The performance of modern processors depends on their ability to execute multiple instructions per cycle, requiring increasing levels of instruction-level parallelism (ILP) to be exposed in programs. One of the major challenges to increasing available ILP is overcoming the limitations imposed by branch instructions. One paradigm presented to help overcome problems with branches both at run time and at compile time is predication. Predication allows the elimination of branch instructions by providing an alternate means of controlling the condition of execution for individual instructions. Effective use of hardware support for predication requires advanced compiler support.; Many existing techniques for compilation using predication make approximations or use heuristics, the inaccuracy of which reduces achievable performance. In some cases, these techniques realize a performance loss compared to codes without predication. This dissertation proposes and investigates a systematic approach to compiling for predication which consistently generates efficient predicated code.; The four key compiler technologies presented here work synergistically to realize the potential of predication. The Partial Reverse If-Conversion Framework provides the first compilation framework to accurately balance control and predication, while providing other compiler components with complete access to the predicated code for further optimization. Though the full potential of the Partial Reverse If-Conversion Framework remains unexplored, current compiler technology justifies its worth.; To operate on predicated code, the optimizer, scheduler, and register allocator require accurate information regarding the relationships among predicates. The Predicate Analysis System is the first efficient predicate relationship database to provide an approximation-free representation.; Optimization, scheduling, and register allocation also require accurate knowledge of the flow of information in the predicated code. Using the Predicate Analysis System, the Predicate Dataflow Graph is built to provide dataflow information. The Predicate Dataflow Graph was developed to provide accurate dataflow information in the presence of predication without requiring a change in the dataflow analysis equations.; In the context of the Partial Reverse If-Conversion Framework, some otherwise unnecessarily complex optimizations become feasible. One example is the Program Decision Logic Optimizer, which uses predicated code and the Predicate Analysis System to represent program decision logic as Boolean expressions. These expressions are minimized, factored, and reformulated to create programs with optimized decision components. Within the given compilation framework, the PDLO has been demonstrated to break many performance bottlenecks due to program decision.; The Partial Reverse If-Conversion Framework, the Predicate Analysis System, the Predicate Dataflow Graph, and the Program Decision Logic Optimizer are demonstrated to work synergistically in the compiler to generate efficient code for nonnumeric programs. Using predication, these techniques effectively optimize codes in the presence of complex branch control flow.
展开▼
