Дослідження та реалізація паралельного алгоритму сортування злиттям за допомогою технології CUDA

Abstract

The development of a parallel merge sort algorithm using CUDA architecture and GPUs is considered. The issue of analyzing existing sorting methods to achieve high performance is considered, the choice of software module development tools is justified, class diagrams of the software module are developed, and the choice of software implementation environment is justified. In the work, a software implementation of a parallel algorithm using CUDA is created and its performance is tested on various data sets. Using the results will allow improving the speed and performance of programs and algorithms that require processing and ordering large amounts of data.