(5) An Optimization Technique of Efficient Goal Scheduling for KL1 Programs

This study aims to optimize the KL1 goal scheduling according to the program characteristics, and thus greatly reduce the overhead associated with parallelism control, which is the biggest performance problem for the KL1 run-time system. For that purpose, we will develop an optimizing compiler to obtain the optimal sequential execution sequence through data dependency analysis between goals, and a new run-time system that executes the resultant sequential program pieces efficiently by making full use of sequentializing.