Automating runtime optimizations for parallel object-oriented programming

Software development for parallel computers has been recognized as one of the bottlenecks preventing their widespread use. In this thesis we examine two complementary approaches for addressing the challenges of high performance and enhanced programmability in parallel programs: automated optimizations and object-orientation. We have developed the parallel object-oriented language Charm++ (an extension of C++), which enables the benefits of object-orientation to be applied to the problems of parallel programming. In order to improve parallel program performance without extra effort, we explore the use of automated optimizations. In particular, we have developed techniques for automating run-time optimizations for parallel object-oriented languages. These techniques have been embodied in the Paradise post-mortem analysis tool which automates several run-time optimizations without programmer intervention. Paradise builds a program representation from traces, analyzes characteristics, chooses and parameterizes optimizations, and generates hints to the Charm++ run-time libraries. The optimizations researched are for static and dynamic object placement, scheduling, granularity control and communication reduction. We also evaluate Charm++, Paradise and several run-time optimization techniques using real applications, including an N-body simulation program, a program from the NAS benchmark suite, and several other programs.U of I OnlyETDs are only available to UIUC Users without author permissio