-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathChangeLog
27526 lines (21589 loc) · 769 KB
/
ChangeLog
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
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
Sat Nov 24 22:37:54 EST 2012 [email protected]
* fixed deadlock bug caused by bogosity flag getting out of synch between processes; thanks to Michael Pippig for the bug report
M ./kernel/ifftw.h +1
M ./kernel/planner.c -3 +6
M ./mpi/api.c +12
Wed Nov 21 18:34:29 EST 2012 [email protected]
* Updated NEWS
M ./NEWS -2 +7
Wed Nov 21 18:33:15 EST 2012 [email protected]
* use 2x2 AVX transposition instead of individual stores.
This seems to improve single-precision AVX on Sandy Bridge machines.
M ./simd-support/simd-avx.h -2 +14
Tue Nov 20 12:18:00 EST 2012 [email protected]
* revert part of Taylor patch to acx_mpi.m4: do not link -lmpi if mpicc works without libraries, as -lmpi may be some completely different MPI implementation
M ./m4/acx_mpi.m4 -3 +3
Tue Nov 20 11:44:57 EST 2012 [email protected]
* fix deadlock bug (thanks to Michael Pippig for the bug report and patch, and to Graham Dennis for the bug report) in which some processes called MPI_Alltoall and some called MPI_Alltoallv
M ./mpi/transpose-alltoall.c -3 +2
Mon Oct 29 15:20:01 EDT 2012 [email protected]
* fix texinfo quirk
M ./doc/tutorial.texi -2 +2
Mon Oct 29 09:16:43 EDT 2012 [email protected]
* clarify that padding only applies to in-place transforms
M ./doc/tutorial.texi -5 +10
Sun Oct 28 18:42:48 EDT 2012 [email protected]
* make the index-computation logic less paranoid
The problem is that for each K and for each expression of the form P[I
+ STRIDE * K] in a loop, most compilers will try to lift an induction
variable PK := &P[I + STRIDE * K]. In large codelets we have many
such values of K. For example, a codelet of size 32 with 4 input
pointers will generate O(128) induction variables, which will likely
overflow the register set, which is likely worse than doing the index
computation in the first place.
In the past we (wisely and correctly) assumed that compilers will do
the wrong thing, and consequently we disabled the induction-variable
"optimization" altogether by setting STRIDE ^= ZERO, where ZERO is a
value guaranteed to be 0. Since the compiler does not know that
ZERO=0, it cannot perform its "optimization" and it is forced to
behave sensibly.
With this patch, FFTW is a little bit less paranoid. FFTW now
disables the induction-variable optimization" only when we estimate
that the codelet uses more than ESTIMATED_AVAILABLE_INDEX_REGISTERS
induction variables.
Currently we set ESTIMATED_AVAILABLE_INDEX_REGISTERS=16. 16 registers ought
to be enough for anybody (or so the amd64 and ARM ISA's seem to imply).
M ./genfft/gen_hc2c.ml -1 +1
M ./genfft/gen_hc2cdft.ml -1 +1
M ./genfft/gen_hc2cdft_c.ml -1 +1
M ./genfft/gen_hc2hc.ml -1 +1
M ./genfft/gen_notw.ml -2 +2
M ./genfft/gen_notw_c.ml -2 +2
M ./genfft/gen_r2cb.ml -3 +3
M ./genfft/gen_r2cf.ml -3 +3
M ./genfft/gen_r2r.ml -2 +2
M ./genfft/gen_twiddle.ml -1 +1
M ./genfft/gen_twiddle_c.ml -1 +1
M ./genfft/gen_twidsq.ml -2 +2
M ./genfft/gen_twidsq_c.ml -2 +2
M ./genfft/genutil.ml -1 +2
M ./kernel/ifftw.h -3 +20
Sun Oct 28 18:33:24 EDT 2012 [email protected]
* silence warnings
M ./kernel/buffered.c +1
M ./rdft/rank0.c +1
Sat Oct 27 09:58:49 EDT 2012 [email protected]
* bump version to 3.3.3
M ./NEWS +7
M ./configure.ac -1 +1
Sat Oct 27 09:55:15 EDT 2012 [email protected]
* evaluate plans for >1ms when using gettimeofday()
The previous limit 10ms was too paranoid, and it made life difficult
on machines without an "official" cycle counter, such as ARM.
M ./kernel/timer.c -1 +1
Sat Oct 27 09:46:04 EDT 2012 [email protected]
* use 4-way NEON SIMD instead of 2-way
Kai-Uwe Bloem tried to warn me a year ago that 128-bit NEON was better
than 64-bit NEON even on machines with a 64-bit pipe, but I foolishly
did not listen. Now that 128-bit NEON pipes are starting to appear on
the market it is definitely time to switch.
M ./simd-support/simd-neon.h -55 +100
Wed Sep 26 14:21:12 EDT 2012 [email protected]
* Note that fftw-3.3 includes MPI support
M ./doc/intro.texi -5 +4
Wed Jul 18 11:25:40 EDT 2012 [email protected]
* remove obsolete unused function
M ./dft/bluestein.c -14
Fri Jun 29 15:57:14 EDT 2012 [email protected]
* whoops, call omp_get_max_threads; thanks to Hanno Rein for the bug report
M ./doc/threads.texi -1 +1
Sat Apr 28 10:55:09 EDT 2012 [email protected]
* Fix libfftw3/libfftw3_threads chicken-egg problem
On most systems we want to build libfftw3 first, so that
libfftw3_threads can depend upon libfftw3. When producing a single
combined-thread library (e.g. on Windows) we want the opposite,
so that libfftw3 can include libfftw3_threads.
