cannam@167: cannam@167: cannam@167: cannam@167: cannam@167: cannam@167: FFTW 3.3.8: FFTW MPI Transposes cannam@167: cannam@167: cannam@167: cannam@167: cannam@167: cannam@167: cannam@167: cannam@167: cannam@167: cannam@167: cannam@167: cannam@167: cannam@167: cannam@167: cannam@167: cannam@167: cannam@167: cannam@167: cannam@167: cannam@167: cannam@167:
cannam@167:

cannam@167: Next: , Previous: , Up: Distributed-memory FFTW with MPI   [Contents][Index]

cannam@167:
cannam@167:
cannam@167: cannam@167:

6.7 FFTW MPI Transposes

cannam@167: cannam@167: cannam@167:

The FFTW’s MPI Fourier transforms rely on one or more global cannam@167: transposition step for their communications. For example, the cannam@167: multidimensional transforms work by transforming along some cannam@167: dimensions, then transposing to make the first dimension local and cannam@167: transforming that, then transposing back. Because global cannam@167: transposition of a block-distributed matrix has many other potential cannam@167: uses besides FFTs, FFTW’s transpose routines can be called directly, cannam@167: as documented in this section. cannam@167:

cannam@167: cannam@167: cannam@167: cannam@167: cannam@167:
• Basic distributed-transpose interface:   cannam@167:
• Advanced distributed-transpose interface:   cannam@167:
• An improved replacement for MPI_Alltoall:   cannam@167:
cannam@167: cannam@167: cannam@167: cannam@167: cannam@167: cannam@167: