]>
Commit | Line | Data |
---|---|---|
644cb69f FXC |
1 | /* Implementation of the SUM intrinsic |
2 | Copyright 2002 Free Software Foundation, Inc. | |
3 | Contributed by Paul Brook <paul@nowt.org> | |
4 | ||
5 | This file is part of the GNU Fortran 95 runtime library (libgfortran). | |
6 | ||
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. | |
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.) | |
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 | |
24 | GNU General Public License for more details. | |
25 | ||
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. */ | |
30 | ||
31 | #include "config.h" | |
32 | #include <stdlib.h> | |
33 | #include <assert.h> | |
34 | #include "libgfortran.h" | |
35 | ||
36 | ||
37 | #if defined (HAVE_GFC_INTEGER_16) && defined (HAVE_GFC_INTEGER_16) | |
38 | ||
39 | ||
64acfd99 JB |
40 | extern void sum_i16 (gfc_array_i16 * const restrict, |
41 | gfc_array_i16 * const restrict, const index_type * const restrict); | |
644cb69f FXC |
42 | export_proto(sum_i16); |
43 | ||
44 | void | |
64acfd99 JB |
45 | sum_i16 (gfc_array_i16 * const restrict retarray, |
46 | gfc_array_i16 * const restrict array, | |
47 | const index_type * const restrict pdim) | |
644cb69f FXC |
48 | { |
49 | index_type count[GFC_MAX_DIMENSIONS]; | |
50 | index_type extent[GFC_MAX_DIMENSIONS]; | |
51 | index_type sstride[GFC_MAX_DIMENSIONS]; | |
52 | index_type dstride[GFC_MAX_DIMENSIONS]; | |
64acfd99 JB |
53 | const GFC_INTEGER_16 * restrict base; |
54 | GFC_INTEGER_16 * restrict dest; | |
644cb69f FXC |
55 | index_type rank; |
56 | index_type n; | |
57 | index_type len; | |
58 | index_type delta; | |
59 | index_type dim; | |
60 | ||
61 | /* Make dim zero based to avoid confusion. */ | |
62 | dim = (*pdim) - 1; | |
63 | rank = GFC_DESCRIPTOR_RANK (array) - 1; | |
64 | ||
644cb69f FXC |
65 | len = array->dim[dim].ubound + 1 - array->dim[dim].lbound; |
66 | delta = array->dim[dim].stride; | |
67 | ||
68 | for (n = 0; n < dim; n++) | |
69 | { | |
70 | sstride[n] = array->dim[n].stride; | |
71 | extent[n] = array->dim[n].ubound + 1 - array->dim[n].lbound; | |
80ee04b9 TK |
72 | |
73 | if (extent[n] < 0) | |
74 | extent[n] = 0; | |
644cb69f FXC |
75 | } |
76 | for (n = dim; n < rank; n++) | |
77 | { | |
78 | sstride[n] = array->dim[n + 1].stride; | |
79 | extent[n] = | |
80 | array->dim[n + 1].ubound + 1 - array->dim[n + 1].lbound; | |
80ee04b9 TK |
81 | |
82 | if (extent[n] < 0) | |
83 | extent[n] = 0; | |
644cb69f FXC |
84 | } |
85 | ||
86 | if (retarray->data == NULL) | |
87 | { | |
80ee04b9 TK |
88 | size_t alloc_size; |
89 | ||
644cb69f FXC |
90 | for (n = 0; n < rank; n++) |
91 | { | |
92 | retarray->dim[n].lbound = 0; | |
93 | retarray->dim[n].ubound = extent[n]-1; | |
94 | if (n == 0) | |
95 | retarray->dim[n].stride = 1; | |
96 | else | |
97 | retarray->dim[n].stride = retarray->dim[n-1].stride * extent[n-1]; | |
98 | } | |
99 | ||
644cb69f FXC |
100 | retarray->offset = 0; |
101 | retarray->dtype = (array->dtype & ~GFC_DTYPE_RANK_MASK) | rank; | |
80ee04b9 TK |
102 | |
