high performance computing on graphics processing units: hgpu.org

SIMD Parallelization is one of the most useful ways of decreasing the computation time and increases the performance of computation intensive algorithms. To do such process, we could execute some processes on several machines by using different platforms like MPI, OpenMP and distribute the workload by using message passing and shared memory. One of the most popular and high performance methods is using an array of graphical processors (GPU) which is used in this paper to present a new technique to save data and do computation by overclocking sieve algorithm make use of CUDA coding. This method shows a good performance upgrade in computation time and memory usage on generating prime numbers in compare with CPU handling.