cannam@127: cannam@127: cannam@127: cannam@127: cannam@127: cannam@127: FFTW 3.3.5: Using MPI Plans cannam@127: cannam@127: cannam@127: cannam@127: cannam@127: cannam@127: cannam@127: cannam@127: cannam@127: cannam@127: cannam@127: cannam@127: cannam@127: cannam@127: cannam@127: cannam@127: cannam@127: cannam@127: cannam@127: cannam@127: cannam@127:
cannam@127:

cannam@127: Next: , Previous: , Up: FFTW MPI Reference   [Contents][Index]

cannam@127:
cannam@127:
cannam@127: cannam@127:

6.12.3 Using MPI Plans

cannam@127: cannam@127:

Once an MPI plan is created, you can execute and destroy it using cannam@127: fftw_execute, fftw_destroy_plan, and the other functions cannam@127: in the serial interface that operate on generic plans (see Using Plans). cannam@127:

cannam@127: cannam@127: cannam@127:

The fftw_execute and fftw_destroy_plan functions, applied to cannam@127: MPI plans, are collective calls: they must be called for all processes cannam@127: in the communicator that was used to create the plan. cannam@127:

cannam@127: cannam@127:

You must not use the serial new-array plan-execution functions cannam@127: fftw_execute_dft and so on (see New-array Execute Functions) with MPI plans. Such functions are specialized to the cannam@127: problem type, and there are specific new-array execute functions for MPI plans: cannam@127:

cannam@127: cannam@127: cannam@127: cannam@127: cannam@127:
cannam@127:
void fftw_mpi_execute_dft(fftw_plan p, fftw_complex *in, fftw_complex *out);
cannam@127: void fftw_mpi_execute_dft_r2c(fftw_plan p, double *in, fftw_complex *out);
cannam@127: void fftw_mpi_execute_dft_c2r(fftw_plan p, fftw_complex *in, double *out);
cannam@127: void fftw_mpi_execute_r2r(fftw_plan p, double *in, double *out);
cannam@127: 
cannam@127: cannam@127: cannam@127: cannam@127:

These functions have the same restrictions as those of the serial cannam@127: new-array execute functions. They are always safe to apply to cannam@127: the same in and out arrays that were used to cannam@127: create the plan. They can only be applied to new arrarys if those cannam@127: arrays have the same types, dimensions, in-placeness, and alignment as cannam@127: the original arrays, where the best way to ensure the same alignment cannam@127: is to use FFTW’s fftw_malloc and related allocation functions cannam@127: for all arrays (see Memory Allocation). Note that distributed cannam@127: transposes (see FFTW MPI Transposes) use cannam@127: fftw_mpi_execute_r2r, since they count as rank-zero r2r plans cannam@127: from FFTW’s perspective. cannam@127:

cannam@127:
cannam@127:
cannam@127:

cannam@127: Next: , Previous: , Up: FFTW MPI Reference   [Contents][Index]

cannam@127:
cannam@127: cannam@127: cannam@127: cannam@127: cannam@127: