Merge from mainline (gomp-merge-2005-02-26).
[official-gcc.git] / libgfortran / generated / matmul_l8.c
blob206097975e8ec76b5e7cb9200c423723a3e54a97
1 /* Implementation of the MATMUL intrinsic
2 Copyright 2002 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., 59 Temple Place - Suite 330,
29 Boston, MA 02111-1307, USA. */
31 #include "config.h"
32 #include <stdlib.h>
33 #include <assert.h>
34 #include "libgfortran.h"
36 /* Dimensions: retarray(x,y) a(x, count) b(count,y).
37 Either a or b can be rank 1. In this case x or y is 1. */
39 extern void matmul_l8 (gfc_array_l8 *, gfc_array_l4 *, gfc_array_l4 *);
40 export_proto(matmul_l8);
42 void
43 matmul_l8 (gfc_array_l8 * retarray, gfc_array_l4 * a, gfc_array_l4 * b)
45 GFC_INTEGER_4 *abase;
46 GFC_INTEGER_4 *bbase;
47 GFC_LOGICAL_8 *dest;
48 index_type rxstride;
49 index_type rystride;
50 index_type xcount;
51 index_type ycount;
52 index_type xstride;
53 index_type ystride;
54 index_type x;
55 index_type y;
57 GFC_INTEGER_4 *pa;
58 GFC_INTEGER_4 *pb;
59 index_type astride;
60 index_type bstride;
61 index_type count;
62 index_type n;
64 assert (GFC_DESCRIPTOR_RANK (a) == 2
65 || GFC_DESCRIPTOR_RANK (b) == 2);
67 if (retarray->data == NULL)
69 if (GFC_DESCRIPTOR_RANK (a) == 1)
71 retarray->dim[0].lbound = 0;
72 retarray->dim[0].ubound = b->dim[1].ubound - b->dim[1].lbound;
73 retarray->dim[0].stride = 1;
75 else if (GFC_DESCRIPTOR_RANK (b) == 1)
77 retarray->dim[0].lbound = 0;
78 retarray->dim[0].ubound = a->dim[0].ubound - a->dim[0].lbound;
79 retarray->dim[0].stride = 1;
81 else
83 retarray->dim[0].lbound = 0;
84 retarray->dim[0].ubound = a->dim[0].ubound - a->dim[0].lbound;
85 retarray->dim[0].stride = 1;
87 retarray->dim[1].lbound = 0;
88 retarray->dim[1].ubound = b->dim[1].ubound - b->dim[1].lbound;
89 retarray->dim[1].stride = retarray->dim[0].ubound+1;
92 retarray->data
93 = internal_malloc_size (sizeof (GFC_LOGICAL_8) * size0 (retarray));
94 retarray->base = 0;
97 abase = a->data;
98 if (GFC_DESCRIPTOR_SIZE (a) != 4)
100 assert (GFC_DESCRIPTOR_SIZE (a) == 8);
101 abase = GFOR_POINTER_L8_TO_L4 (abase);
102 astride <<= 1;
104 bbase = b->data;
105 if (GFC_DESCRIPTOR_SIZE (b) != 4)
107 assert (GFC_DESCRIPTOR_SIZE (b) == 8);
108 bbase = GFOR_POINTER_L8_TO_L4 (bbase);
109 bstride <<= 1;
111 dest = retarray->data;
113 if (retarray->dim[0].stride == 0)
114 retarray->dim[0].stride = 1;
115 if (a->dim[0].stride == 0)
116 a->dim[0].stride = 1;
117 if (b->dim[0].stride == 0)
118 b->dim[0].stride = 1;
121 if (GFC_DESCRIPTOR_RANK (retarray) == 1)
123 rxstride = retarray->dim[0].stride;
124 rystride = rxstride;
126 else
128 rxstride = retarray->dim[0].stride;
129 rystride = retarray->dim[1].stride;
132 /* If we have rank 1 parameters, zero the absent stride, and set the size to
133 one. */
134 if (GFC_DESCRIPTOR_RANK (a) == 1)
136 astride = a->dim[0].stride;
137 count = a->dim[0].ubound + 1 - a->dim[0].lbound;
138 xstride = 0;
139 rxstride = 0;
140 xcount = 1;
142 else
144 astride = a->dim[1].stride;
145 count = a->dim[1].ubound + 1 - a->dim[1].lbound;
146 xstride = a->dim[0].stride;
147 xcount = a->dim[0].ubound + 1 - a->dim[0].lbound;
149 if (GFC_DESCRIPTOR_RANK (b) == 1)
151 bstride = b->dim[0].stride;
152 assert(count == b->dim[0].ubound + 1 - b->dim[0].lbound);
153 ystride = 0;
154 rystride = 0;
155 ycount = 1;
157 else
159 bstride = b->dim[0].stride;
160 assert(count == b->dim[0].ubound + 1 - b->dim[0].lbound);
161 ystride = b->dim[1].stride;
162 ycount = b->dim[1].ubound + 1 - b->dim[1].lbound;
165 for (y = 0; y < ycount; y++)
167 for (x = 0; x < xcount; x++)
169 /* Do the summation for this element. For real and integer types
170 this is the same as DOT_PRODUCT. For complex types we use do
171 a*b, not conjg(a)*b. */
172 pa = abase;
173 pb = bbase;
174 *dest = 0;
176 for (n = 0; n < count; n++)
178 if (*pa && *pb)
180 *dest = 1;
181 break;
183 pa += astride;
184 pb += bstride;
187 dest += rxstride;
188 abase += xstride;
190 abase -= xstride * xcount;
191 bbase += ystride;
192 dest += rystride - (rxstride * xcount);