]>
Commit | Line | Data |
---|---|---|
6de9cd9a DN |
1 | /* Implementation of the MAXLOC intrinsic |
2 | Copyright 2002 Free Software Foundation, Inc. | |
3 | Contributed by Paul Brook <paul@nowt.org> | |
4 | ||
57dea9f6 | 5 | This file is part of the GNU Fortran 95 runtime library (libgfortran). |
6de9cd9a DN |
6 | |
7 | Libgfortran is free software; you can redistribute it and/or | |
57dea9f6 | 8 | modify it under the terms of the GNU General Public |
6de9cd9a | 9 | License as published by the Free Software Foundation; either |
57dea9f6 TM |
10 | version 2 of the License, or (at your option) any later version. |
11 | ||
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.) | |
6de9cd9a DN |
20 | |
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 | |
57dea9f6 | 24 | GNU General Public License for more details. |
6de9cd9a | 25 | |
57dea9f6 TM |
26 | You should have received a copy of the GNU General Public |
27 | License along with libgfortran; see the file COPYING. If not, | |
fe2ae685 KC |
28 | write to the Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor, |
29 | Boston, MA 02110-1301, USA. */ | |
6de9cd9a DN |
30 | |
31 | #include "config.h" | |
32 | #include <stdlib.h> | |
33 | #include <assert.h> | |
34 | #include <float.h> | |
35 | #include <limits.h> | |
36 | #include "libgfortran.h" | |
37 | ||
7d7b8bfe | 38 | |
644cb69f FXC |
39 | #if defined (HAVE_GFC_INTEGER_8) && defined (HAVE_GFC_INTEGER_4) |
40 | ||
41 | ||
64acfd99 JB |
42 | extern void maxloc1_4_i8 (gfc_array_i4 * const restrict, |
43 | gfc_array_i8 * const restrict, const index_type * const restrict); | |
7f68c75f | 44 | export_proto(maxloc1_4_i8); |
7d7b8bfe | 45 | |
6de9cd9a | 46 | void |
64acfd99 JB |
47 | maxloc1_4_i8 (gfc_array_i4 * const restrict retarray, |
48 | gfc_array_i8 * const restrict array, | |
49 | const index_type * const restrict pdim) | |
6de9cd9a | 50 | { |
e33e218b TK |
51 | index_type count[GFC_MAX_DIMENSIONS]; |
52 | index_type extent[GFC_MAX_DIMENSIONS]; | |
53 | index_type sstride[GFC_MAX_DIMENSIONS]; | |
54 | index_type dstride[GFC_MAX_DIMENSIONS]; | |
64acfd99 JB |
55 | const GFC_INTEGER_8 * restrict base; |
56 | GFC_INTEGER_4 * restrict dest; | |
6de9cd9a DN |
57 | index_type rank; |
58 | index_type n; | |
59 | index_type len; | |
60 | index_type delta; | |
61 | index_type dim; | |
62 | ||
63 | /* Make dim zero based to avoid confusion. */ | |
64 | dim = (*pdim) - 1; | |
65 | rank = GFC_DESCRIPTOR_RANK (array) - 1; | |
e33e218b TK |
66 | |
67 | /* TODO: It should be a front end job to correctly set the strides. */ | |
68 | ||
6de9cd9a DN |
69 | if (array->dim[0].stride == 0) |
70 | array->dim[0].stride = 1; | |
6de9cd9a DN |
71 | |
72 | len = array->dim[dim].ubound + 1 - array->dim[dim].lbound; | |
73 | delta = array->dim[dim].stride; | |
74 | ||
75 | for (n = 0; n < dim; n++) | |
76 | { | |
77 | sstride[n] = array->dim[n].stride; | |
78 | extent[n] = array->dim[n].ubound + 1 - array->dim[n].lbound; | |
79 | } | |
80 | for (n = dim; n < rank; n++) | |
81 | { | |
82 | sstride[n] = array->dim[n + 1].stride; | |
83 | extent[n] = | |
84 | array->dim[n + 1].ubound + 1 - array->dim[n + 1].lbound; | |
85 | } | |
86 | ||
6c167c45 VL |
87 | if (retarray->data == NULL) |
88 | { | |
89 | for (n = 0; n < rank; n++) | |
90 | { | |
91 | retarray->dim[n].lbound = 0; | |
92 | retarray->dim[n].ubound = extent[n]-1; | |
93 | if (n == 0) | |
94 | retarray->dim[n].stride = 1; | |
95 | else | |
96 | retarray->dim[n].stride = retarray->dim[n-1].stride * extent[n-1]; | |
97 | } | |
98 | ||
07d3cebe RH |
99 | retarray->data |
100 | = internal_malloc_size (sizeof (GFC_INTEGER_4) | |
101 | * retarray->dim[rank-1].stride | |
102 | * extent[rank-1]); | |
efd4dc1a | 103 | retarray->offset = 0; |
