high performance computing on graphics processing units: hgpu.org

This thesis reviews the parallel languages according to their computational complexities, in terms of time, while using sorting algorithms coded in CUDA and OpenMP. The thesis evaluates the solution for parallelism at a maintainable cost of money and other efforts, for achieving acceptable results of timing when compared to parallel languages together, as well as the sequential execution. The detailed timetables comparing timings from CUDA, OpenMP and sequential code, are present in related chapters. Eventually, the findings are examined to find the benefits of the parallel languages. The findings from the Chapter "Test and Comparisons" indicate that refactoring an algorithm to OpenMP language is easier when compared to CUDA language. However, results from comparing time values indicate that it is always the design of the algorithms, which makes it possible to achieve the performance that is expected of using the parallel languages.