1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
|
/*
* Copyright (c) 1997-1999, 2003 Massachusetts Institute of Technology
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
*
*/
/*
* planner.c -- find the optimal plan
*/
/* $Id: planner.c,v 1.1 2008/10/17 06:13:18 scuri Exp $ */
#include "fftw-int.h"
#include <stdlib.h>
#include <stdio.h>
extern fftw_generic_codelet fftw_twiddle_generic;
extern fftw_generic_codelet fftwi_twiddle_generic;
extern fftw_codelet_desc *fftw_config[];
fftw_plan_hook_ptr fftw_plan_hook = (fftw_plan_hook_ptr) NULL;
static void init_test_array(fftw_complex *arr, int stride, int n)
{
int j;
for (j = 0; j < n; ++j) {
c_re(arr[stride * j]) = 0.0;
c_im(arr[stride * j]) = 0.0;
}
}
/*
* The timer keeps doubling the number of iterations
* until the program runs for more than FFTW_TIME_MIN
*/
static double fftw_measure_runtime(fftw_plan plan,
fftw_complex *in, int istride,
fftw_complex *out, int ostride)
{
fftw_time begin, end, start;
double t, tmax, tmin;
int i, iter;
int n;
int repeat;
int howmany = plan->vector_size;
n = plan->n;
iter = 1;
for (;;) {
tmin = 1.0E10;
tmax = -1.0E10;
init_test_array(in, istride, n * howmany);
start = fftw_get_time();
/* repeat the measurement FFTW_TIME_REPEAT times */
for (repeat = 0; repeat < FFTW_TIME_REPEAT; ++repeat) {
begin = fftw_get_time();
for (i = 0; i < iter; ++i) {
fftw(plan, howmany, in, istride, istride,
out, ostride, ostride);
}
end = fftw_get_time();
t = fftw_time_to_sec(fftw_time_diff(end, begin));
if (t < tmin)
tmin = t;
if (t > tmax)
tmax = t;
/* do not run for too long */
t = fftw_time_to_sec(fftw_time_diff(end, start));
if (t > FFTW_TIME_LIMIT)
break;
}
if (tmin >= FFTW_TIME_MIN)
break;
iter *= 2;
}
tmin /= (double) iter;
tmax /= (double) iter;
return tmin;
}
/* auxiliary functions */
static void compute_cost(fftw_plan plan,
fftw_complex *in, int istride,
fftw_complex *out, int ostride)
{
if (plan->flags & FFTW_MEASURE)
plan->cost = fftw_measure_runtime(plan, in, istride, out, ostride);
else {
double c;
c = plan->n * fftw_estimate_node(plan->root) * plan->vector_size;
plan->cost = c;
}
}
static void run_plan_hooks(fftw_plan p)
{
if (fftw_plan_hook && p) {
fftw_complete_twiddle(p->root, p->n);
fftw_plan_hook(p);
}
}
/* macrology */
#define FOR_ALL_CODELETS(p) \
fftw_codelet_desc **__q, *p; \
for (__q = &fftw_config[0]; (p = (*__q)); ++__q)
/******************************************
* Recursive planner *
******************************************/
static fftw_plan planner(fftw_plan *table, int n, fftw_direction dir,
int flags, int vector_size,
fftw_complex *, int, fftw_complex *, int);
/*
* the planner consists of two parts: one that tries to
* use accumulated wisdom, and one that does not.
* A small driver invokes both parts in sequence
*/
/* planner with wisdom: look up the codelet suggested by the wisdom */
static fftw_plan planner_wisdom(fftw_plan *table, int n,
fftw_direction dir, int flags,
int vector_size,
fftw_complex *in, int istride,
fftw_complex *out, int ostride)
{
fftw_plan best = (fftw_plan) 0;
fftw_plan_node *node;
int have_wisdom;
enum fftw_node_type wisdom_type;
int wisdom_signature;
fftw_recurse_kind wisdom_recurse_kind;
/* see if we remember any wisdom for this case */
have_wisdom = fftw_wisdom_lookup(n, flags, dir, FFTW_WISDOM,
istride, ostride,
&wisdom_type, &wisdom_signature,
&wisdom_recurse_kind, 0);
if (!have_wisdom)
return best;
if (wisdom_type == FFTW_NOTW) {
FOR_ALL_CODELETS(p) {
if (p->dir == dir && p->type == wisdom_type) {
/* see if wisdom applies */
if (wisdom_signature == p->signature &&
p->size == n) {
node = fftw_make_node_notw(n, p);
best = fftw_make_plan(n, dir, node, flags,
p->type, p->signature,
FFTW_NORMAL_RECURSE,
vector_size);
fftw_use_plan(best);
run_plan_hooks(best);
return best;
}
}
}
}
if (wisdom_type == FFTW_TWIDDLE) {
FOR_ALL_CODELETS(p) {
if (p->dir == dir && p->type == wisdom_type) {
/* see if wisdom applies */
if (wisdom_signature == p->signature &&
p->size > 1 &&
(n % p->size) == 0) {
fftw_plan r = planner(table, n / p->size, dir,
flags | FFTW_NO_VECTOR_RECURSE,
wisdom_recurse_kind ==
FFTW_VECTOR_RECURSE ?
