forked from blitzpp/blitz
-
Notifications
You must be signed in to change notification settings - Fork 0
/
ChangeLog
1900 lines (1429 loc) · 91.6 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
changeset: 1917:77c9002404a7
tag: tip
user: Julian Cummings <[email protected]>
date: Tue Jul 03 16:30:09 2012 -0700
files: COPYRIGHT LEGAL LICENSE README
description:
Updated legal information in README and LEGAL files. Updated LICENSE file to use version 2.0 of the Perl Artistic License as required for use with Fedora. Updated language in COPYRIGHT file for BSD license.
changeset: 1916:7c6844d1a805
user: Julian Cummings <[email protected]>
date: Wed Jun 27 18:36:46 2012 -0700
files: NEWS
description:
A brief note on the new 0.10 snapshot release.
changeset: 1915:a8ea119bc0dc
user: Julian Cummings <[email protected]>
date: Wed Jun 27 18:27:21 2012 -0700
files: doc/stamp-vti doc/version.texi
description:
Update datestamp and version number for documentation.
changeset: 1914:fc43218c3b7e
user: Julian Cummings <[email protected]>
date: Wed Jun 27 18:23:08 2012 -0700
files: Makefile.am
description:
Added Blitz-VS2010.zip to the list of files included in the blitz package.
changeset: 1913:4014cbb0ff08
user: Julian Cummings <[email protected]>
date: Wed Jun 27 18:19:24 2012 -0700
files: Blitz-VS2010.zip
description:
New version of support files for compiling blitz with Microsoft Visual Studio 2010.
changeset: 1912:927504bb5536
user: Julian Cummings <[email protected]>
date: Tue Jun 26 17:33:54 2012 -0700
files: blitz/array/expr.h blitz/array/functorExpr.h blitz/array/stencil-et-macros.h blitz/array/where.h
description:
Patch from Sylwester Arabas for blitz compatibility with clang 3.1 compiler. Changes seem fine with g++ compiler.
changeset: 1911:e1a8ec015bde
user: Julian Cummings <[email protected]>
date: Tue Jun 26 17:15:12 2012 -0700
files: benchmarks/daxpy.cpp benchmarks/daxpy2.cpp benchmarks/loop4.cpp benchmarks/makeloops.cpp benchmarks/stencil.cpp
description:
Corrected typo in BenchmarkExt class member name that was discovered in several of the benchmark test codes.
changeset: 1910:8f8423701853
user: Julian Cummings <[email protected]>
date: Tue Jun 26 16:47:42 2012 -0700
files: blitz/globeval.cc blitz/memblock.h blitz/tmevaluate.h blitz/tvevaluate.h
description:
Patch from Matt Bingen to eliminate compiler warnings about using pragmas within macros and to use ptrdiff_t instead of int in a few more places for 64-bit safety.
changeset: 1909:11a167db8e5f
user: Julian Cummings <[email protected]>
date: Tue Jun 26 16:44:55 2012 -0700
files: blitz/array-impl.h
description:
Patch from Matt Bingen to allow the default storage type for Array to be configurable.
changeset: 1908:a46cacb46e8f
user: Julian Cummings <[email protected]>
date: Tue Jun 26 16:38:50 2012 -0700
files: testsuite/initialize.cpp
description:
Workarounds for testsuite code that attempts to use list initialization on Vectors that reference data stored non-consecutively in another Vector. Such code no longer works because the current stand-in Vector class is just a wrapper for the 1D Array class.
changeset: 1907:ac2b17fed6c4
user: Julian Cummings <[email protected]>
date: Tue Jun 26 14:28:43 2012 -0700
files: examples/Makefile.am
description:
Commented out several example programs which rely on the Vector classes and are therefore not currently functional. The Vector classes need to be updated to use the new style expression template machinery in blitz. The old style ET machinery has been removed to reduce the overall amount of blitz code.
changeset: 1906:f88ef337efee
user: Julian Cummings <[email protected]>
date: Tue Jun 26 14:25:30 2012 -0700
files: examples/where.cpp
description:
Modified example code to use 1D Array instead of Vector class, which is currently non-functional.
changeset: 1905:2fa73f1ae57a
user: Julian Cummings <[email protected]>
date: Tue Jun 26 14:24:22 2012 -0700
files: examples/tiny.cpp
description:
Updated include file list to eliminate <tinyvec-et.h>, which is obsolete.
changeset: 1904:ef381c095e49
user: Julian Cummings <[email protected]>
date: Tue Jun 26 14:13:11 2012 -0700
files: blitz/array-impl.h
description:
Revert ListInitializationSwitch back to using a T_numtype* rather than a T_iterator to traverse Array memory for list initialization. By definition, a comma-delimited list will initialize an Array in memory storage order, so a standard pointer is adequate as long as the memory storage is contiguous. This change fixes an error that was occurring in the storage testsuite code.
changeset: 1903:d1291bffbb19
user: Julian Cummings <[email protected]>
date: Tue Jun 26 14:07:16 2012 -0700
files: testsuite/reduce.cpp
description:
Removed sum expression that was not used in any test and was triggering an error.
changeset: 1902:56a36918ccf1
user: Patrik Jonsson <[email protected]>
date: Tue Apr 24 16:04:24 2012 -0400
files: configure.ac
description:
Removed check for boost::mpi library since we only need the include file.
changeset: 1901:20ebbb315dd7
user: Patrik Jonsson <[email protected]>
date: Tue Apr 24 11:56:15 2012 -0700
files: blitz/array/storage.h blitz/tinyvec2.h configure.ac
description:
Added a separate check for the Boost::MPI library when serialization is enabled, since one can imagine wanting serialization without MPI.
