(redirected from parallelizations)
Also found in: Dictionary, Thesaurus.
See: compare
Mentioned in ?
References in periodicals archive ?
Furthermore, we are also interested in observing how effective global analysis is in reducing the overhead due to run-time independence checking when using conditional parallelization algorithms (such as MEL or CDG).
The main conclusion from the dynamic tests is that data flow analysis is quite effective in automatic parallelization.
The obvious drawback is that sometimes unconditional parallelization results in smaller speedups or even no speedup at all in some programs for which conditional parallelization achieves good parallel performance.
We have studied the effectiveness of global analysis in the parallelization of logic programs using strict independence.
It allows not only to significantly reduce the number of run-time independence checks and/or increase the amount of unconditionally parallel code, but it is also quite successful at detecting sequential code statically, thus saving any wasteful detection of sequentiality at run-time, for tile cases in which conditional parallelization is selected.
One such notion is "nonstrict independence" [Hermenegildo and Rossi 1995] for which we have recently developed automatic parallelization technology based on global analysis [Cabeza and Hermenegildo 1994].
1) See [Hermenegildo 1997] for a more extended discussion of the relationships between parallelization techniques used for logic programs and those used for other programming paradigms.
3) See also [Chang and Chiang 1989; DeGroot 1987; Gupta and Costa 1992] for other methods of handling side-effects in the context of logic program parallelization.
A Comparative Study of Methods for Automatic Compile-time Parallelization of Logic Programs.
Automatic Parallelization of Irregular and Pointer-Based Computations: Perspectives from Logic and Constraint Programming.
Automatic Compile-time Parallelization of Logic Programs for Restricted, Goal-level, Independent And-parallelism.
The CDG, UDG, and MEL Methods for Automatic Compile-time Parallelization of Logic Programs for Independent And-parallelism.