M ./Makefile.am -1 +18
M ./threads/Makefile.am +5
Sat Apr 28 10:11:28 EDT 2012 [email protected]
* updated NEWS for 3.3.2
M ./NEWS +15
Thu Apr 26 19:36:11 EDT 2012 [email protected]
* change revision to 3.3.2
M ./configure.ac -2 +2
Thu Apr 26 19:31:02 EDT 2012 [email protected]
* Remove old aligned_main() hack.
On i386, in our benchmark program we used to manually aligned the
stack to 16-byte boundary via asm trickery. This was a good idea in
1999 (and it was actually necessary to make things work) but the hack
is now obsolete and it seems to break gcc-4.7. So the hack is now
gone.
M ./libbench2/Makefile.am -6 +6
R ./libbench2/aligned-main.c
M ./libbench2/main.c -1 +1
Thu Mar 29 16:26:16 EDT 2012 [email protected]
* Bugfix: a couple of uninitialized values in the benchmark program
M ./libbench2/problem.c -1 +1
Tue Mar 20 19:03:47 EDT 2012 [email protected]
* make libfftw{threads,mpi} depend upon libfftw for libtool purposes
Thanks Julian Taylor for the patch
M ./Makefile.am -1 +1
M ./m4/acx_mpi.m4 -3 +3
M ./mpi/Makefile.am +1
M ./threads/Makefile.am +1
Tue Mar 6 04:44:00 EST 2012 [email protected]
* formatting tweak
M ./doc/reference.texi -1 +1
Mon Mar 5 21:05:27 EST 2012 [email protected]
* destroying => overwriting
M ./doc/tutorial.texi -1 +1
Fri Mar 2 10:31:20 EST 2012 [email protected]
* note that WISDOM_ONLY is a documented flag
M ./api/fftw3.h -1 +1
Fri Mar 2 10:27:08 EST 2012 [email protected]
* check for icc pretending to be gcc before including quad-precision decls; thanks to Michael Anselmi for the bug report
M ./api/fftw3.h +1
Fri Mar 2 10:23:19 EST 2012 [email protected]
* foo_CFLAGS needs to manually include AM_CFLAGS; thanks to Henry Gomersall for the Windows bug report
M ./threads/Makefile.am -2 +2
Sat Feb 25 15:21:39 EST 2012 [email protected]
* update for latest mingw
M ./support/BUILD-MINGW32.sh -3 +3
M ./support/BUILD-MINGW64.sh -2 +7
Sat Feb 25 15:21:17 EST 2012 [email protected]
tagged fftw-3.3.1
Mon Feb 20 23:06:13 EST 2012 [email protected]
* added Fortran NEWS
M ./NEWS +6
Mon Feb 20 23:00:13 EST 2012 [email protected]
* move non-portable extended/quad precision F03 interfaces into separate .f03 files (while keeping double/single in fftw3.f03 for minimal ABI breakage)
M ./api/Makefile.am -3 +10
M ./api/f03api.sh -2 +4
M ./api/genf03.pl -5 +5
M ./doc/modern-fortran.texi -2 +39
M ./doc/mpi.texi -2 +2
M ./mpi/Makefile.am -5 +11
M ./mpi/f03api.sh -2 +4
Mon Feb 20 11:21:57 EST 2012 [email protected]
* rm mpi/fftw3-mpi.f03 at make clean time; thanks Tyler Luchko for the bug report.
M ./mpi/Makefile.am +1
Mon Feb 20 11:18:24 EST 2012 [email protected]
* Disable a Visual Studion warning that was obnoxious enough for Sebastian Schuberth to send us a patch.
M ./simd-support/simd-sse2.h +14
Mon Feb 20 11:18:06 EST 2012 [email protected]
* Change version to 3.3.1
M ./configure.ac -1 +1
Mon Feb 20 11:03:15 EST 2012 [email protected]
* Integrated Visual Studio AVX patches by Carsten Steger
M ./NEWS +3
M ./simd-support/amd64-cpuid.h -10 +40
M ./simd-support/simd-avx.h +6
M ./simd-support/simd-sse2.h +8
M ./simd-support/x86-cpuid.h +7
Wed Nov 9 10:13:32 EST 2011 [email protected]
* typo
M ./doc/modern-fortran.texi -2 +2
Tue Nov 8 22:45:09 EST 2011 [email protected]
* add missing F77 set_timelimit function; thanks to Martin Diehl for the bug repory
M ./api/f77funcs.h +5
Sun Sep 25 10:54:56 EDT 2011 [email protected]
* note requirement of /machine:x64 in windows x64 README
M ./support/BUILD-MINGW64.sh +7
Sun Sep 18 09:28:20 EDT 2011 [email protected]
* AVX detection for MSVC
M ./simd-support/amd64-cpuid.h -2 +10
Tue Sep 13 14:58:29 EDT 2011 [email protected]
* compile with C89
M ./dft/bluestein.c -1 +1
M ./dft/generic.c -1 +1
M ./rdft/dht-rader.c -1 +1
Sat Sep 3 16:25:50 EDT 2011 [email protected]
* use the same search pruning heuristics for threaded plans as for nonthreaded plans
M ./threads/ct.c -1 +3
M ./threads/hc2hc.c -1 +3
Sat Sep 3 16:12:11 EDT 2011 [email protected]
* shorten ESTIMATE planning time for certain weird sizes
FFTW includes a collection of "solvers" that apply to a subset of
"problems". Assume for simplicity that a "problem" is a single 1D
complex transform of size N, even though real "problems" are much more
general than that. FFTW includes three "prime" solvers called
"generic", "bluestein", and "rader", which implement different
algorithms for prime sizes.
