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: Chris@10: #include "ifftw.h" Chris@10: Chris@10: INT X(tensor_max_index)(const tensor *sz) Chris@10: { Chris@10: int i; Chris@10: INT ni = 0, no = 0; Chris@10: Chris@10: A(FINITE_RNK(sz->rnk)); Chris@10: for (i = 0; i < sz->rnk; ++i) { Chris@10: const iodim *p = sz->dims + i; Chris@10: ni += (p->n - 1) * X(iabs)(p->is); Chris@10: no += (p->n - 1) * X(iabs)(p->os); Chris@10: } Chris@10: return X(imax)(ni, no); Chris@10: } Chris@10: Chris@10: #define tensor_min_xstride(sz, xs) { \ Chris@10: A(FINITE_RNK(sz->rnk)); \ Chris@10: if (sz->rnk == 0) return 0; \ Chris@10: else { \ Chris@10: int i; \ Chris@10: INT s = X(iabs)(sz->dims[0].xs); \ Chris@10: for (i = 1; i < sz->rnk; ++i) \ Chris@10: s = X(imin)(s, X(iabs)(sz->dims[i].xs)); \ Chris@10: return s; \ Chris@10: } \ Chris@10: } Chris@10: Chris@10: INT X(tensor_min_istride)(const tensor *sz) tensor_min_xstride(sz, is) Chris@10: INT X(tensor_min_ostride)(const tensor *sz) tensor_min_xstride(sz, os) Chris@10: Chris@10: INT X(tensor_min_stride)(const tensor *sz) Chris@10: { Chris@10: return X(imin)(X(tensor_min_istride)(sz), X(tensor_min_ostride)(sz)); Chris@10: } Chris@10: Chris@10: int X(tensor_inplace_strides)(const tensor *sz) Chris@10: { Chris@10: int i; Chris@10: A(FINITE_RNK(sz->rnk)); Chris@10: for (i = 0; i < sz->rnk; ++i) { Chris@10: const iodim *p = sz->dims + i; Chris@10: if (p->is != p->os) Chris@10: return 0; Chris@10: } Chris@10: return 1; Chris@10: } Chris@10: Chris@10: int X(tensor_inplace_strides2)(const tensor *a, const tensor *b) Chris@10: { Chris@10: return X(tensor_inplace_strides(a)) && X(tensor_inplace_strides(b)); Chris@10: } Chris@10: Chris@10: /* return true (1) iff *any* strides of sz decrease when we Chris@10: tensor_inplace_copy(sz, k). */ Chris@10: static int tensor_strides_decrease(const tensor *sz, inplace_kind k) Chris@10: { Chris@10: if (FINITE_RNK(sz->rnk)) { Chris@10: int i; Chris@10: for (i = 0; i < sz->rnk; ++i) Chris@10: if ((sz->dims[i].os - sz->dims[i].is) Chris@10: * (k == INPLACE_OS ? (INT)1 : (INT)-1) < 0) Chris@10: return 1; Chris@10: } Chris@10: return 0; Chris@10: } Chris@10: Chris@10: /* Return true (1) iff *any* strides of sz decrease when we Chris@10: tensor_inplace_copy(k) *or* if *all* strides of sz are unchanged Chris@10: but *any* strides of vecsz decrease. This is used in indirect.c Chris@10: to determine whether to use INPLACE_IS or INPLACE_OS. Chris@10: Chris@10: Note: X(tensor_strides_decrease)(sz, vecsz, INPLACE_IS) Chris@10: || X(tensor_strides_decrease)(sz, vecsz, INPLACE_OS) Chris@10: || X(tensor_inplace_strides2)(p->sz, p->vecsz) Chris@10: must always be true. */ Chris@10: int X(tensor_strides_decrease)(const tensor *sz, const tensor *vecsz, Chris@10: inplace_kind k) Chris@10: { Chris@10: return(tensor_strides_decrease(sz, k) Chris@10: || (X(tensor_inplace_strides)(sz) Chris@10: && tensor_strides_decrease(vecsz, k))); Chris@10: }