Using inter-procedural side-effect information in JIT optimizations

• Main Author: Le, Anatole
• Format: Others
• Language: en
• Published: McGill University 2005
• Subjects: Computer Science.
• Online Access: http://digitool.Library.McGill.CA:80/R/?func=dbin-jump-full&object_id=82273

Side-effect analysis gives information about the set of locations that a statement may read or modify. This analysis can provide information useful in a compiler for performing aggressive optimizations. The impact of the use of side-effect analysis in compiler optimizations has been studied for programming languages such as Modula-3 and C, but no thorough investigation for Java has been done. We present a study of whether side-effect information improves performance in Java just-in-time (JIT) compilers, and if so, what level of analysis precision is needed. We also analyze the optimizations and benchmarks that benefit most from side-effect analysis. === We used SPARK, the inter-procedural analysis component of the SOOT Java analysis and optimization framework, to compute side-effect information and encode it in class files. We modified Jikes RVM, a research JIT, to make use of side-effect analysis in various local and global analyses and optimizations such as local common sub-expression elimination, heap SSA, redundant load elimination and loop-invariant code motion. On the SpecJVM98 benchmarks, we measured the static number of memory operations removed, the dynamic counts of memory reads eliminated, and the execution time. === Our results show that the use of side-effect analysis increases the number of static opportunities for load elimination by up to 98%, and reduces dynamic field read instructions by up to 27%. Side-effect information enabled speedups of up to 20% for some benchmarks. The main cause of the speedups is the use of side-effect information in load elimination. Finally, among the different levels of precision of side-effect information, a simple side-effect analysis is usually sufficient to obtain most of these speedups.