Abstract
Tabled evaluation is a recognized and powerful technique that overcomes some limitations of traditional Prolog systems in dealing with recursion and redundant subcomputations. We can distinguish two main categories of tabling mechanisms: suspension-based tabling and linear tabling. While suspension-based mechanisms are considered to obtain better results in general, they have more memory space requirements and are more complex and harder to implement than linear tabling mechanisms. Arguably, the SLDT and Dynamic Reordering of Alternatives (DRA) strategies are the two most successful extensions to standard linear tabled evaluation. In this work, we propose a new strategy, named dynamic reordering of solutions, and we present a framework, on top of the Yap system, that supports the combination of all these three strategies. Our implementation shares the underlying execution environment and most of the data structures used to implement tabling in Yap. We thus argue that all these common features allows us to make a first and fair comparison between these different linear tabling strategies and, therefore, better understand the advantages and weaknesses of each, when used solely or combined with the others.

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

Abstract
Tabling is a technique of resolution that overcomes some limitations of traditional Prolog systems in dealing with recursion and redundant sub-computations. We can distinguish two main categories of tabling mechanisms: suspension-based tabling and linear tabling. In suspension-based tabling, a tabled evaluation can be seen as a sequence of sub-computations that suspend and later resume. Linear tabling mechanisms maintain a single execution tree where tabled subgoals always extend the current computation without requiring suspension and resumption of sub-computations. In this work, we present a new and efficient implementation of linear tabling, but for that we have extended an already existent suspension-based implementation, the YapTab engine. Our design is based on dynamic reordering of alternatives but it innovates by considering a strategy that schedules the re-evaluation of tabled calls in a similar manner to the suspension-based strategies of YapTab. Our implementation also shares the underlying execution environment and most of the data structures used to implement tabling in YapTab. We thus argue that all these common features allows us to make a first and fair comparison between suspension-based and linear tabling and, therefore, better understand the advantages and weaknesses of each.

Abstract
The execution model on which most tabling engines are based allocates a choice point whenever a new tabled subgoal is called. This happens even when the call is deterministic. however, sortie of the information from the choice point; is never used when evaluating deterministic tabled calls with batched scheduling. Moreover, when a deterministic answer is found for a. deterministic tabled call, the call can be completed early and the corresponding choice point can be removed. Thus, if applying batched scheduling to a long deterministic computation the system may end up consuming memory and evaluating calls unnecessarily. In this paper, we propose a solution that, tries to reduce this memory and execution overhead to a minimum. Our experimental results show that;, for deterministic tabled calls and tabled answers with batched scheduling, it; is possible not. only to reduce the memory usage overhead, but also the running time of tire execution.

Abstract