Using types to analyze and optimize object-oriented programs

Abstract

Object-oriented programming languages provide many software engineering benefits, but these often come at a performance cost. Object-oriented programs make extensive use of method invocations and pointer dereferences, both of which are potentially costly on modern machines. We show how to use types to produce effective, yet simple, techniques that reduce the costs of these features in Modula-3, a statically typed, object-oriented language. Our compiler performs type-based alias analysis to disambiguate memory references. It uses the results of the type-based alias analysis to eliminate redundant memory references and to replace monomorphic method invocation sites with direct calls. Using limit, static, and running time evaluation, we demonstrate that these techniques are effective, and sometimes perfect for a set of Modula-3 benchmarks.

Fast online pointer analysis

“…This is similar to limit studies that other researchers have used to evaluate and debug various compiler analyses [Larus and Chandra 1993;Mock et al 2001;Diwan et al 2001;. Kotzmann and Mössenböck use their escape analysis for stack allocation [2005].…”

“…We evaluate their effect on the precision of our pointer analysis with respect to a client of the pointer analysis: the partitioner in CBGC (connectivity-based garbage collection ). Since these optimizations change precision in incomparable ways (they use a different heap model), it is not appropriate to compare their relative effects using pointsTo sets or alias sets [Diwan et al 2001]. The CBGC partitioner works by placing each allocation site in its own partition and then collapsing strongly-connected components in the partition graph.…”

“…The body of literature on pointer analyses is vast [Hind 2001]. At one extreme, exemplified by Steensgaard [1996] and analyses with type-based heap abstraction [Harris 1999;Tip and Palsberg 2000;Diwan et al 2001], the analyses are fast, but imprecise. At the other extreme, exemplified by shape analyses [Hendren 1990;Sagiv et al 1999], the analyses are slow, but precise enough to discover the shapes of many data structures.…”

A probabilistic pointer analysis for speculative optimizations

“…Several approaches balance this trade-off well by providing an appropriate combination of precision and scalability [2,8,30,32]. However, studies suggest that a more accurate pointer analysis does not necessarily provide increased optimization opportunities [10,17] because ambiguous pointers will persist. Hence speculative optimizations have recently been proposed to bridge this persistent disparity between safety and optimization.…”

Using Inter-Procedural Side-Effect Information in JIT Optimizations

“…However, they only measured the statie number of optimizations performed (rather than their mn-time effect), and their benchmarks are mostly pointer-free C programs, sorne translated directly from FORTRAN, so they found, unsurprisingly, that alias analysis precision had little effect. Other work studying the effect of alias analysis on scalar optimizations also suggests that a simple alias analysis may be sufficient [DLFR01,DMM01].…”