University of Houston

The Fast Fourier Transform (FFT) is one of the most widely used algorithms in science and engineering. However, an efficient FFT implementations for computing on parallel machines are difficult to build because of the growing heterogeneity of hardware architectures, complexity of memory systems and diversity of communication systems of parallel machines. This thesis describes how to use the six-step algorithm to implement an MPI-UHFFT library which runs on distributed-memory architectures. In this library, the implementation of the FFT on parallel machines is selected automatically by attempting to minimize the execution time on those machines at run-time. MPI-UHFFT library is adaptive, portable, efficient, and easy-to-use.