Now, for a "problem" of size 13 (say) FFTW also includes special code
that handles that size at high speed. It would be a waste of time to
measure the execution time of the prime solvers, since we know that
the special code is way faster. However, FFTW is modular and one may
or may not include the special code for size 13, in which case we must
resort to one of the "prime" solvers. To address this issue, the
"prime" solvers (and others) are proclaimed to be SLOW". When
planning, FFTW first tries to produce a plan ignoring all the SLOW
solvers, and if this fails FFTW tries again allowing SLOW solvers.
This heuristic works ok unless the sizes are too large. For example
for 1044000=2*2*2*2*2*3*3*5*5*5*29 FFTW explores a huge search tree of
all zillion factorizations of 1044000/29, failing every time because
29 is SLOW; then it finally allows SLOW solvers and finds a solution
immediately.
This patch proclaims solvers to be SLOW only for small values of N.
For example, the "generic" solver implements an O(n^2) DFT algorithm;
we say that it is SLOW only for N<=16.
The side effects of this choice are as follows. If one modifies FFTW to
include a fast solver of size 17, then planning for N=17*K will be
slower than today, because FFTW till try both the fast solver and the
generic solver (which is SLOW today and therefore not tried, but is no
longer SLOW after the patch). If one removes a fast solver, of size say
13, then he may still fall into the current exponential-search behavior
for "problems" of size 13*HIGHLY_FACTORIZABLE_N.
If somebody had compleined about transforms of size 1044000 ten years
ago, "don't do that" would have been an acceptable answer. I guess the
bar is higher today, so I am going to include this patch in our 3.3.1
release despite their side-effects for people who want to modify FFTW.
M ./dft/bluestein.c -5 +14
M ./dft/generic.c -15 +6
M ./dft/rader.c -7 +7
M ./kernel/ifftw.h +8
M ./kernel/primes.c +6
M ./rdft/dht-rader.c -7 +8
M ./rdft/generic.c -14 +4
Sat Aug 27 13:55:24 EDT 2011 [email protected]
* Fix typo fftw_execute_dft_r2r => fftw_execute_r2r
Thanks KIU Shueng Chuan for the bug report.
M ./doc/reference.texi -1 +1
Fri Aug 26 06:13:55 EDT 2011 [email protected]
* In Rader's algorithm, compute the generator lazily.
The planner was spending a lot of time computing generators for
plans that were immediately discarded. Now we compute generators
only when absolutely needed.
M ./NEWS +5
M ./dft/rader.c -3 +4
M ./rdft/dht-rader.c -3 +4
Sun Aug 21 16:27:31 EDT 2011 [email protected]
tagged fftw-3.3.1-beta1
Sun Aug 21 16:16:38 EDT 2011 [email protected]
* Release notes for 3.3.1-beta1
M ./NEWS +4
M ./configure.ac -1 +1
M ./doc/install.texi -8 +18
M ./doc/other.texi -2 +3
Fri Aug 19 19:59:17 EDT 2011 [email protected]
* make fftw_mpi_block routine 10x faster, since it is being called zillions of times (thanks to Tom Vacek for the profiling)
M ./mpi/block.c -6 +2
Thu Aug 18 14:19:36 EDT 2011 [email protected]
* Implement autodetection of NEON extensions
M ./simd-support/neon.c -4 +51
Sun Aug 14 14:12:29 EDT 2011 [email protected]
* Update the FSF address.
The FSF moved downtown.
M ./COPYRIGHT -1 +1
M ./api/api.h -1 +1
M ./api/apiplan.c -1 +1
M ./api/configure.c -1 +1
M ./api/execute-dft-c2r.c -1 +1
M ./api/execute-dft-r2c.c -1 +1
M ./api/execute-dft.c -1 +1
M ./api/execute-r2r.c -1 +1
M ./api/execute-split-dft-c2r.c -1 +1
M ./api/execute-split-dft-r2c.c -1 +1
M ./api/execute-split-dft.c -1 +1
M ./api/execute.c -1 +1
M ./api/export-wisdom-to-file.c -1 +1
M ./api/export-wisdom-to-string.c -1 +1
M ./api/export-wisdom.c -1 +1
M ./api/f77api.c -1 +1
M ./api/f77funcs.h -1 +1
M ./api/flops.c -1 +1
M ./api/forget-wisdom.c -1 +1
M ./api/import-system-wisdom.c -1 +1
M ./api/import-wisdom-from-file.c -1 +1
M ./api/import-wisdom-from-string.c -1 +1
M ./api/import-wisdom.c -1 +1
M ./api/malloc.c -1 +1
M ./api/map-r2r-kind.c -1 +1
M ./api/mapflags.c -1 +1
M ./api/mkprinter-file.c -1 +1
M ./api/mktensor-iodims.h -1 +1
M ./api/mktensor-rowmajor.c -1 +1
M ./api/plan-dft-1d.c -1 +1
M ./api/plan-dft-2d.c -1 +1
M ./api/plan-dft-3d.c -1 +1
M ./api/plan-dft-c2r-1d.c -1 +1
M ./api/plan-dft-c2r-2d.c -1 +1
M ./api/plan-dft-c2r-3d.c -1 +1
M ./api/plan-dft-c2r.c -1 +1
M ./api/plan-dft-r2c-1d.c -1 +1
M ./api/plan-dft-r2c-2d.c -1 +1
M ./api/plan-dft-r2c-3d.c -1 +1
M ./api/plan-dft-r2c.c -1 +1
M ./api/plan-dft.c -1 +1
M ./api/plan-guru-dft-c2r.h -1 +1
M ./api/plan-guru-dft-r2c.h -1 +1
M ./api/plan-guru-dft.h -1 +1
M ./api/plan-guru-r2r.h -1 +1
M ./api/plan-guru-split-dft-c2r.h -1 +1
M ./api/plan-guru-split-dft-r2c.h -1 +1
M ./api/plan-guru-split-dft.h -1 +1
M ./api/plan-many-dft-c2r.c -1 +1
M ./api/plan-many-dft-r2c.c -1 +1
M ./api/plan-many-dft.c -1 +1
M ./api/plan-many-r2r.c -1 +1
M ./api/plan-r2r-1d.c -1 +1
M ./api/plan-r2r-2d.c -1 +1
M ./api/plan-r2r-3d.c -1 +1
M ./api/plan-r2r.c -1 +1
M ./api/print-plan.c -1 +1
M ./api/rdft2-pad.c -1 +1
M ./api/the-planner.c -1 +1
M ./api/version.c -1 +1
M ./api/x77.h -1 +1
M ./commercialize.sh -2 +2
M ./dft/bluestein.c -1 +1
M ./dft/buffered.c -1 +1
M ./dft/codelet-dft.h -1 +1
M ./dft/conf.c -1 +1
M ./dft/ct.c -1 +1
M ./dft/ct.h -1 +1
M ./dft/dft.h -1 +1
M ./dft/dftw-direct.c -1 +1
M ./dft/dftw-directsq.c -1 +1
M ./dft/dftw-generic.c -1 +1
M ./dft/dftw-genericbuf.c -1 +1
M ./dft/direct.c -1 +1
M ./dft/generic.c -1 +1
M ./dft/indirect-transpose.c -1 +1
M ./dft/indirect.c -1 +1
M ./dft/kdft-dif.c -1 +1
M ./dft/kdft-difsq.c -1 +1
M ./dft/kdft-dit.c -1 +1
M ./dft/kdft.c -1 +1
M ./dft/nop.c -1 +1
M ./dft/plan.c -1 +1
M ./dft/problem.c -1 +1
M ./dft/rader.c -1 +1
M ./dft/rank-geq2.c -1 +1
M ./dft/scalar/n.c -1 +1
M ./dft/scalar/n.h -1 +1
M ./dft/scalar/t.c -1 +1
M ./dft/scalar/t.h -1 +1
M ./dft/simd/common/genus.c -1 +1
M ./dft/simd/n1b.h -1 +1
M ./dft/simd/n1f.h -1 +1
M ./dft/simd/n2b.h -1 +1
M ./dft/simd/n2f.h -1 +1
M ./dft/simd/n2s.h -1 +1
M ./dft/simd/q1b.h -1 +1
M ./dft/simd/q1f.h -1 +1
M ./dft/simd/t1b.h -1 +1
M ./dft/simd/t1bu.h -1 +1
M ./dft/simd/t1f.h -1 +1
M ./dft/simd/t1fu.h -1 +1
M ./dft/simd/t2b.h -1 +1
M ./dft/simd/t2f.h -1 +1
M ./dft/simd/t3b.h -1 +1
M ./dft/simd/t3f.h -1 +1
M ./dft/simd/ts.h -1 +1
M ./dft/solve.c -1 +1
M ./dft/vrank-geq1.c -1 +1
M ./dft/zero.c -1 +1
M ./doc/f77_wisdom.f -1 +1
M ./doc/license.texi -3 +3
M ./genfft/algsimp.ml -1 +1
M ./genfft/algsimp.mli -1 +1
M ./genfft/annotate.ml -1 +1
M ./genfft/annotate.mli -1 +1
M ./genfft/assoctable.ml -1 +1
M ./genfft/assoctable.mli -1 +1
M ./genfft/c.ml -1 +1
M ./genfft/c.mli -1 +1
M ./genfft/complex.ml -1 +1
M ./genfft/complex.mli -1 +1
M ./genfft/conv.ml -1 +1
M ./genfft/conv.mli -1 +1
M ./genfft/dag.ml -1 +1
M ./genfft/dag.mli -1 +1
M ./genfft/expr.ml -1 +1
M ./genfft/expr.mli -1 +1
M ./genfft/fft.ml -1 +1
M ./genfft/fft.mli -1 +1
