作者: Barry Steven Fagin , None
DOI:
关键词: Concurrency 、 Execution model 、 Backtracking 、 Parallel computing 、 Abstract machine 、 Instruction set 、 Benchmark (computing) 、 Computer science 、 Programming language 、 Prolog 、 Multiprocessing
摘要: As single computer systems approach the technological limits of their performance, scientists are turning to multiprocessing as a means solving computationally difficult problems. In addition, symbolic computing has emerged promising field scientific endeavor. These two fields have recently combined into new area research: parallel computing. One candidate language for is Prolog. Numerous ways executing Prolog in been proposed, but current efforts suffer from several deficiencies. Many cannot support fundamental types concurrency Other models purely theoretical interest, ignoring implementation costs. Detailed simulation studies execution scarce; at present little known about costs and benefits parallel. In this thesis, model presented: PPP or Parallel Processor. The supports AND-parallelism, OR-parallelism, intelligent backtracking. An described, through extension an existing abstract machine architecture. Several examples presented, compilation instruction set discussed. performance effects reported, based on large benchmark set. implications these results discussed, directions future work indicated.