changeset: 1900:2663e207d21f
user: Patrik Jonsson <[email protected]>
date: Tue Apr 17 13:14:16 2012 -0400
files: blitz/array-impl.h blitz/array/ops.cc testsuite/Makefile.am testsuite/update.cpp
description:
Fixed Array update operators for scalars, which the previous change broke. Added testcase update to exercise the Array update operators for scalars, Arrays, and expressions.
changeset: 1899:5e7778d680d6
user: Patrik Jonsson <[email protected]>
date: Tue Apr 17 12:49:59 2012 -0400
files: blitz/array-impl.h blitz/array/ops.cc testsuite/multicomponent.cpp
description:
The Array update operators (like +=) are now defined exactly like the assignment operator, which was not the case before, so some expressions that worked as assignments didn't work as updates. Updated multicomponent test case to exercise this.
changeset: 1898:3acfc76c2df9
user: Patrik Jonsson <[email protected]>
date: Tue Mar 06 13:12:04 2012 -0500
files: blitz/array/ops.h testsuite/Makefile.am testsuite/bitwise.cpp
description:
Restored left and right shift operators, which were commented out, and added scalar versions of them as well. Added a test case for the bitwise operators (currently only for TinyVector). There appears to be no ambiguity to me, the operators take ETBase arguments and should not conflict with the I/O operators. At least, all tests pass and my code compiles fine.
changeset: 1897:262deb840568
user: Patrik Jonsson <[email protected]>
date: Wed Jan 25 10:48:31 2012 -0500
files: blitz/tinyvec2.h
description:
Added extent() member to TinyVector in analogy with Array and TinyMatrix.
changeset: 1896:d4a8862b671a
user: Patrik Jonsson <[email protected]>
date: Mon Dec 12 17:11:18 2011 -0500
files: blitz/tinymat2.cc
description:
One more warning fix.
changeset: 1895:4d8789826636
user: Patrik Jonsson <[email protected]>
date: Mon Dec 12 17:00:20 2011 -0500
files: blitz/array/reduce.h blitz/array/storage.h blitz/globeval.cc blitz/indexexpr.h blitz/memblock.cc blitz/shapecheck.h
description:
Fixed some warnings with gcc -Wall.
changeset: 1894:90c2f00d236e
user: Patrik Jonsson <[email protected]>
date: Tue Dec 06 18:55:54 2011 -0500
files: blitz/array/newet-macros.h blitz/funcs.h testsuite/Makefile.am testsuite/int-math-func.cpp
description:
Changed the BZ_DEFINE_UNARY_FUNC macro so the functors now have the option of using a double return type for integer types. This avoids bug 3381318 where doing sin(intarray) returns ints even if the results go into a floating array. Turned off this option for functions like abs() that can sensibly be called with integer types. Added testcase int-math-func to test this.
changeset: 1893:e29277a6148b
user: Patrik Jonsson <[email protected]>
date: Tue Dec 06 16:27:08 2011 -0500
files: blitz/array/funcs.h blitz/funcs.h
description:
Removed the special definitions of abs() in array/funcs.h. Since all types are overloaded as abs() in current cmath, this should no longer be necessary. This also fixes bug 3427014.
changeset: 1892:a8f6391064b0
user: Patrik Jonsson <[email protected]>
date: Tue Dec 06 16:24:18 2011 -0500
files: testsuite/Makefile.am
description:
Fixed a bug in the testsuite Makefile.am.
changeset: 1891:2b8b3ea5cd97
user: Patrik Jonsson <[email protected]>
date: Tue Dec 06 15:45:36 2011 -0500
files: testsuite/Makefile.am
description:
Changed the testsuite Makefile so all tests are run even if one fails.
changeset: 1890:4f7d8ef18b48
user: Patrik Jonsson <[email protected]>
date: Mon Dec 05 15:14:49 2011 -0500
files: blitz/blitz.h random/mt.h testsuite/serialize.cpp
description:
Added ability to serialize the Mersenne Twister state, and a test of this to the serialize test case. Also, blitz.h now includes bzconfig.h.
changeset: 1889:d4ea469afaa0
user: Patrik Jonsson <[email protected]>
date: Tue Nov 29 22:07:47 2011 -0500
files: blitz/globeval.cc testsuite/Makefile.am testsuite/indexexpr-base.cpp
description:
Fixed a bug in the 1D index traversal evaluation that caused results for arrays with nonzero base to be written out of bounds. Fixes issue 3441913. Added indexexpr_base to the testsuite to check this.
changeset: 1888:9a7e4e6adaa2
user: Patrik Jonsson <[email protected]>
date: Wed Nov 23 09:42:33 2011 -0500
files: blitz/array-impl.h blitz/array/storage.h blitz/tinymat2.h blitz/tinyvec2.h
description:
Moved declaration of boost::mpi::is_mpi_datatype out of the blitz namespace.
changeset: 1887:c3f9ffd1a3d3
user: Patrik Jonsson <[email protected]>
date: Tue Nov 22 15:39:45 2011 -0500
files: blitz/array-impl.h blitz/array/storage.h blitz/memblock.cc blitz/memblock.h blitz/tinymat2.h blitz/tinyvec2.h
description:
Improved the Boost::Serialization support so that it works with the skeleton/content functionality in boost::MPI. Declared boost::mpi::is_mpi_datatype appropriate data types.
changeset: 1886:9527b95a58b5
user: Patrik Jonsson <[email protected]>
date: Thu Oct 27 14:07:17 2011 -0400
files: blitz/memblock.h
description:
Fixed a segfault when serializing MemoryBlockReferences without a block.
changeset: 1885:41b58087d239
user: Patrik Jonsson <[email protected]>
date: Wed Oct 26 17:15:35 2011 -0400
files: configure.ac
description:
Added --enable-serialization configure option.
changeset: 1884:d05049b205e5
user: Patrik Jonsson <[email protected]>
date: Wed Oct 26 16:35:20 2011 -0400
files: benchmarks/Makefile.am examples/Makefile.am
description:
Fixed makefiles for benchmarks and examples dirs.
changeset: 1883:daba8c883045
user: Patrik Jonsson <[email protected]>
date: Wed Oct 26 16:24:41 2011 -0400
files: blitz/array-impl.h blitz/array/storage.h blitz/memblock.h blitz/tinymat2.h blitz/tinyvec2.h lib/Makefile.am testsuite/Makefile.am testsuite/serialize.cpp
description:
Updated flags for serialization support, tweaked Makefiles.
changeset: 1882:68f7fb7d191e
user: Patrik Jonsson <[email protected]>
date: Wed Oct 26 16:02:13 2011 -0400
files: blitz/array-impl.h blitz/array/storage.h blitz/memblock.cc blitz/memblock.h blitz/tinymat2.h blitz/tinyvec2.h configure.ac m4/ax_boost_base.m4 m4/ax_boost_serialization.m4 testsuite/Makefile.am testsuite/serialize.cpp
description:
Added Boost::Serialization support for Array, TinyVector and TinyMatrix. Added --with-boost and --with-boost-serialization configure options.
changeset: 1881:4de0b8fd18cd
user: Patrik Jonsson <[email protected]>
date: Thu Sep 29 15:35:03 2011 -0400
files: blitz/tmevaluate.h blitz/tvevaluate.h configure.ac testsuite/alignment.cpp
description:
Added macro BZ_USE_ALIGNMENT_PRAGMAS, which is only set to true if a SIMD width is specified. If this is not defined, the alignment pragmas are not used. This fixes bug 3415448.
changeset: 1880:f360eaedb125
user: Patrik Jonsson <[email protected]>
date: Wed Sep 28 08:47:19 2011 -0400
files: testsuite/64bit.cpp
description:
Updated 64bit test case to also test that fastRead for the expression classes also works with large arguments.
changeset: 1879:54b961d29fc0
user: Patrik Jonsson <[email protected]>
date: Tue Sep 27 16:12:21 2011 -0400
files: blitz/array/expr.h blitz/array/fastiter.h blitz/array/functorExpr.h blitz/array/where.h blitz/generate/genstencils.py blitz/indexexpr.h blitz/levicivita.h blitz/range.h blitz/tinymat2.h blitz/tinyvec2.h blitz/tm2fastiter.h blitz/tv2fastiter.h
description:
Changed argument of the fastRead() functions from int to diffType, which is also needed for 64-bit offsets to work.
changeset: 1878:ed2dfe67697b
parent: 1877:da484a923d69
parent: 1875:eb2f78ddf0b4
user: Patrik Jonsson <[email protected]>
date: Tue Sep 27 15:43:22 2011 -0400
files: testsuite/Makefile.am
description:
Merge.
changeset: 1877:da484a923d69
user: Patrik Jonsson <[email protected]>
date: Tue Sep 27 15:29:55 2011 -0400
files: blitz/globeval.cc testsuite/64bit.cpp testsuite/Makefile.am
description:
Further 64-bit fixes. Changed the 64bit test case to catch these problems, which unfortunately precludes only evaluating over a subrange. Therefore, I've enabled optimization for the 64bit test only so it doesn't take forever to run.
changeset: 1876:84f0b039f76d
parent: 1874:450f9b1ee2d0
user: Patrik Jonsson <[email protected]>
date: Tue Sep 27 14:31:34 2011 -0400
files: blitz/globeval.cc
description:
Changed offset type in evaluation from int to diffType, because otherwise evaluations with offsets > 1<<31 fails.
changeset: 1875:eb2f78ddf0b4
user: Patrik Jonsson <[email protected]>
date: Mon Sep 26 19:11:43 2011 -0700
files: blitz/array/expr.h testsuite/Makefile.am testsuite/cast.cpp
description:
Fixed broken cast float->double by removing a ArrayExprUnaryOp constructor that caused ambiguous resolution and apparently isn't actually used. Added test case cast.
changeset: 1874:450f9b1ee2d0
parent: 1873:1e77ae68003b
parent: 1871:4223fee30010
user: Patrik Jonsson <[email protected]>
date: Mon Sep 26 21:07:37 2011 -0400
description:
Merged.
changeset: 1873:1e77ae68003b
user: Patrik Jonsson <[email protected]>
date: Mon Sep 26 21:07:13 2011 -0400
files: testsuite/reduce.cpp
description:
Added test for bug 2058441.
changeset: 1872:760e2d93d633
parent: 1862:691355238c25
user: Patrik Jonsson <[email protected]>
date: Tue Jul 26 17:01:08 2011 -0400
files: blitz/array/reduce.h
description:
Added comment to reduce.h documenting bug 2058441.
changeset: 1871:4223fee30010
user: Patrik Jonsson <[email protected]>
date: Thu Aug 11 14:42:04 2011 -0700
files: blitz/tinyvec2.h
description:
Removed spurious inline declarations.
changeset: 1870:3448111f885d
user: Patrik Jonsson <[email protected]>
date: Thu Aug 11 11:40:38 2011 -0700
files: blitz/tmevaluate.h blitz/tvevaluate.h
description:
Fixed clobbering of template parameter.
changeset: 1869:fa21544fdcbd
user: Patrik Jonsson <[email protected]>
date: Thu Aug 11 09:49:15 2011 -0700
files: blitz/globeval.cc
description:
Fixed gcc complaining about rank() not being a constant-expression.
changeset: 1868:a6f80db77b94
user: patricg
date: Thu Jul 28 22:38:01 2011 +0100
files: doc/stamp-vti doc/version.texi
description:
Updated.
changeset: 1867:c73aba3b07e9
user: patricg
date: Thu Jul 28 22:23:42 2011 +0100
files: blitz/generate/Makefile.am
description:
Added python script to EXTRA_DIST in order to pass checkdist.
changeset: 1866:c243451132a6
user: patricg
date: Thu Jul 28 22:21:30 2011 +0100
files: doc/stencils/Makefile.am
description:
Renamed the preprocessorp macro "stenciltoapply" to the new name
corresponding to the new internal stencil operator (that take
iterators as opposed to the stencil ET expression operators that
take arrays), i.e. appended '_stencilop' to the operator name.
changeset: 1865:06bd97b19920
user: patricg
date: Thu Jul 28 21:21:48 2011 +0100
files: configure.ac
description:
Added clang++ in the list of C++ compilers to check.
changeset: 1864:fa98fbc6bd4d
user: patricg
date: Thu Jul 28 21:19:58 2011 +0100
files: m4/ac_cxx_flags_preset.m4
description:
Added an entry for the clang++ compiler (http://clang.llvm.org/).
changeset: 1863:d68ac0a95712
user: patricg
date: Thu Jul 28 21:16:18 2011 +0100
files: m4/ac_prog_doxygen.m4
description:
Changed default behaviour to doxygen disabled.
changeset: 1862:691355238c25
user: Patrik Jonsson <[email protected]>
date: Fri Jul 22 12:18:54 2011 -0400
files: blitz/array/ops.cc
description:
Added _bz_forceinline to array assignment operators.
changeset: 1861:85ea8a3b7b34
user: Patrik Jonsson <[email protected]>
date: Fri Jul 22 12:07:53 2011 -0400
files: blitz/globeval.cc
description:
Commented out '#pragma forceinline recursive' in the evaluation functions, since it seems to increase compilation time catastrophically.
changeset: 1860:e89dadf02f56
user: Patrik Jonsson <[email protected]>
date: Fri Jul 22 11:36:25 2011 -0400
files: blitz/Makefile.am
description:
Added tmevaluate.h to blitz_HEADERS.
changeset: 1859:ec1d33eba15b
user: Patrik Jonsson <[email protected]>
date: Fri Jul 22 11:24:21 2011 -0400
files: blitz/Makefile.am blitz/tinymat2.cc blitz/tinymat2.h blitz/tinyvec2.h blitz/tm2ops.cc blitz/tmevaluate.h blitz/tv2ops.cc blitz/tvevaluate.h
description:
Added compile-time selection of TinyVector-only expressions so the full evaluator is not instantiated in these cases. Ported the TinyVector-only evaluation machinery also for TinyMatrix, so TM expressions use the lightweight evaluation. (With some thinking, these two could probably be merged into one code.) With these changes, icpc v12 is now able to compile the multicomponent testcase.
changeset: 1858:c7413cdd3fac
user: Patrik Jonsson <[email protected]>
date: Thu Jul 21 16:26:31 2011 -0400
files: blitz/funcs.h
description:
pow2-8 now works also for complex scalars.
changeset: 1857:9a8d54c82f39
user: Patrik Jonsson <[email protected]>
date: Thu Jul 21 15:38:33 2011 -0400
files: blitz/funcs.h
description:
Restored pow2-8 functionality for builtin types.
changeset: 1856:d0150e83962d
user: Patrik Jonsson <[email protected]>
date: Thu Jul 21 13:38:50 2011 -0400
files: configure.ac
description:
Configure now defines BZ_ALIGN_VARIABLE to just a standard declaration if a simd width is not specified, rather than use alignment 1 as that gives warnings.
changeset: 1855:494a2ee3abf9
user: Patrik Jonsson <[email protected]>
date: Wed Jul 20 14:27:13 2011 -0700
files: blitz/globeval.cc
description:
Fixed compilation error with gcc, which apparently insists that fastRead_tv be explicitly template qualified.
changeset: 1854:f21b1f853e06
user: Patrik Jonsson <[email protected]>
date: Wed Jul 20 14:26:08 2011 -0700
files: blitz/generate/genstencils.py
description:
Fixed a typo in the generated stencil ET code.
changeset: 1853:75e5322949b6
user: Patrik Jonsson <[email protected]>
date: Wed Jul 20 14:25:26 2011 -0700
files: blitz/simdtypes.h
description:
Added inclusion of stdint.h to find uintptr_t.
changeset: 1852:1642b5d3f4df
user: Patrik Jonsson <[email protected]>
date: Wed Jul 20 17:15:28 2011 -0400
files: blitz/array/expr.cc blitz/indexexpr.h blitz/timer.h
description:
Fixed some compilation errors with gcc.
changeset: 1851:2cccdb4ade84
user: Patrik Jonsson <[email protected]>
date: Wed Jul 20 11:21:13 2011 -0400
files: blitz/array/asexpr.h
description:
Renamed template parameter T_result in the results traits classes.
changeset: 1850:e61bfbb651fb
user: Patrik Jonsson <[email protected]>
date: Tue Jul 19 15:25:27 2011 -0400
files: blitz/array-impl.h
description:
Removed 'restrict' from the getInitializationIterator return type since it doesn't make sense.
changeset: 1849:dc958e1d2c5b
user: Patrik Jonsson <[email protected]>
date: Tue Jul 19 15:24:29 2011 -0400
files: blitz/Makefile.am
description:
Added tvevaluate.h to blitz_HEADERS.
changeset: 1848:fbe66ccb53ee
user: Patrik Jonsson <[email protected]>
date: Tue Jul 19 15:23:36 2011 -0400
files: blitz/simdtypes.h
description:
Fixed an error where some compilers complain about the '<I>>1>' construct.
changeset: 1847:6e8b33f8b407
user: Patrik Jonsson <[email protected]>
date: Thu Jun 30 20:54:52 2011 -0700
files: blitz/memblock.cc blitz/memblock.h blitz/simdtypes.h
description:
Fixed a bug in the shift-to-cache-line during memoryBlock allocation that would cause badness when T_numtype didn't fit evenly into a cache line (like for multicomponent Arrays.)
changeset: 1846:4d4602e6c605
user: Patrik Jonsson <[email protected]>
date: Thu Jun 30 22:16:33 2011 -0400
files: blitz/tvevaluate.h
description:
Added tvevaluate.h which was forgotten in r1845.
changeset: 1845:d2b487a42b58
user: Patrik Jonsson <[email protected]>
date: Thu Jun 30 18:46:06 2011 -0400
files: blitz/globeval.cc blitz/tinyvec2.cc blitz/tuning.h
description:
Moved the _tv_evalute_(un)aligned functions into a class _tv_evaluator for the purpose of avoiding meta-unrolling very long TinyVector expressions. It is advantageous to binary-unroll short array expressions and then use wider vector widths than the simd width itself during evaluation. Added defines BZ_MAX_BITS_FOR_BINARY_UNROLL and BZ_VECTORIZED_LOOP_WIDTH to tuning.h to control this.
changeset: 1844:4db0fbbcc036
user: Patrik Jonsson <[email protected]>
date: Thu Jun 30 14:18:23 2011 -0400
files: blitz/memblock.cc
description:
Fixed typo in memblock.cc.
changeset: 1843:6049c40c2237
user: Patrik Jonsson <[email protected]>
date: Thu Jun 30 10:35:05 2011 -0700
files: blitz/array-impl.h blitz/levicivita.h blitz/range.h
description:
Fixed a few more warnings.
changeset: 1842:981ac7763b8d
user: Patrik Jonsson <[email protected]>
date: Thu Jun 30 10:16:39 2011 -0700
files: blitz/tuning.h
description:
Fixed warning about redefined macros in tuning.h
changeset: 1841:2adbcdadf146
user: Patrik Jonsson <[email protected]>
date: Thu Jun 30 10:15:18 2011 -0700
files: blitz/tuning.h
description:
Fixed warning about redefined macros in tuning.h
changeset: 1840:6e1433edffa5
user: Patrik Jonsson <[email protected]>
date: Thu Jun 30 09:50:15 2011 -0700
files: benchmarks/floop1.cpp benchmarks/floop10.cpp benchmarks/floop11.cpp benchmarks/floop12.cpp benchmarks/floop13.cpp benchmarks/floop14.cpp benchmarks/floop15.cpp benchmarks/floop16.cpp benchmarks/floop17.cpp benchmarks/floop18.cpp benchmarks/floop19.cpp benchmarks/floop2.cpp benchmarks/floop21.cpp benchmarks/floop22.cpp benchmarks/floop23.cpp benchmarks/floop24.cpp benchmarks/floop25.cpp benchmarks/floop3.cpp benchmarks/floop36.cpp benchmarks/floop5.cpp benchmarks/floop6.cpp benchmarks/floop8.cpp benchmarks/floop9.cpp benchmarks/loop1.cpp benchmarks/loop10.cpp benchmarks/loop100.cpp benchmarks/loop11.cpp benchmarks/loop12.cpp benchmarks/loop13.cpp benchmarks/loop14.cpp benchmarks/loop15.cpp benchmarks/loop16.cpp benchmarks/loop17.cpp benchmarks/loop18.cpp benchmarks/loop19.cpp benchmarks/loop2.cpp benchmarks/loop21.cpp benchmarks/loop22.cpp benchmarks/loop23.cpp benchmarks/loop24.cpp benchmarks/loop25.cpp benchmarks/loop3.cpp benchmarks/loop36.cpp benchmarks/loop5.cpp benchmarks/loop6.cpp benchmarks/loop8.cpp benchmarks/loop9.cpp blitz/generate/makeloops.py
description:
Removed the TinyVector version in the benchmark loops. Since we can't measure it reliably anyway, it seems more useful to run a denser set of array sizes.
changeset: 1839:d79f66a7f6e7
user: Patrik Jonsson <[email protected]>
date: Thu Jun 30 09:38:51 2011 -0700
files: blitz/memblock.cc blitz/simdtypes.h blitz/tuning.h
description:
Turned on the align blocks on cache line option, as it makes a substantial difference at least on my machine. Moved the definitions of the L1 cache line size and critical memory block size for aligning to tuning.h.
changeset: 1838:06aa6f2889f6
user: Patrik Jonsson <[email protected]>
date: Wed Jun 29 12:27:50 2011 -0700
files: blitz/array/ops.cc blitz/globeval.cc blitz/tinyvec2.cc blitz/tuning.h
description:
Added macro _bz_forceinline, which expands to __forceinline for icpc, and used that for the evaluation functions. Unfortunately, the directive doesn't actually do what it promises, leaving intermediate function calls. Asked Intel about this.
