Chris@42: /* Chris@42: * Copyright (c) 2003, 2007-14 Matteo Frigo Chris@42: * Copyright (c) 2003, 2007-14 Massachusetts Institute of Technology Chris@42: * Chris@42: * This program is free software; you can redistribute it and/or modify Chris@42: * it under the terms of the GNU General Public License as published by Chris@42: * the Free Software Foundation; either version 2 of the License, or Chris@42: * (at your option) any later version. Chris@42: * Chris@42: * This program is distributed in the hope that it will be useful, Chris@42: * but WITHOUT ANY WARRANTY; without even the implied warranty of Chris@42: * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the Chris@42: * GNU General Public License for more details. Chris@42: * Chris@42: * You should have received a copy of the GNU General Public License Chris@42: * along with this program; if not, write to the Free Software Chris@42: * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA Chris@42: * Chris@42: */ Chris@42: Chris@42: Chris@42: #include "ifftw.h" Chris@42: Chris@42: INT X(tensor_max_index)(const tensor *sz) Chris@42: { Chris@42: int i; Chris@42: INT ni = 0, no = 0; Chris@42: Chris@42: A(FINITE_RNK(sz->rnk)); Chris@42: for (i = 0; i < sz->rnk; ++i) { Chris@42: const iodim *p = sz->dims + i; Chris@42: ni += (p->n - 1) * X(iabs)(p->is); Chris@42: no += (p->n - 1) * X(iabs)(p->os); Chris@42: } Chris@42: return X(imax)(ni, no); Chris@42: } Chris@42: Chris@42: #define tensor_min_xstride(sz, xs) { \ Chris@42: A(FINITE_RNK(sz->rnk)); \ Chris@42: if (sz->rnk == 0) return 0; \ Chris@42: else { \ Chris@42: int i; \ Chris@42: INT s = X(iabs)(sz->dims[0].xs); \ Chris@42: for (i = 1; i < sz->rnk; ++i) \ Chris@42: s = X(imin)(s, X(iabs)(sz->dims[i].xs)); \ Chris@42: return s; \ Chris@42: } \ Chris@42: } Chris@42: Chris@42: INT X(tensor_min_istride)(const tensor *sz) tensor_min_xstride(sz, is) Chris@42: INT X(tensor_min_ostride)(const tensor *sz) tensor_min_xstride(sz, os) Chris@42: Chris@42: INT X(tensor_min_stride)(const tensor *sz) Chris@42: { Chris@42: return X(imin)(X(tensor_min_istride)(sz), X(tensor_min_ostride)(sz)); Chris@42: } Chris@42: Chris@42: int X(tensor_inplace_strides)(const tensor *sz) Chris@42: { Chris@42: int i; Chris@42: A(FINITE_RNK(sz->rnk)); Chris@42: for (i = 0; i < sz->rnk; ++i) { Chris@42: const iodim *p = sz->dims + i; Chris@42: if (p->is != p->os) Chris@42: return 0; Chris@42: } Chris@42: return 1; Chris@42: } Chris@42: Chris@42: int X(tensor_inplace_strides2)(const tensor *a, const tensor *b) Chris@42: { Chris@42: return X(tensor_inplace_strides(a)) && X(tensor_inplace_strides(b)); Chris@42: } Chris@42: Chris@42: /* return true (1) iff *any* strides of sz decrease when we Chris@42: tensor_inplace_copy(sz, k). */ Chris@42: static int tensor_strides_decrease(const tensor *sz, inplace_kind k) Chris@42: { Chris@42: if (FINITE_RNK(sz->rnk)) { Chris@42: int i; Chris@42: for (i = 0; i < sz->rnk; ++i) Chris@42: if ((sz->dims[i].os - sz->dims[i].is) Chris@42: * (k == INPLACE_OS ? (INT)1 : (INT)-1) < 0) Chris@42: return 1; Chris@42: } Chris@42: return 0; Chris@42: } Chris@42: Chris@42: /* Return true (1) iff *any* strides of sz decrease when we Chris@42: tensor_inplace_copy(k) *or* if *all* strides of sz are unchanged Chris@42: but *any* strides of vecsz decrease. This is used in indirect.c Chris@42: to determine whether to use INPLACE_IS or INPLACE_OS. Chris@42: Chris@42: Note: X(tensor_strides_decrease)(sz, vecsz, INPLACE_IS) Chris@42: || X(tensor_strides_decrease)(sz, vecsz, INPLACE_OS) Chris@42: || X(tensor_inplace_strides2)(p->sz, p->vecsz) Chris@42: must always be true. */ Chris@42: int X(tensor_strides_decrease)(const tensor *sz, const tensor *vecsz, Chris@42: inplace_kind k) Chris@42: { Chris@42: return(tensor_strides_decrease(sz, k) Chris@42: || (X(tensor_inplace_strides)(sz) Chris@42: && tensor_strides_decrease(vecsz, k))); Chris@42: }