作者: Miguel Areias , Ricardo Rocha
关键词: Overlay 、 Memory management 、 Memory map 、 Data structure 、 Operating system 、 Execution time 、 Multithreading 、 Parallel computing 、 Thread (computing) 、 Allocator 、 Interleaved memory 、 Extended memory 、 Computer science
摘要: Despite the availability of both multithreading and tabling in some Prolog systems, implementation these two features, such that they work together, implies complex ties to one another underlying engine. In recent work, we have proposed an approach combine with tabling, implemented on top Yap system, whose primary goal was reduce memory usage for table space. Regarding execution times, observed problems related Yap's allocator, which is based operating system's default when running programs allocate a higher number data structures this paper, propose more efficient scalable allocator multithreaded tabled evaluation logic programs. Our minimize performance degradation system suffers it exposed simultaneous requests made by multiple threads. For that, local global pages, split among specific different threads, together strategy where same type are pre-allocated within page. Experimental results show our new can effectively time scale better, increasing than original allocator.