important role in optimizing compilers. Many optimizations, such as elimination of redundant expressions, communication optimizations, and load-reuse optimizations, employ PRE as an underlying technique for improving the efficiency of a program. Classical approaches are conservative and fail to exploit many opportunities for optimization. Therefore, new PRE approaches have been developed that greatly increase the number of eliminated redundancies. However, they either cause the code to explode in size or they cannot handle statements with side-effects. First, we describe a basic transformation for PRE that employs predication to achieve a complete removal of partial redundancies. The control flow is not restructured, however, predication might cause performance overhead. Second, a cost-analysis based on probabilistic data-flow analysis decides whether a PRE transformation is profitable and should be applied. In contrast to other approaches our transformation is strictly semantics preserving.