M ./genfft/gen_hc2c.ml -1 +1
M ./genfft/gen_hc2cdft.ml -1 +1
M ./genfft/gen_hc2cdft_c.ml -1 +1
M ./genfft/gen_hc2hc.ml -1 +1
M ./genfft/gen_mdct.ml -1 +1
M ./genfft/gen_notw.ml -1 +1
M ./genfft/gen_notw_c.ml -1 +1
M ./genfft/gen_r2cb.ml -1 +1
M ./genfft/gen_r2cf.ml -1 +1
M ./genfft/gen_r2r.ml -1 +1
M ./genfft/gen_twiddle.ml -1 +1
M ./genfft/gen_twiddle_c.ml -1 +1
M ./genfft/gen_twidsq.ml -1 +1
M ./genfft/gen_twidsq_c.ml -1 +1
M ./genfft/genutil.ml -1 +1
M ./genfft/littlesimp.ml -1 +1
M ./genfft/littlesimp.mli -1 +1
M ./genfft/magic.ml -1 +1
M ./genfft/monads.ml -1 +1
M ./genfft/number.ml -1 +1
M ./genfft/number.mli -1 +1
M ./genfft/oracle.ml -1 +1
M ./genfft/oracle.mli -1 +1
M ./genfft/schedule.ml -1 +1
M ./genfft/schedule.mli -1 +1
M ./genfft/simd.ml -1 +1
M ./genfft/simd.mli -1 +1
M ./genfft/simdmagic.ml -1 +1
M ./genfft/to_alist.ml -1 +1
M ./genfft/to_alist.mli -1 +1
M ./genfft/trig.ml -1 +1
M ./genfft/trig.mli -1 +1
M ./genfft/twiddle.ml -1 +1
M ./genfft/twiddle.mli -1 +1
M ./genfft/unique.ml -1 +1
M ./genfft/unique.mli -1 +1
M ./genfft/util.ml -1 +1
M ./genfft/util.mli -1 +1
M ./genfft/variable.ml -1 +1
M ./genfft/variable.mli -1 +1
M ./kernel/align.c -1 +1
M ./kernel/alloc.c -1 +1
M ./kernel/assert.c -1 +1
M ./kernel/awake.c -1 +1
M ./kernel/buffered.c -1 +1
M ./kernel/cpy1d.c -1 +1
M ./kernel/cpy2d-pair.c -1 +1
M ./kernel/cpy2d.c -1 +1
M ./kernel/ct.c -1 +1
M ./kernel/debug.c -1 +1
M ./kernel/extract-reim.c -1 +1
M ./kernel/hash.c -1 +1
M ./kernel/iabs.c -1 +1
M ./kernel/ifftw.h -1 +1
M ./kernel/kalloc.c -1 +1
M ./kernel/md5-1.c -1 +1
M ./kernel/md5.c -1 +1
M ./kernel/minmax.c -1 +1
M ./kernel/ops.c -1 +1
M ./kernel/pickdim.c -1 +1
M ./kernel/plan.c -1 +1
M ./kernel/planner.c -1 +1
M ./kernel/primes.c -1 +1
M ./kernel/print.c -1 +1
M ./kernel/problem.c -1 +1
M ./kernel/rader.c -1 +1
M ./kernel/scan.c -1 +1
M ./kernel/solver.c -1 +1
M ./kernel/solvtab.c -1 +1
M ./kernel/stride.c -1 +1
M ./kernel/tensor.c -1 +1
M ./kernel/tensor1.c -1 +1
M ./kernel/tensor2.c -1 +1
M ./kernel/tensor3.c -1 +1
M ./kernel/tensor4.c -1 +1
M ./kernel/tensor5.c -1 +1
M ./kernel/tensor7.c -1 +1
M ./kernel/tensor8.c -1 +1
M ./kernel/tensor9.c -1 +1
M ./kernel/tile2d.c -1 +1
M ./kernel/timer.c -1 +1
M ./kernel/transpose.c -1 +1
M ./kernel/trig.c -1 +1
M ./kernel/twiddle.c -1 +1
M ./libbench2/aligned-main.c -1 +1
M ./libbench2/bench-main.c -1 +1
M ./libbench2/bench-user.h -1 +1
M ./libbench2/bench.h -1 +1
M ./libbench2/can-do.c -1 +1
M ./libbench2/dotens2.c -1 +1
M ./libbench2/info.c -1 +1
M ./libbench2/main.c -1 +1
M ./libbench2/my-getopt.c -1 +1
M ./libbench2/my-getopt.h -1 +1
M ./libbench2/problem.c -1 +1
M ./libbench2/report.c -1 +1
M ./libbench2/speed.c -1 +1
M ./libbench2/tensor.c -1 +1
M ./libbench2/timer.c -1 +1
M ./libbench2/useropt.c -1 +1
M ./libbench2/util.c -1 +1
M ./libbench2/verify-dft.c -1 +1
M ./libbench2/verify-lib.c -1 +1
M ./libbench2/verify-r2r.c -1 +1
M ./libbench2/verify-rdft2.c -1 +1
M ./libbench2/verify.c -1 +1
M ./libbench2/verify.h -1 +1
M ./libbench2/zero.c -1 +1
M ./mpi/any-true.c -1 +1
M ./mpi/api.c -1 +1
M ./mpi/block.c -1 +1
M ./mpi/choose-radix.c -1 +1
M ./mpi/conf.c -1 +1
M ./mpi/dft-problem.c -1 +1
M ./mpi/dft-rank-geq2-transposed.c -1 +1
M ./mpi/dft-rank-geq2.c -1 +1
M ./mpi/dft-rank1-bigvec.c -1 +1
M ./mpi/dft-rank1.c -1 +1
M ./mpi/dft-serial.c -1 +1
M ./mpi/dft-solve.c -1 +1
M ./mpi/dtensor.c -1 +1
M ./mpi/ifftw-mpi.h -1 +1
M ./mpi/mpi-dft.h -1 +1
M ./mpi/mpi-rdft.h -1 +1
M ./mpi/mpi-rdft2.h -1 +1
M ./mpi/mpi-transpose.h -1 +1
M ./mpi/rdft-problem.c -1 +1
M ./mpi/rdft-rank-geq2-transposed.c -1 +1
M ./mpi/rdft-rank-geq2.c -1 +1