103 | alloc_size = sizeof (GFC_INTEGER_16) * retarray->dim[rank-1].stride | |
104 | * extent[rank-1]; | |
105 | ||
106 | if (alloc_size == 0) | |
107 | { | |
108 | /* Make sure we have a zero-sized array. */ | |
109 | retarray->dim[0].lbound = 0; | |
110 | retarray->dim[0].ubound = -1; | |
111 | return; | |
112 | } | |
113 | else | |
114 | retarray->data = internal_malloc_size (alloc_size); | |
644cb69f FXC |
115 | } |
116 | else | |
117 | { | |
644cb69f FXC |
118 | if (rank != GFC_DESCRIPTOR_RANK (retarray)) |
119 | runtime_error ("rank of return array incorrect"); | |
120 | } | |
121 | ||
122 | for (n = 0; n < rank; n++) | |
123 | { | |
124 | count[n] = 0; | |
125 | dstride[n] = retarray->dim[n].stride; | |
126 | if (extent[n] <= 0) | |
127 | len = 0; | |
128 | } | |
129 | ||
130 | base = array->data; | |
131 | dest = retarray->data; | |
132 | ||
133 | while (base) | |
134 | { | |
64acfd99 | 135 | const GFC_INTEGER_16 * restrict src; |
644cb69f FXC |
136 | GFC_INTEGER_16 result; |
137 | src = base; | |
138 | { | |
139 | ||
140 | result = 0; | |
141 | if (len <= 0) | |
142 | *dest = 0; | |
143 | else | |
144 | { | |
145 | for (n = 0; n < len; n++, src += delta) | |
146 | { | |
147 | ||
148 | result += *src; | |
149 | } | |
150 | *dest = result; | |
151 | } | |
152 | } | |
153 | /* Advance to the next element. */ | |
154 | count[0]++; | |
155 | base += sstride[0]; | |
156 | dest += dstride[0]; | |
157 | n = 0; | |
158 | while (count[n] == extent[n]) | |
159 | { | |
160 | /* When we get to the end of a dimension, reset it and increment | |
161 | the next dimension. */ | |
162 | count[n] = 0; | |
163 | /* We could precalculate these products, but this is a less | |
5d7adf7a | 164 | frequently used path so probably not worth it. */ |
644cb69f FXC |
165 | base -= sstride[n] * extent[n]; |
166 | dest -= dstride[n] * extent[n]; | |
167 | n++; | |
168 | if (n == rank) | |
169 | { | |
170 | /* Break out of the look. */ | |
171 | base = NULL; | |
172 | break; | |
173 | } | |
174 | else | |
175 | { | |
176 | count[n]++; | |
177 | base += sstride[n]; | |
178 | dest += dstride[n]; | |
179 | } | |
180 | } | |
181 | } | |
182 | } | |
183 | ||
184 | ||
64acfd99 JB |
185 | extern void msum_i16 (gfc_array_i16 * const restrict, |
186 | gfc_array_i16 * const restrict, const index_type * const restrict, | |
187 | gfc_array_l4 * const restrict); | |
644cb69f FXC |
188 | export_proto(msum_i16); |
189 | ||
190 | void | |
64acfd99 JB |
191 | msum_i16 (gfc_array_i16 * const restrict retarray, |
192 | gfc_array_i16 * const restrict array, | |
193 | const index_type * const restrict pdim, | |
194 | gfc_array_l4 * const restrict mask) | |
644cb69f FXC |
195 | { |
196 | index_type count[GFC_MAX_DIMENSIONS]; | |
197 | index_type extent[GFC_MAX_DIMENSIONS]; | |
198 | index_type sstride[GFC_MAX_DIMENSIONS]; | |
199 | index_type dstride[GFC_MAX_DIMENSIONS]; | |
200 | index_type mstride[GFC_MAX_DIMENSIONS]; | |
64acfd99 JB |
201 | GFC_INTEGER_16 * restrict dest; |
202 | const GFC_INTEGER_16 * restrict base; | |
203 | const GFC_LOGICAL_4 * restrict mbase; | |
644cb69f FXC |
204 | int rank; |
