1 /* Implementation of the MATMUL intrinsic
2 Copyright 2002, 2005, 2006, 2007 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
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 "libgfortran.h"
36 #if defined (HAVE_GFC_LOGICAL_8)
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_l8 (gfc_array_l8
* const restrict
,
42 gfc_array_l1
* const restrict
, gfc_array_l1
* const restrict
);
43 export_proto(matmul_l8
);
46 matmul_l8 (gfc_array_l8
* const restrict retarray
,
47 gfc_array_l1
* const restrict a
, gfc_array_l1
* const restrict b
)
49 const GFC_LOGICAL_1
* restrict abase
;
50 const GFC_LOGICAL_1
* restrict bbase
;
51 GFC_LOGICAL_8
* restrict dest
;
63 const GFC_LOGICAL_1
* restrict pa
;
64 const GFC_LOGICAL_1
* restrict pb
;
70 assert (GFC_DESCRIPTOR_RANK (a
) == 2
71 || GFC_DESCRIPTOR_RANK (b
) == 2);
73 if (retarray
->data
== NULL
)
75 if (GFC_DESCRIPTOR_RANK (a
) == 1)
77 retarray
->dim
[0].lbound
= 0;
78 retarray
->dim
[0].ubound
= b
->dim
[1].ubound
- b
->dim
[1].lbound
;
79 retarray
->dim
[0].stride
= 1;
81 else if (GFC_DESCRIPTOR_RANK (b
) == 1)
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;
89 retarray
->dim
[0].lbound
= 0;
90 retarray
->dim
[0].ubound
= a
->dim
[0].ubound
- a
->dim
[0].lbound
;
91 retarray
->dim
[0].stride
= 1;
93 retarray
->dim
[1].lbound
= 0;
94 retarray
->dim
[1].ubound
= b
->dim
[1].ubound
- b
->dim
[1].lbound
;
95 retarray
->dim
[1].stride
= retarray
->dim
[0].ubound
+1;
99 = internal_malloc_size (sizeof (GFC_LOGICAL_8
) * size0 ((array_t
*) retarray
));
100 retarray
->offset
= 0;
102 else if (compile_options
.bounds_check
)
104 index_type ret_extent
, arg_extent
;
106 if (GFC_DESCRIPTOR_RANK (a
) == 1)
108 arg_extent
= b
->dim
[1].ubound
+ 1 - b
->dim
[1].lbound
;
109 ret_extent
= retarray
->dim
[0].ubound
+ 1 - retarray
->dim
[0].lbound
;
110 if (arg_extent
!= ret_extent
)
111 runtime_error ("Incorrect extent in return array in"
112 " MATMUL intrinsic: is %ld, should be %ld",
113 (long int) ret_extent
, (long int) arg_extent
);
115 else if (GFC_DESCRIPTOR_RANK (b
) == 1)
117 arg_extent
= a
->dim
[0].ubound
+ 1 - a
->dim
[0].lbound
;
118 ret_extent
= retarray
->dim
[0].ubound
+ 1 - retarray
->dim
[0].lbound
;
119 if (arg_extent
!= ret_extent
)
120 runtime_error ("Incorrect extent in return array in"
121 " MATMUL intrinsic: is %ld, should be %ld",
122 (long int) ret_extent
, (long int) arg_extent
);
126 arg_extent
= a
->dim
[0].ubound
+ 1 - a
->dim
[0].lbound
;
127 ret_extent
= retarray
->dim
[0].ubound
+ 1 - retarray
->dim
[0].lbound
;
128 if (arg_extent
!= ret_extent
)
129 runtime_error ("Incorrect extent in return array in"
130 " MATMUL intrinsic for dimension 1:"
131 " is %ld, should be %ld",
132 (long int) ret_extent
, (long int) arg_extent
);
134 arg_extent
= b
->dim
[1].ubound
+ 1 - b
->dim
[1].lbound
;
135 ret_extent
= retarray
->dim
[1].ubound
+ 1 - retarray
->dim
[1].lbound
;
136 if (arg_extent
!= ret_extent
)
137 runtime_error ("Incorrect extent in return array in"
138 " MATMUL intrinsic for dimension 2:"
139 " is %ld, should be %ld",
140 (long int) ret_extent
, (long int) arg_extent
);
145 a_kind
= GFC_DESCRIPTOR_SIZE (a
);
147 if (a_kind
== 1 || a_kind
== 2 || a_kind
== 4 || a_kind
== 8
148 #ifdef HAVE_GFC_LOGICAL_16
152 abase
= GFOR_POINTER_TO_L1 (abase
, a_kind
);
154 internal_error (NULL
, "Funny sized logical array");
157 b_kind
= GFC_DESCRIPTOR_SIZE (b
);
159 if (b_kind
== 1 || b_kind
== 2 || b_kind
== 4 || b_kind
== 8
160 #ifdef HAVE_GFC_LOGICAL_16
164 bbase
= GFOR_POINTER_TO_L1 (bbase
, b_kind
);
166 internal_error (NULL
, "Funny sized logical array");
168 dest
= retarray
->data
;
171 if (GFC_DESCRIPTOR_RANK (retarray
) == 1)
173 rxstride
= retarray
->dim
[0].stride
;
178 rxstride
= retarray
->dim
[0].stride
;
179 rystride
= retarray
->dim
[1].stride
;
182 /* If we have rank 1 parameters, zero the absent stride, and set the size to
184 if (GFC_DESCRIPTOR_RANK (a
) == 1)
186 astride
= a
->dim
[0].stride
* a_kind
;
187 count
= a
->dim
[0].ubound
+ 1 - a
->dim
[0].lbound
;
194 astride
= a
->dim
[1].stride
* a_kind
;
195 count
= a
->dim
[1].ubound
+ 1 - a
->dim
[1].lbound
;
196 xstride
= a
->dim
[0].stride
* a_kind
;
197 xcount
= a
->dim
[0].ubound
+ 1 - a
->dim
[0].lbound
;
199 if (GFC_DESCRIPTOR_RANK (b
) == 1)
201 bstride
= b
->dim
[0].stride
* b_kind
;
202 assert(count
== b
->dim
[0].ubound
+ 1 - b
->dim
[0].lbound
);
209 bstride
= b
->dim
[0].stride
* b_kind
;
210 assert(count
== b
->dim
[0].ubound
+ 1 - b
->dim
[0].lbound
);
211 ystride
= b
->dim
[1].stride
* b_kind
;
212 ycount
= b
->dim
[1].ubound
+ 1 - b
->dim
[1].lbound
;
215 for (y
= 0; y
< ycount
; y
++)
217 for (x
= 0; x
< xcount
; x
++)
219 /* Do the summation for this element. For real and integer types
220 this is the same as DOT_PRODUCT. For complex types we use do
221 a*b, not conjg(a)*b. */
226 for (n
= 0; n
< count
; n
++)
240 abase
-= xstride
* xcount
;
242 dest
+= rystride
- (rxstride
* xcount
);