cannam@167: cannam@167: cannam@167: cannam@167: cannam@167: cannam@167: FFTW 3.3.8: Using MPI Plans 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: FFTW MPI Reference   [Contents][Index]

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

6.12.3 Using MPI Plans

cannam@167: cannam@167:

Once an MPI plan is created, you can execute and destroy it using cannam@167: fftw_execute, fftw_destroy_plan, and the other functions cannam@167: in the serial interface that operate on generic plans (see Using Plans). cannam@167:

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

The fftw_execute and fftw_destroy_plan functions, applied to cannam@167: MPI plans, are collective calls: they must be called for all processes cannam@167: in the communicator that was used to create the plan. cannam@167:

cannam@167: cannam@167:

You must not use the serial new-array plan-execution functions cannam@167: fftw_execute_dft and so on (see New-array Execute Functions) with MPI plans. Such functions are specialized to the cannam@167: problem type, and there are specific new-array execute functions for MPI plans: cannam@167:

cannam@167: cannam@167: cannam@167: cannam@167: cannam@167:
cannam@167:
void fftw_mpi_execute_dft(fftw_plan p, fftw_complex *in, fftw_complex *out);
cannam@167: void fftw_mpi_execute_dft_r2c(fftw_plan p, double *in, fftw_complex *out);
cannam@167: void fftw_mpi_execute_dft_c2r(fftw_plan p, fftw_complex *in, double *out);
cannam@167: void fftw_mpi_execute_r2r(fftw_plan p, double *in, double *out);
cannam@167: 
cannam@167: cannam@167: cannam@167: cannam@167:

These functions have the same restrictions as those of the serial cannam@167: new-array execute functions. They are always safe to apply to cannam@167: the same in and out arrays that were used to cannam@167: create the plan. They can only be applied to new arrarys if those cannam@167: arrays have the same types, dimensions, in-placeness, and alignment as cannam@167: the original arrays, where the best way to ensure the same alignment cannam@167: is to use FFTW’s fftw_malloc and related allocation functions cannam@167: for all arrays (see Memory Allocation). Note that distributed cannam@167: transposes (see FFTW MPI Transposes) use cannam@167: fftw_mpi_execute_r2r, since they count as rank-zero r2r plans cannam@167: from FFTW’s perspective. cannam@167:

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

cannam@167: Next: , Previous: , Up: FFTW MPI Reference   [Contents][Index]

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