205 | int dim; | |
206 | index_type n; | |
207 | index_type len; | |
208 | index_type delta; | |
209 | index_type mdelta; | |
210 | ||
211 | dim = (*pdim) - 1; | |
212 | rank = GFC_DESCRIPTOR_RANK (array) - 1; | |
213 | ||
644cb69f FXC |
214 | len = array->dim[dim].ubound + 1 - array->dim[dim].lbound; |
215 | if (len <= 0) | |
216 | return; | |
217 | delta = array->dim[dim].stride; | |
218 | mdelta = mask->dim[dim].stride; | |
219 | ||
220 | for (n = 0; n < dim; n++) | |
221 | { | |
222 | sstride[n] = array->dim[n].stride; | |
223 | mstride[n] = mask->dim[n].stride; | |
224 | extent[n] = array->dim[n].ubound + 1 - array->dim[n].lbound; | |
80ee04b9 TK |
225 | |
226 | if (extent[n] < 0) | |
227 | extent[n] = 0; | |
228 | ||
644cb69f FXC |
229 | } |
230 | for (n = dim; n < rank; n++) | |
231 | { | |
232 | sstride[n] = array->dim[n + 1].stride; | |
233 | mstride[n] = mask->dim[n + 1].stride; | |
234 | extent[n] = | |
235 | array->dim[n + 1].ubound + 1 - array->dim[n + 1].lbound; | |
80ee04b9 TK |
236 | |
237 | if (extent[n] < 0) | |
238 | extent[n] = 0; | |
644cb69f FXC |
239 | } |
240 | ||
241 | if (retarray->data == NULL) | |
242 | { | |
80ee04b9 TK |
243 | size_t alloc_size; |
244 | ||
644cb69f FXC |
245 | for (n = 0; n < rank; n++) |
246 | { | |
247 | retarray->dim[n].lbound = 0; | |
248 | retarray->dim[n].ubound = extent[n]-1; | |
249 | if (n == 0) | |
250 | retarray->dim[n].stride = 1; | |
251 | else | |
252 | retarray->dim[n].stride = retarray->dim[n-1].stride * extent[n-1]; | |
253 | } | |
254 | ||
80ee04b9 TK |
255 | alloc_size = sizeof (GFC_INTEGER_16) * retarray->dim[rank-1].stride |
256 | * extent[rank-1]; | |
257 | ||
644cb69f FXC |
258 | retarray->offset = 0; |
259 | retarray->dtype = (array->dtype & ~GFC_DTYPE_RANK_MASK) | rank; | |
80ee04b9 TK |
260 | |
261 | if (alloc_size == 0) | |
262 | { | |
263 | /* Make sure we have a zero-sized array. */ | |
264 | retarray->dim[0].lbound = 0; | |
265 | retarray->dim[0].ubound = -1; | |
266 | return; | |
267 | } | |
268 | else | |
269 | retarray->data = internal_malloc_size (alloc_size); | |
270 | ||
644cb69f FXC |
271 | } |
272 | else | |
273 | { | |
644cb69f FXC |
274 | if (rank != GFC_DESCRIPTOR_RANK (retarray)) |
275 | runtime_error ("rank of return array incorrect"); | |
276 | } | |
277 | ||
278 | for (n = 0; n < rank; n++) | |
279 | { | |
280 | count[n] = 0; | |
281 | dstride[n] = retarray->dim[n].stride; | |
282 | if (extent[n] <= 0) | |
283 | return; | |
284 | } | |
285 | ||
286 | dest = retarray->data; | |
287 | base = array->data; | |
288 | mbase = mask->data; | |
289 | ||
290 | if (GFC_DESCRIPTOR_SIZE (mask) != 4) | |
291 | { | |
292 | /* This allows the same loop to be used for all logical types. */ | |
293 | assert (GFC_DESCRIPTOR_SIZE (mask) == 8); | |
294 | for (n = 0; n < rank; n++) | |
295 | mstride[n] <<= 1; | |
296 | mdelta <<= 1; | |
297 | mbase = (GFOR_POINTER_L8_TO_L4 (mbase)); | |
298 | } | |
299 | ||
300 | while (base) | |
301 | { | |
64acfd99 JB |
302 | const GFC_INTEGER_16 * restrict src; |
303 | const GFC_LOGICAL_4 * restrict msrc; | |
644cb69f FXC |
