The characterization of data-accumulating algorithms

A Case Study in Real-Time Parallel Computation: Correcting Algorithms

“…conforming to a result given in [10] and improved in [7], where S$(1, P) is the speedup offered in the static case by an algorithm that uses P processors, and c dp is a positive constant. The parallel complexity is in this case c dp N v ÂS$(1, P).…”

“…Definition 2.1 [7]. An algorithm A is a d-algorithm if: (I) A works on a set of data which is not entirely available at the beginning of computation.…”

“…Indeed, the definition from [10] is similar to Definition 2.2, except that``performs the same computation'' is replaced by``solves the same problem,'' and, in addition, A s is the best-known algorithm that solves the given problem. However, as noted in [7], such a definition is of little relevance, since there are d-algorithms whose performance is inherently worse than in the static case as defined in [10]. Therefore we chose the above definition, as suggested in [7].…”

“…However, as noted in [7], such a definition is of little relevance, since there are d-algorithms whose performance is inherently worse than in the static case as defined in [10]. Therefore we chose the above definition, as suggested in [7]. In addition, this definition allows us to define the time complexity of a d-algorithm in a more elegant manner, as we shall see below.…”

“…In this paradigm, data that arrive during the computation are considered as an endless stream, the algorithm terminating when all the currently arrived data have been processed. The study of d-algorithms was continued in [6,7,10], where it was shown that such an algorithm manifests superunitary behavior if and only if the parallel speedup for the static case is strictly larger than 1.…”