PFFT is a general frame work for computing Fast Fourier Transforms on parallel machines. This frame work includes a ready to use library and several extra routines. The library can compute parallel multidimensional real-to-complex and complex-to-complex FFT's. It can also do the housekeeping for parallel application codes. PFFT uses either the vendor supplied scalar FFT routines, or FFTW to do the computations in processors. MPI is used for the interprocessor communication and I/O.
Download pfft and parallel transposes
For help contact dubey@tagore.uchicago.edu