M ./mpi/rdft-rank1-bigvec.c -1 +1
M ./mpi/rdft-serial.c -1 +1
M ./mpi/rdft-solve.c -1 +1
M ./mpi/rdft2-problem.c -1 +1
M ./mpi/rdft2-rank-geq2-transposed.c -1 +1
M ./mpi/rdft2-rank-geq2.c -1 +1
M ./mpi/rdft2-serial.c -1 +1
M ./mpi/rdft2-solve.c -1 +1
M ./mpi/rearrange.c -1 +1
M ./mpi/testsched.c -1 +1
M ./mpi/transpose-alltoall.c -1 +1
M ./mpi/transpose-pairwise.c -1 +1
M ./mpi/transpose-problem.c -1 +1
M ./mpi/transpose-recurse.c -1 +1
M ./mpi/transpose-solve.c -1 +1
M ./mpi/wisdom-api.c -1 +1
M ./rdft/buffered.c -1 +1
M ./rdft/buffered2.c -1 +1
M ./rdft/codelet-rdft.h -1 +1
M ./rdft/conf.c -1 +1
M ./rdft/ct-hc2c-direct.c -1 +1
M ./rdft/ct-hc2c.c -1 +1
M ./rdft/ct-hc2c.h -1 +1
M ./rdft/dft-r2hc.c -1 +1
M ./rdft/dht-r2hc.c -1 +1
M ./rdft/dht-rader.c -1 +1
M ./rdft/direct-r2c.c -1 +1
M ./rdft/direct-r2r.c -1 +1
M ./rdft/direct2.c -1 +1
M ./rdft/generic.c -1 +1
M ./rdft/hc2hc-direct.c -1 +1
M ./rdft/hc2hc-generic.c -1 +1
M ./rdft/hc2hc.c -1 +1
M ./rdft/hc2hc.h -1 +1
M ./rdft/indirect.c -1 +1
M ./rdft/khc2c.c -1 +1
M ./rdft/khc2hc.c -1 +1
M ./rdft/kr2c.c -1 +1
M ./rdft/kr2r.c -1 +1
M ./rdft/nop.c -1 +1
M ./rdft/nop2.c -1 +1
M ./rdft/plan.c -1 +1
M ./rdft/plan2.c -1 +1
M ./rdft/problem.c -1 +1
M ./rdft/problem2.c -1 +1
M ./rdft/rank-geq2-rdft2.c -1 +1
M ./rdft/rank-geq2.c -1 +1
M ./rdft/rank0-rdft2.c -1 +1
M ./rdft/rank0.c -1 +1
M ./rdft/rdft-dht.c -1 +1
M ./rdft/rdft.h -1 +1
M ./rdft/rdft2-inplace-strides.c -1 +1
M ./rdft/rdft2-rdft.c -1 +1
M ./rdft/rdft2-strides.c -1 +1
M ./rdft/rdft2-tensor-max-index.c -1 +1
M ./rdft/scalar/hb.h -1 +1
M ./rdft/scalar/hc2c.c -1 +1
M ./rdft/scalar/hc2cb.h -1 +1
M ./rdft/scalar/hc2cf.h -1 +1
M ./rdft/scalar/hf.h -1 +1
M ./rdft/scalar/hfb.c -1 +1
M ./rdft/scalar/r2c.c -1 +1
M ./rdft/scalar/r2cb.h -1 +1
M ./rdft/scalar/r2cbIII.h -1 +1
M ./rdft/scalar/r2cf.h -1 +1
M ./rdft/scalar/r2cfII.h -1 +1
M ./rdft/scalar/r2r.c -1 +1
M ./rdft/scalar/r2r.h -1 +1
M ./rdft/simd/common/genus.c -1 +1
M ./rdft/simd/hc2cbv.h -1 +1
M ./rdft/simd/hc2cfv.h -1 +1
M ./rdft/solve.c -1 +1
M ./rdft/solve2.c -1 +1
M ./rdft/vrank-geq1-rdft2.c -1 +1
M ./rdft/vrank-geq1.c -1 +1
M ./rdft/vrank3-transpose.c -1 +1
M ./reodft/conf.c -1 +1
M ./reodft/redft00e-r2hc-pad.c -1 +1
M ./reodft/redft00e-r2hc.c -1 +1
M ./reodft/reodft.h -1 +1
M ./reodft/reodft00e-splitradix.c -1 +1
M ./reodft/reodft010e-r2hc.c -1 +1
M ./reodft/reodft11e-r2hc-odd.c -1 +1
M ./reodft/reodft11e-r2hc.c -1 +1
M ./reodft/reodft11e-radix2.c -1 +1
M ./reodft/rodft00e-r2hc-pad.c -1 +1
M ./reodft/rodft00e-r2hc.c -1 +1
M ./simd-support/altivec.c -1 +1
M ./simd-support/amd64-cpuid.h -1 +1
M ./simd-support/avx.c -1 +1
M ./simd-support/neon.c -1 +1
M ./simd-support/simd-altivec.h -1 +1
M ./simd-support/simd-avx.h -1 +1
M ./simd-support/simd-common.h -1 +1
M ./simd-support/simd-neon.h -1 +1
M ./simd-support/simd-sse2.h -1 +1
M ./simd-support/sse2-nonportable.c -1 +1
M ./simd-support/sse2.c -1 +1
M ./simd-support/taint.c -1 +1
M ./simd-support/x86-cpuid.h -1 +1
M ./simd/altivec.c -1 +1
M ./simd/mips_ps.c -1 +1
M ./simd/mips_ps.h -1 +1
M ./simd/nonportable/sse.c -1 +1
M ./simd/nonportable/sse2.c -1 +1
M ./simd/simd-altivec.h -1 +1
M ./simd/simd-mips_ps.h -1 +1
M ./simd/simd-sse.h -1 +1
M ./simd/simd-sse2.h -1 +1
M ./simd/simd.h -1 +1
M ./simd/sse.c -1 +1
M ./simd/sse2.c -1 +1
M ./simd/taint.c -1 +1
M ./simd/x86-cpuid.h -1 +1
M ./threads/api.c -1 +1
M ./threads/conf.c -1 +1
M ./threads/ct.c -1 +1
M ./threads/dft-vrank-geq1.c -1 +1
M ./threads/f77api.c -1 +1
M ./threads/f77funcs.h -1 +1
M ./threads/hc2hc.c -1 +1
M ./threads/openmp.c -1 +1
M ./threads/rdft-vrank-geq1.c -1 +1