50dd63a9 | 104 | retarray->dtype = (array->dtype & ~GFC_DTYPE_RANK_MASK) | rank; |
6c167c45 | 105 | } |
50dd63a9 TK |
106 | else |
107 | { | |
108 | if (retarray->dim[0].stride == 0) | |
109 | retarray->dim[0].stride = 1; | |
110 | ||
111 | if (rank != GFC_DESCRIPTOR_RANK (retarray)) | |
112 | runtime_error ("rank of return array incorrect"); | |
113 | } | |
114 | ||
6de9cd9a DN |
115 | for (n = 0; n < rank; n++) |
116 | { | |
117 | count[n] = 0; | |
118 | dstride[n] = retarray->dim[n].stride; | |
119 | if (extent[n] <= 0) | |
120 | len = 0; | |
121 | } | |
122 | ||
123 | base = array->data; | |
124 | dest = retarray->data; | |
125 | ||
126 | while (base) | |
127 | { | |
64acfd99 | 128 | const GFC_INTEGER_8 * restrict src; |
6de9cd9a DN |
129 | GFC_INTEGER_4 result; |
130 | src = base; | |
131 | { | |
132 | ||
133 | GFC_INTEGER_8 maxval; | |
134 | maxval = -GFC_INTEGER_8_HUGE; | |
135 | result = 1; | |
136 | if (len <= 0) | |
137 | *dest = 0; | |
138 | else | |
139 | { | |
140 | for (n = 0; n < len; n++, src += delta) | |
141 | { | |
142 | ||
143 | if (*src > maxval) | |
144 | { | |
145 | maxval = *src; | |
146 | result = (GFC_INTEGER_4)n + 1; | |
147 | } | |
148 | } | |
149 | *dest = result; | |
150 | } | |
151 | } | |
152 | /* Advance to the next element. */ | |
153 | count[0]++; | |
154 | base += sstride[0]; | |
155 | dest += dstride[0]; | |
156 | n = 0; | |
157 | while (count[n] == extent[n]) | |
158 | { | |
159 | /* When we get to the end of a dimension, reset it and increment | |
160 | the next dimension. */ | |
161 | count[n] = 0; | |
162 | /* We could precalculate these products, but this is a less | |
163 | frequently used path so proabably not worth it. */ | |
164 | base -= sstride[n] * extent[n]; | |
165 | dest -= dstride[n] * extent[n]; | |
166 | n++; | |
167 | if (n == rank) | |
168 | { | |
169 | /* Break out of the look. */ | |
170 | base = NULL; | |
171 | break; | |
172 | } | |
173 | else | |
174 | { | |
175 | count[n]++; | |
176 | base += sstride[n]; | |
177 | dest += dstride[n]; | |
178 | } | |
179 | } | |
180 | } | |
181 | } | |
182 | ||
7d7b8bfe | 183 | |
64acfd99 JB |
184 | extern void mmaxloc1_4_i8 (gfc_array_i4 * const restrict, |
185 | gfc_array_i8 * const restrict, const index_type * const restrict, | |
186 | gfc_array_l4 * const restrict); | |
7f68c75f | 187 | export_proto(mmaxloc1_4_i8); |
7d7b8bfe | 188 | |
6de9cd9a | 189 | void |
64acfd99 JB |
190 | mmaxloc1_4_i8 (gfc_array_i4 * const restrict retarray, |
191 | gfc_array_i8 * const restrict array, | |
192 | const index_type * const restrict pdim, | |
193 | gfc_array_l4 * const restrict mask) | |
6de9cd9a | 194 | { |
e33e218b TK |
195 | index_type count[GFC_MAX_DIMENSIONS]; |
196 | index_type extent[GFC_MAX_DIMENSIONS]; | |
197 | index_type sstride[GFC_MAX_DIMENSIONS]; | |
198 | index_type dstride[GFC_MAX_DIMENSIONS]; | |
199 | index_type mstride[GFC_MAX_DIMENSIONS]; | |
64acfd99 JB |
200 | GFC_INTEGER_4 * restrict dest; |
201 | const GFC_INTEGER_8 * restrict base; | |
202 | const GFC_LOGICAL_4 * restrict mbase; | |
6de9cd9a DN |
203 | int rank; |
204 | int dim; | |
205 | index_type n; | |
206 | index_type len; | |
207 | index_type delta; | |
208 | index_type mdelta; | |
209 | ||
210 | dim = (*pdim) - 1; | |
211 | rank = GFC_DESCRIPTOR_RANK (array) - 1; | |
e33e218b TK |
212 | |
213 | /* TODO: It should be a front end job to correctly set the strides. */ | |
214 | ||
6de9cd9a DN |
215 | if (array->dim[0].stride == 0) |
216 | array->dim[0].stride = 1; | |
6de9cd9a | 217 | |
c6abe94d TK |
218 | if (mask->dim[0].stride == 0) |
219 | mask->dim[0].stride = 1; | |
220 | ||
6de9cd9a DN |
221 | len = array->dim[dim].ubound + 1 - array->dim[dim].lbound; |
222 | if (len <= 0) | |
223 | return; | |
224 | delta = array->dim[dim].stride; | |
