Multi-threading is currently supported by several well-known Prolog systems providing a highly portable solution for applications that can benefit from concurrency. When multi-threading is combined with tabling, we can exploit the power of higher procedural control and declarative semantics. However, despite the availability of both threads and tabling in some Prolog systems, the implementation of these two features implies complex ties to each other and to the underlying engine. Until now, XSB was the only Prolog system combining multi-threading with tabling. In XSB, tables may be either private or shared between threads. While thread-private tables are easier to implement, shared tables have all the associated issues of locking, synchronization and potential deadlocks. In this paper, we propose an alternative view to XSB's approach. In our proposal, each thread views its tables as private but, at the engine level, we use a common table space where tables are shared among all threads. We present three designs for our common table space approach: No-Sharing (NS) (similar to XSB's private tables), Subgoal-Sharing (SS) and Full-Sharing (FS). The primary goal of this work was to reduce the memory usage for the table space but, our experimental results, using the YapTab tabling system with a local evaluation strategy, show that we can also achieve significant reductions on running time.

A critical component in the implementation of a concurrent tabling system is the design of the table space. One of the most successful proposals for representing tables is based on a two-level trie data structure, where one trie level stores the tabled subgoal calls and the other stores the computed answers. In this work, we present a simple and efficient lock-free design where both levels of the tries can be shared among threads in a concurrent environment. To implement lock-freedom we took advantage of the CAS atomic instruction that nowadays can be widely found on many common architectures. CAS reduces the granularity of the synchronization when threads access concurrent areas, but still suffers from low-level problems such as false sharing or cache memory side-effects. In order to be as effective as possible in the concurrent search and insert operations over the table space data structures, we based our design on a hash trie data structure in such a way that it minimizes potential low-level synchronization problems by dispersing as much as possible the concurrent areas. Experimental results in the Yap Prolog system show that our new lock-free hash trie design can effectively reduce the execution time and scale better than previous designs.

No abstract

Tabling is a powerful implementation technique that improves the declarativeness and expressiveness of traditional Prolog systems in dealing with recursion and redundant computations. It can be viewed as a natural tool to implement dynamic programming problems, where a general recursive strategy divides a problem in simple sub-problems that are often the same. When tabling is combined with multithreading, we have the best of both worlds, since we can exploit the combination of higher declarative semantics with higher procedural control. However, at the engine level, such combination for dynamic programming problems is very difficult to exploit in order to achieve execution scalability as we increase the number of running threads. In this work, we focus on two well-known dynamic programming problems, the Knapsack and the Longest Common Subsequence problems, and we discuss how we were able to scale their execution by using the multithreaded tabling engine of the Yap Prolog system. To the best of our knowledge, this is the first work showing a Prolog system to be able to scale the execution of multithreaded dynamic programming problems. Our experiments also show that our system can achieve comparable or even better speedup results than other parallel implementations of the same problems.