M ./threads/threads.c -1 +1
M ./threads/threads.h -1 +1
M ./threads/vrank-geq1-rdft2.c -1 +1
M ./tools/fftw-wisdom-to-conf.1 -1 +1
M ./tools/fftw-wisdom-to-conf.in -1 +1
M ./tools/fftw-wisdom.c -1 +1
M ./tools/fftw_wisdom.1.in -1 +1
Thu Aug 11 14:54:38 EDT 2011 [email protected]
* allow specifying TRANSPOSE_{IN/OUT} transpose plans, since libbench does not canonicalize rnk=1 n=1 plans as rnk=0
M ./mpi/mpi-bench.c -2 +2
Thu Aug 11 14:17:24 EDT 2011 [email protected]
* check.pl should occasionally check DESTROY_INPUT problems too (especially since those enable slightly different algorithms in MPI)
M ./tests/check.pl +1
Thu Aug 11 12:37:51 EDT 2011 [email protected]
* unify post-MPI transpose handling in pairwise and alltoall solvers; should make the former faster in the destroy-input out-of-place case, and the latter more widely applicable
M ./mpi/mpi-transpose.h +5
M ./mpi/transpose-alltoall.c -69 +24
M ./mpi/transpose-pairwise.c -66 +97
Mon Aug 8 10:06:14 EDT 2011 [email protected]
* Add support for ARM NEON
M ./Makefile.am -1 +6
M ./api/version.c +4
M ./configure.ac -5 +23
M ./dft/codelet-dft.h +1
M ./dft/conf.c +4
M ./dft/simd/Makefile.am -1 +1
A ./dft/simd/neon/
A ./dft/simd/neon/Makefile.am
M ./kernel/ifftw.h +2
M ./m4/ax_cc_maxopt.m4 +10
M ./rdft/codelet-rdft.h +1
M ./rdft/conf.c +4
M ./rdft/simd/Makefile.am -1 +1
A ./rdft/simd/neon/
A ./rdft/simd/neon/Makefile.am
M ./simd-support/Makefile.am -1 +2
A ./simd-support/neon.c
M ./simd-support/simd-common.h +3
A ./simd-support/simd-neon.h
Fri Aug 5 17:25:32 EDT 2011 [email protected]
* more C++ paranoia
M ./tests/fftw-bench.h +10
Fri Aug 5 17:02:00 EDT 2011 [email protected]
* tentative version bump for 3.3.1
M ./NEWS +5
M ./configure.ac -2 +2
Fri Aug 5 16:52:28 EDT 2011 [email protected]
* fixes so that MPI code compiles when MPICC is a C++ compiler, even if the serial code is compiled with a C compiler; thanks to Kyle Spyksma for the bug report
M ./api/api.h +12
M ./api/plan-guru-r2r.h -2
M ./api/plan-many-r2r.c -2
M ./dft/dft.h +9
M ./kernel/ifftw.h +9
M ./mpi/api.c -3 +1
M ./mpi/dtensor.c -1 +1
M ./mpi/ifftw-mpi.h -1 +4
M ./mpi/mpi-bench.c -1 +1
M ./rdft/rdft.h +9
Fri Aug 5 16:04:06 EDT 2011 [email protected]
* use correct precision in f03-wrap.c, avoiding a (harmless) implicit pointer cast that prevented compilation under C++; thanks to Kyle Spyksma for the bug report
M ./mpi/genf03-wrap.pl +2
Fri Aug 5 14:04:32 EDT 2011 [email protected]
* manual typo
M ./doc/modern-fortran.texi -1 +1
Wed Jul 27 12:50:25 EDT 2011 [email protected]
tagged fftw-3.3
Tue Jul 26 20:55:45 EDT 2011 [email protected]
* Honor WITH_OUR_MALLOC in libbench2
M ./libbench2/util.c -6 +5
Tue Jul 26 20:27:28 EDT 2011 [email protected]
* fixed typo: incorrect name of combined threads library on Windows
M ./Makefile.am -1 +1
Mon Jul 25 14:38:20 EDT 2011 [email protected]
* 3.3 version bump & NEWS
M ./NEWS -5 +16
M ./configure.ac -1 +1
Mon Jul 25 14:37:48 EDT 2011 [email protected]
* use int(..., C_SIZE_T) rather than declaring another variable in the Fortran examples
M ./doc/modern-fortran.texi -12 +9
Wed Jul 13 05:02:32 EDT 2011 [email protected]
* typo, thanks to Rhys Ulerich for the comment
M ./doc/mpi.texi -1 +1
Mon Jul 11 14:39:52 EDT 2011 [email protected]
* Fix bug in bubblesort
Bubblesort was not sorting. This was a bug in the benchmark library
(not in FFTW per se), and it impacted the benchmark program
with --report-time and --report-mflops causing it to output
an incorrect value for the median. (The minimum, maximum, and
average value were correct.) Thanks Dima Baksheev of Intel for
reporting this bug.