changeset: 1837:6c2b4079cfe6
user: Patrik Jonsson <[email protected]>
date: Wed Jun 29 10:39:20 2011 -0700
files: benchmarks/floop1.cpp benchmarks/floop10.cpp benchmarks/floop10f90.f90 benchmarks/floop11.cpp benchmarks/floop11f90.f90 benchmarks/floop12.cpp benchmarks/floop12f90.f90 benchmarks/floop13.cpp benchmarks/floop13f90.f90 benchmarks/floop14.cpp benchmarks/floop14f90.f90 benchmarks/floop15.cpp benchmarks/floop15f90.f90 benchmarks/floop16.cpp benchmarks/floop16f90.f90 benchmarks/floop17.cpp benchmarks/floop17f90.f90 benchmarks/floop18.cpp benchmarks/floop18f90.f90 benchmarks/floop19.cpp benchmarks/floop19f90.f90 benchmarks/floop1f90.f90 benchmarks/floop2.cpp benchmarks/floop21.cpp benchmarks/floop21f90.f90 benchmarks/floop22.cpp benchmarks/floop22f90.f90 benchmarks/floop23.cpp benchmarks/floop23f90.f90 benchmarks/floop24.cpp benchmarks/floop24f90.f90 benchmarks/floop25.cpp benchmarks/floop25f90.f90 benchmarks/floop2f90.f90 benchmarks/floop3.cpp benchmarks/floop36.cpp benchmarks/floop36f90.f90 benchmarks/floop5.cpp benchmarks/floop5f90.f90 benchmarks/floop6.cpp benchmarks/floop6f90.f90 benchmarks/floop8.cpp benchmarks/floop8f90.f90 benchmarks/floop9.cpp benchmarks/floop9f90.f90 benchmarks/loop1.cpp benchmarks/loop10.cpp benchmarks/loop100.cpp benchmarks/loop100f90.f90 benchmarks/loop10f90.f90 benchmarks/loop11.cpp benchmarks/loop11f90.f90 benchmarks/loop12.cpp benchmarks/loop12f90.f90 benchmarks/loop13.cpp benchmarks/loop13f90.f90 benchmarks/loop14.cpp benchmarks/loop14f90.f90 benchmarks/loop15.cpp benchmarks/loop15f90.f90 benchmarks/loop16.cpp benchmarks/loop16f90.f90 benchmarks/loop17.cpp benchmarks/loop17f90.f90 benchmarks/loop18.cpp benchmarks/loop18f90.f90 benchmarks/loop19.cpp benchmarks/loop19f90.f90 benchmarks/loop1f90.f90 benchmarks/loop2.cpp benchmarks/loop21.cpp benchmarks/loop21f90.f90 benchmarks/loop22.cpp benchmarks/loop22f90.f90 benchmarks/loop23.cpp benchmarks/loop23f90.f90 benchmarks/loop24.cpp benchmarks/loop24f90.f90 benchmarks/loop25.cpp benchmarks/loop25f90.f90 benchmarks/loop2f90.f90 benchmarks/loop3.cpp benchmarks/loop36.cpp benchmarks/loop36f90.f90 benchmarks/loop5.cpp benchmarks/loop5f90.f90 benchmarks/loop6.cpp benchmarks/loop6f90.f90 benchmarks/loop8.cpp benchmarks/loop8f90.f90 benchmarks/loop9.cpp benchmarks/loop9f90.f90 blitz/generate/makeloops.py
description:
Fixed yet another bug in makeloops.
changeset: 1836:52fb01a72be8
user: Patrik Jonsson <[email protected]>
date: Wed Jun 29 10:11:57 2011 -0700
files: benchmarks/floop1.cpp benchmarks/floop10.cpp benchmarks/floop10f.f benchmarks/floop11.cpp benchmarks/floop11f.f benchmarks/floop12.cpp benchmarks/floop12f.f benchmarks/floop13.cpp benchmarks/floop13f.f benchmarks/floop14.cpp benchmarks/floop14f.f benchmarks/floop15.cpp benchmarks/floop15f.f benchmarks/floop16.cpp benchmarks/floop16f.f benchmarks/floop17.cpp benchmarks/floop17f.f benchmarks/floop18.cpp benchmarks/floop18f.f benchmarks/floop19.cpp benchmarks/floop19f.f benchmarks/floop1f.f benchmarks/floop2.cpp benchmarks/floop21.cpp benchmarks/floop21f.f benchmarks/floop22.cpp benchmarks/floop22f.f benchmarks/floop23.cpp benchmarks/floop23f.f benchmarks/floop24.cpp benchmarks/floop24f.f benchmarks/floop25.cpp benchmarks/floop25f.f benchmarks/floop2f.f benchmarks/floop3.cpp benchmarks/floop36.cpp benchmarks/floop36f.f benchmarks/floop3f.f benchmarks/floop5.cpp benchmarks/floop5f.f benchmarks/floop6.cpp benchmarks/floop6f.f benchmarks/floop8.cpp benchmarks/floop8f.f benchmarks/floop9.cpp benchmarks/floop9f.f benchmarks/loop1.cpp benchmarks/loop10.cpp benchmarks/loop100.cpp benchmarks/loop100f.f benchmarks/loop10f.f benchmarks/loop11.cpp benchmarks/loop11f.f benchmarks/loop12.cpp benchmarks/loop12f.f benchmarks/loop13.cpp benchmarks/loop13f.f benchmarks/loop14.cpp benchmarks/loop14f.f benchmarks/loop15.cpp benchmarks/loop15f.f benchmarks/loop16.cpp benchmarks/loop16f.f benchmarks/loop17.cpp benchmarks/loop17f.f benchmarks/loop18.cpp benchmarks/loop18f.f benchmarks/loop19.cpp benchmarks/loop19f.f benchmarks/loop1f.f benchmarks/loop2.cpp benchmarks/loop21.cpp benchmarks/loop21f.f benchmarks/loop22.cpp benchmarks/loop22f.f benchmarks/loop23.cpp benchmarks/loop23f.f benchmarks/loop24.cpp benchmarks/loop24f.f benchmarks/loop25.cpp benchmarks/loop25f.f benchmarks/loop2f.f benchmarks/loop3.cpp benchmarks/loop36.cpp benchmarks/loop36f.f benchmarks/loop3f.f benchmarks/loop5.cpp benchmarks/loop5f.f benchmarks/loop6.cpp benchmarks/loop6f.f benchmarks/loop8.cpp benchmarks/loop8f.f benchmarks/loop9.cpp benchmarks/loop9f.f blitz/generate/makeloops.py
description:
Fixed a bug in makeloops for loop expressions containing several statements.
changeset: 1835:3b5d96bde968
user: Patrik Jonsson <[email protected]>
date: Tue Jun 28 15:08:12 2011 -0700
files: benchmarks/Makefile.am benchmarks/arrdaxpy.cpp benchmarks/arrdaxpy.m benchmarks/arrdaxpy2.m benchmarks/arrdaxpyf.f
description:
Made the arrdaxpy benchmark actually be included in the bench targets and compile.
