* config/rs6000/rs6000.md (floatsitf2): Remove constraints.
[official-gcc.git] / libgfortran / generated / matmul_l16.c
blobcaa1b11910e0273ba6b65dbacc8304c349fff9b1
1 /* Implementation of the MATMUL intrinsic
2 Copyright 2002, 2005 Free Software Foundation, Inc.
3 Contributed by Paul Brook <paul@nowt.org>
5 This file is part of the GNU Fortran 95 runtime library (libgfortran).
7 Libgfortran is free software; you can redistribute it and/or
8 modify it under the terms of the GNU General Public
9 License as published by the Free Software Foundation; either
10 version 2 of the License, or (at your option) any later version.
12 In addition to the permissions in the GNU General Public License, the
13 Free Software Foundation gives you unlimited permission to link the
14 compiled version of this file into combinations with other programs,
15 and to distribute those combinations without any restriction coming
16 from the use of this file. (The General Public License restrictions
17 do apply in other respects; for example, they cover modification of
18 the file, and distribution when not linked into a combine
19 executable.)
21 Libgfortran is distributed in the hope that it will be useful,
22 but WITHOUT ANY WARRANTY; without even the implied warranty of
23 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
24 GNU General Public License for more details.
26 You should have received a copy of the GNU General Public
27 License along with libgfortran; see the file COPYING. If not,
28 write to the Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor,
29 Boston, MA 02110-1301, USA. */
31 #include "config.h"
32 #include <stdlib.h>
33 #include <assert.h>
34 #include "libgfortran.h"
36 #if defined (HAVE_GFC_LOGICAL_16)
38 /* Dimensions: retarray(x,y) a(x, count) b(count,y).
39 Either a or b can be rank 1. In this case x or y is 1. */
41 extern void matmul_l16 (gfc_array_l16 * const restrict,
42 gfc_array_l4 * const restrict, gfc_array_l4 * const restrict);
43 export_proto(matmul_l16);
45 void
46 matmul_l16 (gfc_array_l16 * const restrict retarray,
47 gfc_array_l4 * const restrict a, gfc_array_l4 * const restrict b)
49 const GFC_INTEGER_4 * restrict abase;
50 const GFC_INTEGER_4 * restrict bbase;
51 GFC_LOGICAL_16 * restrict dest;
52 index_type rxstride;
53 index_type rystride;
54 index_type xcount;
55 index_type ycount;
56 index_type xstride;
57 index_type ystride;
58 index_type x;
59 index_type y;
61 const GFC_INTEGER_4 * restrict pa;
62 const GFC_INTEGER_4 * restrict pb;
63 index_type astride;
64 index_type bstride;
65 index_type count;
66 index_type n;
68 assert (GFC_DESCRIPTOR_RANK (a) == 2
69 || GFC_DESCRIPTOR_RANK (b) == 2);
71 if (retarray->data == NULL)
73 if (GFC_DESCRIPTOR_RANK (a) == 1)
75 retarray->dim[0].lbound = 0;
76 retarray->dim[0].ubound = b->dim[1].ubound - b->dim[1].lbound;
77 retarray->dim[0].stride = 1;
79 else if (GFC_DESCRIPTOR_RANK (b) == 1)
81 retarray->dim[0].lbound = 0;
82 retarray->dim[0].ubound = a->dim[0].ubound - a->dim[0].lbound;
83 retarray->dim[0].stride = 1;
85 else
87 retarray->dim[0].lbound = 0;
88 retarray->dim[0].ubound = a->dim[0].ubound - a->dim[0].lbound;
89 retarray->dim[0].stride = 1;
91 retarray->dim[1].lbound = 0;
92 retarray->dim[1].ubound = b->dim[1].ubound - b->dim[1].lbound;
93 retarray->dim[1].stride = retarray->dim[0].ubound+1;
96 retarray->data
97 = internal_malloc_size (sizeof (GFC_LOGICAL_16) * size0 ((array_t *) retarray));
98 retarray->offset = 0;
101 abase = a->data;
102 if (GFC_DESCRIPTOR_SIZE (a) != 4)
104 assert (GFC_DESCRIPTOR_SIZE (a) == 8);
105 abase = GFOR_POINTER_L8_TO_L4 (abase);
107 bbase = b->data;
108 if (GFC_DESCRIPTOR_SIZE (b) != 4)
110 assert (GFC_DESCRIPTOR_SIZE (b) == 8);
111 bbase = GFOR_POINTER_L8_TO_L4 (bbase);
113 dest = retarray->data;
115 if (retarray->dim[0].stride == 0)
116 retarray->dim[0].stride = 1;
117 if (a->dim[0].stride == 0)
118 a->dim[0].stride = 1;
119 if (b->dim[0].stride == 0)
120 b->dim[0].stride = 1;
123 if (GFC_DESCRIPTOR_RANK (retarray) == 1)
125 rxstride = retarray->dim[0].stride;
126 rystride = rxstride;
128 else
130 rxstride = retarray->dim[0].stride;
131 rystride = retarray->dim[1].stride;
134 /* If we have rank 1 parameters, zero the absent stride, and set the size to
135 one. */
136 if (GFC_DESCRIPTOR_RANK (a) == 1)
138 astride = a->dim[0].stride;
139 count = a->dim[0].ubound + 1 - a->dim[0].lbound;
140 xstride = 0;
141 rxstride = 0;
142 xcount = 1;
144 else
146 astride = a->dim[1].stride;
147 count = a->dim[1].ubound + 1 - a->dim[1].lbound;
148 xstride = a->dim[0].stride;
149 xcount = a->dim[0].ubound + 1 - a->dim[0].lbound;
151 if (GFC_DESCRIPTOR_RANK (b) == 1)
153 bstride = b->dim[0].stride;
154 assert(count == b->dim[0].ubound + 1 - b->dim[0].lbound);
155 ystride = 0;
156 rystride = 0;
157 ycount = 1;
159 else
161 bstride = b->dim[0].stride;
162 assert(count == b->dim[0].ubound + 1 - b->dim[0].lbound);
163 ystride = b->dim[1].stride;
164 ycount = b->dim[1].ubound + 1 - b->dim[1].lbound;
167 for (y = 0; y < ycount; y++)
169 for (x = 0; x < xcount; x++)
171 /* Do the summation for this element. For real and integer types
172 this is the same as DOT_PRODUCT. For complex types we use do
173 a*b, not conjg(a)*b. */
174 pa = abase;
175 pb = bbase;
176 *dest = 0;
178 for (n = 0; n < count; n++)
180 if (*pa && *pb)
182 *dest = 1;
183 break;
185 pa += astride;
186 pb += bstride;
189 dest += rxstride;
190 abase += xstride;
192 abase -= xstride * xcount;
193 bbase += ystride;
194 dest += rystride - (rxstride * xcount);
198 #endif