M ./libbench2/report.c -2 +2
Fri Jul 8 13:35:59 EDT 2011 [email protected]
* small manual typos
M ./doc/modern-fortran.texi -2 +2
Wed Jul 6 10:49:40 EDT 2011 [email protected]
* Detection of altivec.h requires $ALTIVEC_CFLAGS
M ./configure.ac -5 +4
Tue Jul 5 19:58:47 EDT 2011 [email protected]
* Introduce fake dependency so that my-getopt.c is recompiled
my-getopt.c does not depend on anything, and so it is not rebuilt when
reconfiguring for a different ISA (e.g., CC="gcc -m32" vs CC="gcc
-m64"). Add a fake dependency on <config.h> so that the file is
recompiled.
M ./libbench2/my-getopt.c +1
Tue Jul 5 18:53:36 EDT 2011 [email protected]
* support compiling/installing --enable-threads --enable-openmp at the same time, although in this case the test program only uses the threads variety. Update documentation accordingly, and in general expand the documentation of the OpenMP support
M ./Makefile.am -1 +1
M ./configure.ac -21 +13
M ./doc/install.texi -14 +13
M ./doc/threads.texi -19 +43
M ./mpi/Makefile.am -2 +6
M ./tests/Makefile.am -2 +6
M ./tests/fftw-bench.c -1 +3
M ./threads/Makefile.am -6 +20
M ./threads/conf.c -4
M ./threads/openmp.c -4
M ./threads/threads.c -3
M ./tools/Makefile.am -4 +8
Tue Jul 5 16:04:03 EDT 2011 [email protected]
* call omp_set_num_threads in fftw-bench so that the number of OpenMP threads corresponds with the number of FFTW threads
M ./tests/fftw-bench.c +5
Tue Jul 5 16:03:06 EDT 2011 [email protected]
* when --enable-openmp, install as fftw3_omp rather than fftw3_threads, so that both the POSIX threads and OpenMP variants of FFTW can be installed at once
M ./Makefile.am -1 +1
M ./configure.ac +4
M ./threads/Makefile.am -4 +4
Sat Jul 2 02:21:22 EDT 2011 [email protected]
* don't even declare an fftw_execute interface in Fortran, since it is unsafe and we recommend against it anyway; thanks to Arjen Markus for the suggestion
M ./api/f03api.sh -1 +1
M ./doc/modern-fortran.texi -2 +4
Fri Jul 1 14:35:44 EDT 2011 [email protected]
* consistently use the order single, double, long double
M ./support/BUILD-MINGW32.sh -1 +1
M ./support/BUILD-MINGW64.sh -1 +1
Wed Jun 29 17:27:06 EDT 2011 [email protected]
* MSVC AVX 64-bit detection does not work, punt for now.
M ./simd-support/amd64-cpuid.h -16 +2
Wed Jun 29 15:52:27 EDT 2011 [email protected]
* fixed typo, added note on transposed flags for r2c/c2r; thanks to Rhys Ulerich for the suggestions
M ./doc/mpi.texi -1 +14
Wed Jun 29 09:41:39 EDT 2011 [email protected]
* fixes for compiling with MSVC (untested)
M ./simd-support/amd64-cpuid.h -2 +16
M ./simd-support/simd-avx.h -11 +16
M ./simd-support/x86-cpuid.h -1 +7
Tue Jun 28 16:48:36 EDT 2011 [email protected]
* comment
M ./simd-support/sse2-nonportable.c +3
Mon Jun 27 21:01:56 EDT 2011 [email protected]
* rm extraneous line break in HTML output ... I hate texinfo
M ./doc/reference.texi -6 +3
Mon Jun 27 06:43:57 EDT 2011 [email protected]
tagged fftw-3.3-beta1
Mon Jun 27 00:47:33 EDT 2011 [email protected]
* maintainer-clean should delete html directory (otherwwise we keep obsolete HTML files in the dist tarball, sigh)
M ./doc/Makefile.am -2 +2
Mon Jun 27 00:45:28 EDT 2011 [email protected]
tagged fftw-3.3-beta1
Sun Jun 26 23:36:32 EDT 2011 [email protected]
* update copyright year in manual
M ./doc/license.texi -2 +2