Run-Time Support for the Automatic Parallelization of Java Programs

摘要: We describe and evaluate a novel approach for the automatic parallelization of programs that use pointer-based dynamic data structures, written in Java. The exploits parallelism among methods by creating an asynchronous thread execution each method invocation program. At compile time, are analyzed to determine they access, parameterized their context. A description these accesses is transmitted run-time system during program execution. utilizes this when may execute, enforce dependences threads. This main focus paper. More specifically, paper details representation framework used detect Experimental evaluation implementation on four-processor Sun multiprocessor indicates close ideal speedup can be obtained number benchmarks. validates our approach.