changeset: 1834:c31ec13b787a
user: Patrik Jonsson <[email protected]>
date: Tue Jun 28 14:34:11 2011 -0700
files: benchmarks/acoustic.cpp
description:
Fixed flops parameter and some warnings in acoustic benchmark.
changeset: 1833:0825284cb582
user: Patrik Jonsson <[email protected]>
date: Tue Jun 28 13:41:12 2011 -0700
files: blitz/array/expr.h blitz/array/stencil-et-macros.h blitz/array/stencil-et.h blitz/generate/genstencils.py blitz/globeval.cc
description:
Operations that can not use the vectorized fastRead_tv can now indicate this by setting minWidth=0. This will bypass the vectorized operations in _bz_evaluateWithUnitStride and go directly to the scalar path. Used this mechanism to turn of vectorization for stencils. Finished updating stencil classes with fastRead_tv. Tests with stencils now work again.
changeset: 1832:8ad1428d6b8d
user: Patrik Jonsson <[email protected]>
date: Tue Jun 28 12:36:50 2011 -0700
files: blitz/array/expr.h blitz/array/functorExpr.h blitz/array/map.h blitz/array/reduce.h blitz/array/stencil-et.h blitz/array/where.h blitz/generate/genstencils.py blitz/indexexpr.h blitz/levicivita.h blitz/range.h blitz/tv2fastiter.h
description:
Further cleanup of the now-unused tvtypeprop typedefs in the ET classes. Added fastRead_tv() and the tvresult classes to the stencil ET classes. Stencils are now attempted to be vectorized, but it's not clear how this functionality should work so at this point we get a precondition failure. Need to either work out how to do it or how to not vectorize them.
changeset: 1831:8e406dc4dd1c
user: Patrik Jonsson <[email protected]>
date: Tue Jun 28 12:34:05 2011 -0700
files: blitz/globeval.cc
description:
Rank>1 stack traversals now also use the evaluateWithUnitStride and evaluateWithCommonStride functions for the inner loop, if appropriate. This should enable vectorization for the inner loops of higher-rank expressions as long as they are unit stride.
changeset: 1830:f3114d86ba95
user: Patrik Jonsson <[email protected]>
date: Tue Jun 28 09:14:08 2011 -0700
files: blitz/array/expr.h blitz/array/functorExpr.h blitz/array/where.h
description:
Cleaned up old fastRead_tv from readHelper classes.
changeset: 1829:469fb9bd9755
user: Patrik Jonsson <[email protected]>
date: Tue Jun 28 09:13:51 2011 -0700
files: testsuite/loop1.cpp
description:
Updated loop1 test case to use new benchext interface.
changeset: 1828:7e076a476614
user: Patrik Jonsson <[email protected]>
date: Tue Jun 28 08:59:41 2011 -0700
files: benchmarks/floop1.cpp benchmarks/floop10.cpp benchmarks/floop11.cpp benchmarks/floop12.cpp benchmarks/floop13.cpp benchmarks/floop14.cpp benchmarks/floop15.cpp benchmarks/floop16.cpp benchmarks/floop17.cpp benchmarks/floop18.cpp benchmarks/floop19.cpp benchmarks/floop2.cpp benchmarks/floop21.cpp benchmarks/floop22.cpp benchmarks/floop23.cpp benchmarks/floop24.cpp benchmarks/floop25.cpp benchmarks/floop3.cpp benchmarks/floop36.cpp benchmarks/floop5.cpp benchmarks/floop6.cpp benchmarks/floop8.cpp benchmarks/floop9.cpp benchmarks/loop1.cpp benchmarks/loop10.cpp benchmarks/loop100.cpp benchmarks/loop11.cpp benchmarks/loop12.cpp benchmarks/loop13.cpp benchmarks/loop14.cpp benchmarks/loop15.cpp benchmarks/loop16.cpp benchmarks/loop17.cpp benchmarks/loop18.cpp benchmarks/loop19.cpp benchmarks/loop2.cpp benchmarks/loop21.cpp benchmarks/loop22.cpp benchmarks/loop23.cpp benchmarks/loop24.cpp benchmarks/loop25.cpp benchmarks/loop3.cpp benchmarks/loop36.cpp benchmarks/loop5.cpp benchmarks/loop6.cpp benchmarks/loop8.cpp benchmarks/loop9.cpp blitz/benchext.cc blitz/benchext.h blitz/generate/makeloops.py blitz/timer.h
description:
Benchext functionality with dependent variable now works correctly. Updated makeloops so loop benchmarks make use of this, and also added array tensor notation to the benchmarks.
changeset: 1827:85b6c418c275
user: Patrik Jonsson <[email protected]>
date: Mon Jun 27 15:32:51 2011 -0700
files: benchmarks/acoustic.cpp
description:
Updated acoustic benchmark to use the benchext and run on several sizes. Not working right yet, though.
changeset: 1826:46a96a0bc719
user: Patrik Jonsson <[email protected]>
date: Mon Jun 27 15:32:28 2011 -0700
files: blitz/benchext.cc blitz/benchext.h
description:
Changed the "rate decription" in benchext to explicitly setting the dependent variable. Depending on whether the timer is reporting seconds or cycles, it will then create the proper output quantity. Also added method currentImplementation() that returns the implementation currently being benchmarked, so it can easily be printed.
changeset: 1825:870827755b84
user: Patrik Jonsson <[email protected]>
date: Mon Jun 27 14:22:10 2011 -0700
files: blitz/globeval.cc
description:
Fixed typo in globeval.cc.