p->size : vector_size,
in, istride, out, ostride);
node = fftw_make_node_twiddle(n, p,
r->root, flags);
best = fftw_make_plan(n, dir, node, flags,
p->type, p->signature,
wisdom_recurse_kind,
vector_size);
fftw_use_plan(best);
run_plan_hooks(best);
fftw_destroy_plan_internal(r);
return best;
}
}
}
}
/*
* BUG (or: TODO) Can we have generic wisdom? This is probably
* an academic question
*/
return best;
}
/*
* planner with no wisdom: try all combinations and pick
* the best
*/
static fftw_plan planner_normal(fftw_plan *table, int n, fftw_direction dir,
int flags, int vector_size,
fftw_complex *in, int istride,
fftw_complex *out, int ostride)
{
fftw_plan best = (fftw_plan) 0;
fftw_plan newplan;
fftw_plan_node *node;
/* see if we have any codelet that solves the problem */
{
FOR_ALL_CODELETS(p) {
if (p->dir == dir && p->type == FFTW_NOTW) {
if (p->size == n) {
node = fftw_make_node_notw(n, p);
newplan = fftw_make_plan(n, dir, node, flags,
p->type, p->signature,
FFTW_NORMAL_RECURSE,
vector_size);
fftw_use_plan(newplan);
compute_cost(newplan, in, istride, out, ostride);
run_plan_hooks(newplan);
best = fftw_pick_better(newplan, best);
}
}
}
}
/* Then, try all available twiddle codelets */
{
FOR_ALL_CODELETS(p) {
if (p->dir == dir && p->type == FFTW_TWIDDLE) {
if ((n % p->size) == 0 &&
p->size > 1 &&
(!best || n != p->size)) {
fftw_plan r = planner(table, n / p->size, dir,
flags | FFTW_NO_VECTOR_RECURSE,
vector_size,
in, istride, out, ostride);
node = fftw_make_node_twiddle(n, p,
r->root, flags);
newplan = fftw_make_plan(n, dir, node, flags,
p->type, p->signature,
FFTW_NORMAL_RECURSE,
vector_size);
fftw_use_plan(newplan);
fftw_destroy_plan_internal(r);
compute_cost(newplan, in, istride, out, ostride);
run_plan_hooks(newplan);
best = fftw_pick_better(newplan, best);
}
}
}
}
/* try vector recursion unless prohibited by the flags: */
if (! (flags & FFTW_NO_VECTOR_RECURSE)) {
FOR_ALL_CODELETS(p) {
if (p->dir == dir && p->type == FFTW_TWIDDLE) {
if ((n % p->size) == 0 &&
p->size > 1 &&
(!best || n != p->size)) {
fftw_plan r = planner(table, n / p->size, dir,
flags | FFTW_NO_VECTOR_RECURSE,
p->size,
in, istride, out, ostride);
node = fftw_make_node_twiddle(n, p,
r->root, flags);
newplan = fftw_make_plan(n, dir, node, flags,
p->type, p->signature,
FFTW_VECTOR_RECURSE,
vector_size);
fftw_use_plan(newplan);
fftw_destroy_plan_internal(r);
compute_cost(newplan, in, istride, out, ostride);
run_plan_hooks(newplan);
best = fftw_pick_better(newplan, best);
}
}
}
}
/*
* resort to generic or rader codelets for unknown factors
*/
{
fftw_generic_codelet *codelet = (dir == FFTW_FORWARD ?