304 | GFC_INTEGER_16 result; |
305 | src = base; | |
306 | msrc = mbase; | |
307 | { | |
308 | ||
309 | result = 0; | |
310 | if (len <= 0) | |
311 | *dest = 0; | |
312 | else | |
313 | { | |
314 | for (n = 0; n < len; n++, src += delta, msrc += mdelta) | |
315 | { | |
316 | ||
317 | if (*msrc) | |
318 | result += *src; | |
319 | } | |
320 | *dest = result; | |
321 | } | |
322 | } | |
323 | /* Advance to the next element. */ | |
324 | count[0]++; | |
325 | base += sstride[0]; | |
326 | mbase += mstride[0]; | |
327 | dest += dstride[0]; | |
328 | n = 0; | |
329 | while (count[n] == extent[n]) | |
330 | { | |
331 | /* When we get to the end of a dimension, reset it and increment | |
332 | the next dimension. */ | |
333 | count[n] = 0; | |
334 | /* We could precalculate these products, but this is a less | |
5d7adf7a | 335 | frequently used path so probably not worth it. */ |
644cb69f FXC |
336 | base -= sstride[n] * extent[n]; |
337 | mbase -= mstride[n] * extent[n]; | |
338 | dest -= dstride[n] * extent[n]; | |
339 | n++; | |
340 | if (n == rank) | |
341 | { | |
342 | /* Break out of the look. */ | |
343 | base = NULL; | |
344 | break; | |
345 | } | |
346 | else | |
347 | { | |
348 | count[n]++; | |
349 | base += sstride[n]; | |
350 | mbase += mstride[n]; | |
351 | dest += dstride[n]; | |
352 | } | |
353 | } | |
354 | } | |
355 | } | |
356 | ||
97a62038 TK |
357 | |
358 | extern void ssum_i16 (gfc_array_i16 * const restrict, | |
359 | gfc_array_i16 * const restrict, const index_type * const restrict, | |
360 | GFC_LOGICAL_4 *); | |
361 | export_proto(ssum_i16); | |
362 | ||
363 | void | |
364 | ssum_i16 (gfc_array_i16 * const restrict retarray, | |
365 | gfc_array_i16 * const restrict array, | |
366 | const index_type * const restrict pdim, | |
367 | GFC_LOGICAL_4 * mask) | |
368 | { | |
369 | index_type rank; | |
370 | index_type n; | |
371 | index_type dstride; | |
372 | GFC_INTEGER_16 *dest; | |
373 | ||
374 | if (*mask) | |
375 | { | |
376 | sum_i16 (retarray, array, pdim); | |
377 | return; | |
378 | } | |
379 | rank = GFC_DESCRIPTOR_RANK (array); | |
380 | if (rank <= 0) | |
381 | runtime_error ("Rank of array needs to be > 0"); | |
382 | ||
383 | if (retarray->data == NULL) | |
384 | { | |
385 | retarray->dim[0].lbound = 0; | |
386 | retarray->dim[0].ubound = rank-1; | |
387 | retarray->dim[0].stride = 1; | |
388 | retarray->dtype = (retarray->dtype & ~GFC_DTYPE_RANK_MASK) | 1; | |
389 | retarray->offset = 0; | |
390 | retarray->data = internal_malloc_size (sizeof (GFC_INTEGER_16) * rank); | |
391 | } | |
392 | else | |
393 | { | |
394 | if (GFC_DESCRIPTOR_RANK (retarray) != 1) | |
395 | runtime_error ("rank of return array does not equal 1"); | |
396 | ||
397 | if (retarray->dim[0].ubound + 1 - retarray->dim[0].lbound != rank) | |
398 | runtime_error ("dimension of return array incorrect"); | |
97a62038 TK |
399 | } |
400 | ||
401 | dstride = retarray->dim[0].stride; | |
402 | dest = retarray->data; | |
403 | ||
404 | for (n = 0; n < rank; n++) | |
405 | dest[n * dstride] = 0 ; | |
406 | } | |
407 | ||
644cb69f | 408 | #endif |