furnace/extern/fftw/rdft/problem2.c

225 lines
6 KiB
C
Raw Normal View History

/*
* Copyright (c) 2003, 2007-14 Matteo Frigo
* Copyright (c) 2003, 2007-14 Massachusetts Institute of Technology
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*
*/
#include "dft/dft.h"
#include "rdft/rdft.h"
#include <stddef.h>
static void destroy(problem *ego_)
{
problem_rdft2 *ego = (problem_rdft2 *) ego_;
X(tensor_destroy2)(ego->vecsz, ego->sz);
X(ifree)(ego_);
}
static void hash(const problem *p_, md5 *m)
{
const problem_rdft2 *p = (const problem_rdft2 *) p_;
X(md5puts)(m, "rdft2");
X(md5int)(m, p->r0 == p->cr);
X(md5INT)(m, p->r1 - p->r0);
X(md5INT)(m, p->ci - p->cr);
X(md5int)(m, X(ialignment_of)(p->r0));
X(md5int)(m, X(ialignment_of)(p->r1));
X(md5int)(m, X(ialignment_of)(p->cr));
X(md5int)(m, X(ialignment_of)(p->ci));
X(md5int)(m, p->kind);
X(tensor_md5)(m, p->sz);
X(tensor_md5)(m, p->vecsz);
}
static void print(const problem *ego_, printer *p)
{
const problem_rdft2 *ego = (const problem_rdft2 *) ego_;
p->print(p, "(rdft2 %d %d %T %T)",
(int)(ego->cr == ego->r0),
(int)(ego->kind),
ego->sz,
ego->vecsz);
}
static void recur(const iodim *dims, int rnk, R *I0, R *I1)
{
if (rnk == RNK_MINFTY)
return;
else if (rnk == 0)
I0[0] = K(0.0);
else if (rnk > 0) {
INT i, n = dims[0].n, is = dims[0].is;
if (rnk == 1) {
for (i = 0; i < n - 1; i += 2) {
*I0 = *I1 = K(0.0);
I0 += is; I1 += is;
}
if (i < n)
*I0 = K(0.0);
} else {
for (i = 0; i < n; ++i)
recur(dims + 1, rnk - 1, I0 + i * is, I1 + i * is);
}
}
}
static void vrecur(const iodim *vdims, int vrnk,
const iodim *dims, int rnk, R *I0, R *I1)
{
if (vrnk == RNK_MINFTY)
return;
else if (vrnk == 0)
recur(dims, rnk, I0, I1);
else if (vrnk > 0) {
INT i, n = vdims[0].n, is = vdims[0].is;
for (i = 0; i < n; ++i)
vrecur(vdims + 1, vrnk - 1,
dims, rnk, I0 + i * is, I1 + i * is);
}
}
INT X(rdft2_complex_n)(INT real_n, rdft_kind kind)
{
switch (kind) {
case R2HC:
case HC2R:
return (real_n / 2) + 1;
case R2HCII:
case HC2RIII:
return (real_n + 1) / 2;
default:
/* can't happen */
A(0);
return 0;
}
}
static void zero(const problem *ego_)
{
const problem_rdft2 *ego = (const problem_rdft2 *) ego_;
if (R2HC_KINDP(ego->kind)) {
/* FIXME: can we avoid the double recursion somehow? */
vrecur(ego->vecsz->dims, ego->vecsz->rnk,
ego->sz->dims, ego->sz->rnk,
UNTAINT(ego->r0), UNTAINT(ego->r1));
} else {
tensor *sz;
tensor *sz2 = X(tensor_copy)(ego->sz);
int rnk = sz2->rnk;
if (rnk > 0) /* ~half as many complex outputs */
sz2->dims[rnk-1].n =
X(rdft2_complex_n)(sz2->dims[rnk-1].n, ego->kind);
sz = X(tensor_append)(ego->vecsz, sz2);
X(tensor_destroy)(sz2);
X(dft_zerotens)(sz, UNTAINT(ego->cr), UNTAINT(ego->ci));
X(tensor_destroy)(sz);
}
}
static const problem_adt padt =
{
PROBLEM_RDFT2,
hash,
zero,
print,
destroy
};
problem *X(mkproblem_rdft2)(const tensor *sz, const tensor *vecsz,
R *r0, R *r1, R *cr, R *ci,
rdft_kind kind)
{
problem_rdft2 *ego;
A(kind == R2HC || kind == R2HCII || kind == HC2R || kind == HC2RIII);
A(X(tensor_kosherp)(sz));
A(X(tensor_kosherp)(vecsz));
A(FINITE_RNK(sz->rnk));
/* require in-place problems to use r0 == cr */
if (UNTAINT(r0) == UNTAINT(ci))
return X(mkproblem_unsolvable)();
/* FIXME: should check UNTAINT(r1) == UNTAINT(cr) but
only if odd elements exist, which requires compressing the
tensors first */
if (UNTAINT(r0) == UNTAINT(cr))
r0 = cr = JOIN_TAINT(r0, cr);
ego = (problem_rdft2 *)X(mkproblem)(sizeof(problem_rdft2), &padt);
if (sz->rnk > 1) { /* have to compress rnk-1 dims separately, ugh */
tensor *szc = X(tensor_copy_except)(sz, sz->rnk - 1);
tensor *szr = X(tensor_copy_sub)(sz, sz->rnk - 1, 1);
tensor *szcc = X(tensor_compress)(szc);
if (szcc->rnk > 0)
ego->sz = X(tensor_append)(szcc, szr);
else
ego->sz = X(tensor_compress)(szr);
X(tensor_destroy2)(szc, szr); X(tensor_destroy)(szcc);
} else {
ego->sz = X(tensor_compress)(sz);
}
ego->vecsz = X(tensor_compress_contiguous)(vecsz);
ego->r0 = r0;
ego->r1 = r1;
ego->cr = cr;
ego->ci = ci;
ego->kind = kind;
A(FINITE_RNK(ego->sz->rnk));
return &(ego->super);
}
/* Same as X(mkproblem_rdft2), but also destroy input tensors. */
problem *X(mkproblem_rdft2_d)(tensor *sz, tensor *vecsz,
R *r0, R *r1, R *cr, R *ci, rdft_kind kind)
{
problem *p = X(mkproblem_rdft2)(sz, vecsz, r0, r1, cr, ci, kind);
X(tensor_destroy2)(vecsz, sz);
return p;
}
/* Same as X(mkproblem_rdft2_d), but with only one R pointer.
Used by the API. */
problem *X(mkproblem_rdft2_d_3pointers)(tensor *sz, tensor *vecsz,
R *r0, R *cr, R *ci, rdft_kind kind)
{
problem *p;
int rnk = sz->rnk;
R *r1;
if (rnk == 0)
r1 = r0;
else if (R2HC_KINDP(kind)) {
r1 = r0 + sz->dims[rnk-1].is;
sz->dims[rnk-1].is *= 2;
} else {
r1 = r0 + sz->dims[rnk-1].os;
sz->dims[rnk-1].os *= 2;
}
p = X(mkproblem_rdft2)(sz, vecsz, r0, r1, cr, ci, kind);
X(tensor_destroy2)(vecsz, sz);
return p;
}