fftw_twiddle_generic :
fftwi_twiddle_generic);
int size, prev_size = 0, remaining_factors = n;
fftw_plan r;
while (remaining_factors > 1) {
size = fftw_factor(remaining_factors);
remaining_factors /= size;
/* don't try the same factor more than once */
if (size == prev_size)
continue;
prev_size = size;
/* Look for codelets corresponding to this factor. */
{
FOR_ALL_CODELETS(p) {
if (p->dir == dir && p->type == FFTW_TWIDDLE
&& p->size == size) {
size = 0;
break;
}
}
}
/*
* only try a generic/rader codelet if there were no
* twiddle codelets for this factor
*/
if (!size)
continue;
r = planner(table, n / size, dir,
flags | FFTW_NO_VECTOR_RECURSE,
vector_size,
in, istride, out, ostride);
/* Try Rader codelet: */
node = fftw_make_node_rader(n, size, dir, r->root, flags);
newplan = fftw_make_plan(n, dir, node, flags, FFTW_RADER, 0,
FFTW_NORMAL_RECURSE, vector_size);
fftw_use_plan(newplan);
compute_cost(newplan, in, istride, out, ostride);
run_plan_hooks(newplan);
best = fftw_pick_better(newplan, best);
if (size < 100) { /*
* only try generic for small
* sizes
*/
/* Try generic codelet: */
node = fftw_make_node_generic(n, size, codelet,
r->root, flags);
newplan = fftw_make_plan(n, dir, node, flags,
FFTW_GENERIC, 0,
FFTW_NORMAL_RECURSE, vector_size);
fftw_use_plan(newplan);
compute_cost(newplan, in, istride, out, ostride);
run_plan_hooks(newplan);
best = fftw_pick_better(newplan, best);
}
fftw_destroy_plan_internal(r);
}
}
if (!best)
fftw_die("bug in planner\n");
return best;
}
static fftw_plan planner(fftw_plan *table, int n, fftw_direction dir,
int flags, int vector_size,
fftw_complex *in, int istride,
fftw_complex *out, int ostride)
{
fftw_plan best = (fftw_plan) 0;
if (vector_size > 1)
flags |= FFTW_NO_VECTOR_RECURSE;
/* see if plan has already been computed */
best = fftw_lookup(table, n, flags, vector_size);
if (best) {
fftw_use_plan(best);
return best;
}
/* try a wise plan */
best = planner_wisdom(table, n, dir, flags, vector_size,
in, istride, out, ostride);
if (!best) {
/* No wisdom. Plan normally. */
best = planner_normal(table, n, dir, flags,
vector_size,
in, istride, out, ostride);
}
if (best) {
fftw_insert(table, best);
/* remember the wisdom */
fftw_wisdom_add(n, flags, dir, FFTW_WISDOM, istride, ostride,
best->wisdom_type,
best->wisdom_signature,
best->recurse_kind);
}
return best;
}
fftw_plan fftw_create_plan_specific(int n, fftw_direction dir, int flags,
fftw_complex *in, int istride,
fftw_complex *out, int ostride)
{
fftw_plan table;
fftw_plan p1;
/* validate parameters */
if (n <= 0)
return (fftw_plan) 0;
#ifndef FFTW_ENABLE_VECTOR_RECURSE
/* TEMPORARY: disable vector recursion until it is more tested. */
flags |= FFTW_NO_VECTOR_RECURSE;
#endif
if ((dir != FFTW_FORWARD) && (dir != FFTW_BACKWARD))
return (fftw_plan) 0;
fftw_make_empty_table(&table);
p1 = planner(&table, n, dir, flags, 1,
in, istride, out, ostride);
fftw_destroy_table(&table);
if (p1)
fftw_complete_twiddle(p1->root, n);
return p1;
}
fftw_plan fftw_create_plan(int n, fftw_direction dir, int flags)
{
fftw_complex *tmp_in, *tmp_out;
fftw_plan p;
if (flags & FFTW_MEASURE) {
tmp_in = (fftw_complex *) fftw_malloc(2 * n * sizeof(fftw_complex));
if (!tmp_in)
return 0;
tmp_out = tmp_in + n;
p = fftw_create_plan_specific(n, dir, flags,
tmp_in, 1, tmp_out, 1);
fftw_free(tmp_in);
} else
p = fftw_create_plan_specific(n, dir, flags,
(fftw_complex *) 0, 1, (fftw_complex *) 0, 1);
return p;
}
void fftw_destroy_plan(fftw_plan plan)
{
fftw_destroy_plan_internal(plan);
}
|