2008/01/01 by Gergö Barany, Barany, Gergö
Computer Science · #Distributed systems and fault tolerance #Parallel Computing and Optimization Techniques #Partial redundancy elimination #Software Engineering Research #source-to-source optimization #value flow analysis
paper · doi:10.4230/dagsemproc.08161.9
openalex publication_date 2008/01/01 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
Partial redundancy elimination is a common program optimization that attempts to improve execution time by removing superfluous computations from a program. There are two well-known classes of such techniques: syntactic and semantic methods. While semantic optimization is more powerful, traditional algorithms based on SSA from are complicated, heuristic in nature, and unable to perform certain useful optimizations. The value flow graph is a syntactic program representation modeling semantic equivalences; it allows the combination of simple syntactic partial redundancy elimination with a powerful semantic analysis. This yields an optimization that is computationally optimal and simpler than traditional semantic methods. This talk discusses partial redundancy elimination using the value flow graph. A source-to-source optimizer for C++ was implemented using the SATIrE program analysis and transformation system. Two tools integrated in SATIrE were used in the implementation: ROSE is a framework for arbitrary analyses and source-to-source transformations of C++ programs, PAG is a tool for generating data flow analyzers from functional specifications.