225 | mdelta = mask->dim[dim].stride; | |
226 | ||
227 | for (n = 0; n < dim; n++) | |
228 | { | |
229 | sstride[n] = array->dim[n].stride; | |
230 | mstride[n] = mask->dim[n].stride; | |
231 | extent[n] = array->dim[n].ubound + 1 - array->dim[n].lbound; | |
232 | } | |
233 | for (n = dim; n < rank; n++) | |
234 | { | |
235 | sstride[n] = array->dim[n + 1].stride; | |
236 | mstride[n] = mask->dim[n + 1].stride; | |
237 | extent[n] = | |
238 | array->dim[n + 1].ubound + 1 - array->dim[n + 1].lbound; | |
239 | } | |
240 | ||
50dd63a9 TK |
241 | if (retarray->data == NULL) |
242 | { | |
243 | for (n = 0; n < rank; n++) | |
244 | { | |
245 | retarray->dim[n].lbound = 0; | |
246 | retarray->dim[n].ubound = extent[n]-1; | |
247 | if (n == 0) | |
248 | retarray->dim[n].stride = 1; | |
249 | else | |
250 | retarray->dim[n].stride = retarray->dim[n-1].stride * extent[n-1]; | |
251 | } | |
252 | ||
253 | retarray->data | |
254 | = internal_malloc_size (sizeof (GFC_INTEGER_4) | |
255 | * retarray->dim[rank-1].stride | |
256 | * extent[rank-1]); | |
efd4dc1a | 257 | retarray->offset = 0; |
50dd63a9 TK |
258 | retarray->dtype = (array->dtype & ~GFC_DTYPE_RANK_MASK) | rank; |
259 | } | |
260 | else | |
261 | { | |
262 | if (retarray->dim[0].stride == 0) | |
263 | retarray->dim[0].stride = 1; | |
264 | ||
265 | if (rank != GFC_DESCRIPTOR_RANK (retarray)) | |
266 | runtime_error ("rank of return array incorrect"); | |
267 | } | |
268 | ||
6de9cd9a DN |
269 | for (n = 0; n < rank; n++) |
270 | { | |
271 | count[n] = 0; | |
272 | dstride[n] = retarray->dim[n].stride; | |
273 | if (extent[n] <= 0) | |
274 | return; | |
275 | } | |
276 | ||
277 | dest = retarray->data; | |
278 | base = array->data; | |
279 | mbase = mask->data; | |
280 | ||
281 | if (GFC_DESCRIPTOR_SIZE (mask) != 4) | |
282 | { | |
283 | /* This allows the same loop to be used for all logical types. */ | |
284 | assert (GFC_DESCRIPTOR_SIZE (mask) == 8); | |
285 | for (n = 0; n < rank; n++) | |
286 | mstride[n] <<= 1; | |
287 | mdelta <<= 1; | |
288 | mbase = (GFOR_POINTER_L8_TO_L4 (mbase)); | |
289 | } | |
290 | ||
291 | while (base) | |
292 | { | |
64acfd99 JB |
293 | const GFC_INTEGER_8 * restrict src; |
294 | const GFC_LOGICAL_4 * restrict msrc; | |
6de9cd9a DN |
295 | GFC_INTEGER_4 result; |
296 | src = base; | |
297 | msrc = mbase; | |
298 | { | |
299 | ||
300 | GFC_INTEGER_8 maxval; | |
301 | maxval = -GFC_INTEGER_8_HUGE; | |
302 | result = 1; | |
303 | if (len <= 0) | |
304 | *dest = 0; | |
305 | else | |
306 | { | |
307 | for (n = 0; n < len; n++, src += delta, msrc += mdelta) | |
308 | { | |
309 | ||
310 | if (*msrc && *src > maxval) | |
311 | { | |
312 | maxval = *src; | |
313 | result = (GFC_INTEGER_4)n + 1; | |
314 | } | |
315 | } | |
316 | *dest = result; | |
317 | } | |
318 | } | |
319 | /* Advance to the next element. */ | |
320 | count[0]++; | |
321 | base += sstride[0]; | |
322 | mbase += mstride[0]; | |
323 | dest += dstride[0]; | |
324 | n = 0; | |
325 | while (count[n] == extent[n]) | |
326 | { | |
327 | /* When we get to the end of a dimension, reset it and increment | |
328 | the next dimension. */ | |
329 | count[n] = 0; | |
330 | /* We could precalculate these products, but this is a less | |
331 | frequently used path so proabably not worth it. */ | |
332 | base -= sstride[n] * extent[n]; | |
333 | mbase -= mstride[n] * extent[n]; | |
334 | dest -= dstride[n] * extent[n]; | |
335 | n++; | |
336 | if (n == rank) | |
337 | { | |
338 | /* Break out of the look. */ | |
339 | base = NULL; | |
340 | break; | |
341 | } | |
342 | else | |
343 | { | |
344 | count[n]++; | |
345 | base += sstride[n]; | |
346 | mbase += mstride[n]; | |
347 | dest += dstride[n]; | |
348 | } | |
349 | } | |
350 | } | |
351 | } | |
352 | ||
644cb69f | 353 | #endif |