P. Lopez, M. Hermenegildo, and S. Debray. A methodology for granularitybased control of parallelism in logic programs. Journal of Symbolic Computation, 21(4/5/6):715-734, 1996.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....analysis, as well as in other analyses in which the size of a term is an important notion. For example in [8] information on the size of the arguments of a predicate is used to determine how many solutions it can produce and to estimate its computational cost. This work is generalised in [18] and applied to control the degree of parallelism in a logic program. Types have been used before to drive the construction of suitable norms. In [2] norms are de ned in terms of type schemata but not derived automatically. Automatic derivation of typed norms is considered in [11] in which ....

P. Lopez, M. Hermenegildo, and S. Debray. A methodology for granularitybased control of parallelism in logic programs. Journal of Symbolic Computation, 21(4/5/6):715-734, 1996.

....has been noticed since the beginning in logic programming (see [6] for an extensive survey) because the simplicity of the logical semantics makes it apparently easy to extract. However, limiting the amount of extracted parallelism in order to maximize resources allocation is a non trivial problem [18, 23]. Compilation towards a data parallel execution model has been studied in the Reform model [20] or in the Multilog system [27] In contrast, explicit parallel languages allow the programmer to describe precisely the parallel execution of his program by specifying independent computations, ....

P. Lopez-Garcia, M. Hermenegildo, and S. K. Debray. A methodology for granularitybased control of parallelism in logic programming. Journal of Logic and Computation, special issue on parallel symbolic computation, 22:715-734, 1996.

....excessive parallelism. The general idea is that if the gain obtained by executing a task in parallel is less than the overheads required to support parallel execution, then the task is better executed sequentially [6] Granularity control have been recently applied to Prolog Logic Programming [20, 9, 8] and Functional Programming [7] In these methods, a task is considered to be a goal or an alternative to be executed in parallel, and useful work the amount of computation performed to solve the goal or try the alternative. Granularity control in these systems thus boils down to a two phase ....

....as there is parallel execution, the direct overheads probably do not vary (on average) greatly from program to program, so the constant of proportionality would not change greatly between programs. Task size, as estimated 1 Or more precisely, an upper or lower bound on the goal s complexity. In [9, 8], a relationship between the inputs and the bound is derived for recursive goals, and this is used at run time to determine the time complexity. by using a goal s time complexity, is therefore a suitable metric for accounting this source of overheads across a wide range of programs. However, in ....

[Article contains additional citation context not shown here]

P. Lopez-Garca, M. V. Hermenegildo, and S. K. Debray. A Methodology for Granularity Based Control of Parallelism in Logic Programs. Journal of Symbolic Computing, Special Issue on Parallel Symbolic Computation, 11(3--4):217-- 242, 1996.

....time complexity analysis of logic programs, does not require any complicated difference equation solving machinery. 1 Introduction Automatic time complexity analysis is useful to the programmer for algorithmic considerations but has a special role in the development of efficient parallel programs [9, 6, 7, 12, 15]. The execution of a parallel program can break down into processes which are too finegrained for a multiprocessor. This can present a mismatch of granularity between the program and the multi processor which, in turn, can degrade performance. Time complexity analysis enables fine grained ....

....is fine grained whereas quicksort is less predictable generating both fine grained and course grained processes. The programs were hand annotated with the thresholding tests, and then timings where taken on a Sequent Symmetry for 1, 2, 4 and 9 processors. We have used similar benchmark programs to [12], and the same 20MHz 80386 processor Sequent. The andparallel Prolog system DASWAM [20] was used. The programs used were limited to independent and parallelism, because suspension complicates the granularity question for general dependent and parallelism. The programs were executed with different ....

[Article contains additional citation context not shown here]

P. Lopez Garca, M. Hermenegildo, and S.K. Debray. A Methodology for Granularity Based Control of Parallelism in Logic Programs. Journal of Symbolic Computing, 11(3--4):217--242, 1996.

....excessive parallelism. The general idea is that if the gain obtained by executing a task in parallel is less than the overheads required to support parallel execution, then the task is better executed sequentially [6] Granularity control have been recently applied to Prolog Logic Programming [18, 9, 8] and to Functional Programming [7] In logic programs, a task is considered to be a goal or an alternative to be executed in parallel, and useful work the amount of computation performed to solve the goal or try the alternative. Granularity control in most of these systems thus boils down to a ....

.... single worker versus running it without annotations on the DASWAM running on a Sequent Symmetry, it is 43 slower, the largest such overhead for any IAP program examined in [12] Moreover, the program is not suitable for the granularity control based on time complexity analysis as described in [9, 8], because of the difficulties of deriving a relationship between input arguments and the complexity. The distance metric, however, suggests a very simple way to increase the distance, by performing a source to source transformation on the program such that and parallelism is generated only for ....

[Article contains additional citation context not shown here]

P. Lopez-Garca, M. V. Hermenegildo, and S. K. Debray. A Methodology for Granularity Based Control of Parallelism in Logic Programs. Journal of Symbolic Computing, Special Issue on Parallel Symbolic Computation, 11(3--4):217--242, 1996.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC