Solución al problema de ordenación usando PipeLine y Binary Tree como composiciones paralelas de alto nivel

Abstract

This paper proposes the model of the High Level Parallel Compositions or CPANs to communication patterns / interaction Pipeline and TreeDV for implementing a sorting algorithm by Structured Parallel Programming approach based on the concept of Parallel Objects. The CPANS TreeDV and PipeLine are displayed using the paradigm of object orientation and sorting problem is solved using two different algorithms; it is using a pipeline process to sort a dataset in disordered (CPAN Pipe) and one that by quick sort uses a binary tree for the ordering of the same dataset disordered by divide and conquer technique (CPAN TreeDVQS). Each proposal of CPAN contains a predefined set of restrictions of synchronization between processes (maximum parallelism, mutual exclusion and synchronization of producer- consumer type), and the use of synchronous, asynchronous and asynchronous future communication modes. Sorting algorithms, their design and implementation as CPANs and comparative performance metrics on a parallel machine 64 processors are shown.