Search

Optimizations in the and-parallel component of the and-or/parallel copying-based execution (ACE) model help to reduce memory consumption and improve speedups. These optimizations also enhance the performance of the system during forward execution and backtracking. The ACE system efficiently utilizes the independent and- and or-parallelism from Prolog programs. The proposed optimizations for ACE systems can also be applied to other and-parallel systems.

Data-parallelism and and-parallelism techniques in logic programs are fundamentally similar, differing only in memory management, fast track startup and management. The reconciliation of both the models results in powerful systems possessing the generality of the and-parallel systems and the performance advantages of the data-parallelism systems. The results of the actual implementation of the reconciliation of the two systems are discussed.