Chris@10: /*
Chris@10:  * Copyright (c) 2003, 2007-11 Matteo Frigo
Chris@10:  * Copyright (c) 2003, 2007-11 Massachusetts Institute of Technology
Chris@10:  *
Chris@10:  * This program is free software; you can redistribute it and/or modify
Chris@10:  * it under the terms of the GNU General Public License as published by
Chris@10:  * the Free Software Foundation; either version 2 of the License, or
Chris@10:  * (at your option) any later version.
Chris@10:  *
Chris@10:  * This program is distributed in the hope that it will be useful,
Chris@10:  * but WITHOUT ANY WARRANTY; without even the implied warranty of
Chris@10:  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
Chris@10:  * GNU General Public License for more details.
Chris@10:  *
Chris@10:  * You should have received a copy of the GNU General Public License
Chris@10:  * along with this program; if not, write to the Free Software
Chris@10:  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA
Chris@10:  *
Chris@10:  */
Chris@10: 
Chris@10: #include "ifftw-mpi.h"
Chris@10: 
Chris@10: /* r2c and c2r transforms.  The sz dtensor, as usual, gives the size
Chris@10:    of the "logical" complex array.  For the last dimension N, however,
Chris@10:    only N/2+1 complex numbers are stored for the complex data.  Moreover,
Chris@10:    for the real data, the last dimension is *always* padded to a size
Chris@10:    2*(N/2+1).  (Contrast this with the serial API, where there is only
Chris@10:    padding for in-place plans.) */
Chris@10: 
Chris@10: /* problem.c: */
Chris@10: typedef struct {
Chris@10:      problem super;
Chris@10:      dtensor *sz;
Chris@10:      INT vn; /* vector length (vector stride 1) */
Chris@10:      R *I, *O; /* contiguous interleaved arrays */
Chris@10: 
Chris@10:      rdft_kind kind; /* assert(kind < DHT) */
Chris@10:      unsigned flags; /* TRANSPOSED_IN/OUT meaningful for rnk>1 only
Chris@10: 			SCRAMBLED_IN/OUT meaningful for 1d transforms only */
Chris@10: 
Chris@10:      MPI_Comm comm;
Chris@10: } problem_mpi_rdft2;
Chris@10: 
Chris@10: problem *XM(mkproblem_rdft2)(const dtensor *sz, INT vn,
Chris@10: 			     R *I, R *O, MPI_Comm comm,
Chris@10: 			     rdft_kind kind, unsigned flags);
Chris@10: problem *XM(mkproblem_rdft2_d)(dtensor *sz, INT vn,
Chris@10: 			       R *I, R *O, MPI_Comm comm,
Chris@10: 			       rdft_kind kind, unsigned flags);
Chris@10: 
Chris@10: /* solve.c: */
Chris@10: void XM(rdft2_solve)(const plan *ego_, const problem *p_);
Chris@10: 
Chris@10: /* plans have same operands as rdft plans, so just re-use */
Chris@10: typedef plan_rdft plan_mpi_rdft2;
Chris@10: #define MKPLAN_MPI_RDFT2(type, adt, apply) \
Chris@10:   (type *)X(mkplan_rdft)(sizeof(type), adt, apply)
Chris@10: 
Chris@10: int XM(rdft2_serial_applicable)(const problem_mpi_rdft2 *p);
Chris@10: 
Chris@10: /* various solvers */
Chris@10: void XM(rdft2_rank_geq2_register)(planner *p);
Chris@10: void XM(rdft2_rank_geq2_transposed_register)(planner *p);
Chris@10: void XM(rdft2_serial_register)(planner *p);