Эффективность параллельной реализации алгоритма Radix-4 быстрого преобразования Фурье
Parallel program for the fast Fourier transform is implemented on the basis of MPI (Message Passing Interface) technology. Radix-4 algorithm is chosen as a basic method to use. The dependence of parallel calculation acceleration on the number of processors is studied for two supercomputers. The formula describing the dependence of the calculation time on the number of processors is proposed for the range of the input data volume and supercomputer characteristics. The number of nodes providing you with maximum acceleration of calculation is estimated.