GPU Matrix Sort (An Efficient Implementation of Merge Sort)

Abstract

Sorting is one of the frequent used operations in computer science. Due to highly parallel computing nature of GPU architecture; it can be utilized for sorting purpose. We have considered the input array that is to be sorted in a 2D matrix form and applied a modified version of merge sort on that matrix. This modification leads to a much efficient sorting algorithm with reduced complexity. Therefore a lot of work has already been done to improve the efficiency of sorting algorithms. In this paper We have used the GPU architecture for solving the sorting problem.