changeset: 1824:729fe7dab4b8
user: Patrik Jonsson <[email protected]>
date: Mon Jun 27 14:21:58 2011 -0700
files: benchmarks/daxpy.cpp benchmarks/tinydaxpy.cpp
description:
Updated daxpy and tinydaxpy tests to run with new ET classes.
changeset: 1823:979407299a71
user: Patrik Jonsson <[email protected]>
date: Mon Jun 27 14:21:06 2011 -0700
files: blitz/timer.h src/globals.cpp
description:
Made timer ivar_ member a string instead of const char*.
changeset: 1822:6e69cd6bf77c
user: Patrik Jonsson <[email protected]>
date: Mon Jun 27 14:15:01 2011 -0700
files: benchmarks/acou3d.cpp benchmarks/acou3db4.cpp benchmarks/acoustic.cpp
description:
Updated acoustic benchmarks to work with new stencil operator names.
changeset: 1821:bf4d2dbfde7e
user: Patrik Jonsson <[email protected]>
date: Mon Jun 27 14:13:50 2011 -0700
files: blitz/timer.h src/globals.cpp
description:
Reverted timer to report seconds instead of us.
changeset: 1820:becc65b42c96
user: Patrik Jonsson <[email protected]>
date: Mon Jun 27 14:13:11 2011 -0700
files: blitz/traversal.cc
description:
Fixed typo in traversal.cc
changeset: 1819:062688a03ef3
user: Patrik Jonsson <[email protected]>
date: Mon Jun 27 13:18:50 2011 -0700
files: blitz/array/stencils.h testsuite/Makefile.am testsuite/stencil-extent.cpp
description:
Made stencilExtent member functions const so the automatic stencil extent detection works, and added testcase stencil-extent to test this.
changeset: 1818:513fd16d68d8
user: Patrik Jonsson <[email protected]>
date: Mon Jun 27 12:01:12 2011 -0700
files: blitz/traversal.cc blitz/traversal.h blitz/vector2.h
description:
Added Vector default constructor and updated traversal functions to use new Vector instead of old.
changeset: 1817:18a93bbfc6e5
user: Patrik Jonsson <[email protected]>
date: Fri Jun 24 12:08:25 2011 -0700
files: blitz/array-impl.h blitz/array/methods.cc blitz/memblock.h
description:
Added method blockLength() to MemoryBlockReference and made Array::storageSize() use that. Converted some comments to doxygen strings.
changeset: 1816:3268cc91465c
user: Patrik Jonsson <[email protected]>
date: Thu Jun 23 17:30:24 2011 -0700
files: blitz/array-impl.h blitz/array/funcs.h blitz/array/interlace.cc blitz/array/methods.cc blitz/array/resize.cc blitz/array/storage.h blitz/blitz.h blitz/memblock.h blitz/tinyvec2.h testsuite/chris-jeffery-2.cpp testsuite/contiguous.cpp testsuite/peter-bienstman-3.cpp testsuite/preexisting.cpp testsuite/transpose.cpp testsuite/troyer-genilloud.cpp
description:
Moved the specification of padded/unpadded storage to the GeneralArrayStorage object, and created shortcuts paddedArray and contiguousArray shortcuts. Preexisting memory constructors will now throw an exception if the specified array is noncontiguous. Tests were updated to specify contiguous arrays if they rely on knowing the strides.
changeset: 1815:25e0fde6666c
user: Patrik Jonsson <[email protected]>
date: Thu Jun 23 17:30:02 2011 -0700
files: blitz/simdtypes.h
description:
Removed specialization of simdTypes for bools as we can't adjust alignment anyway.
changeset: 1814:9ba706d744f5
user: Patrik Jonsson <[email protected]>
date: Thu Jun 23 14:42:32 2011 -0700
files: blitz/array-impl.h blitz/array/interlace.cc blitz/array/methods.cc blitz/array/resize.cc blitz/array/storage.h blitz/blitz.h blitz/simdtypes.h blitz/tinyvec2.h configure.ac testsuite/contiguous.cpp testsuite/peter-bienstman-3.cpp testsuite/preexisting.cpp testsuite/transpose.cpp
description:
Added argument paddingPolicy to Array constructors and resize methods. The default value is set by the --enable-array-length-padding argument to configure. Reduced duplicated code in constructors. Restored tests that test storage to just specify unpadded storage.
changeset: 1813:a672a5056a2d
user: Patrik Jonsson <[email protected]>
date: Thu Jun 23 13:42:24 2011 -0700
files: testsuite/alignment.cpp
description:
Updated alignment test to check paddedLength, offsetToAlignment and _bz_meta_bitwidth.
changeset: 1812:2c9dd168eb7c
parent: 1808:7f1cd86bf252
user: Patrik Jonsson <[email protected]>
date: Thu Jun 23 13:36:16 2011 -0700
files: blitz/simdtypes.h
description:
Added paddedLength() function to simdTypes, and rewrite offsetToAlignment to not do a modulo. Added _bz_meta_bitwidth metaprogram to calculate the number of bits needed to fit a number.
changeset: 1811:e8364831590c
branch: vectorization
parent: 1605:1a3d913c4c80
user: Patrik Jonsson <[email protected]>
date: Wed Jun 22 08:33:02 2011 -0700
description:
Closed vectorization branch, this was only a proof-of-concept.
changeset: 1810:7e7b743d45dc
branch: compat-64bit
parent: 1662:243a3797fdec
user: Patrik Jonsson <[email protected]>
date: Wed Jun 22 08:30:27 2011 -0700
description:
Closed compat-64bit branch, as this has was merged in CVS.
changeset: 1809:115f8f0c4aa3
branch: stencil-et
parent: 1684:f42d3e515fbd
user: Patrik Jonsson <[email protected]>
date: Wed Jun 22 08:27:58 2011 -0700
description:
Closed stencil-et branch, as this has was merged in CVS.
changeset: 1808:7f1cd86bf252
parent: 1807:67c9701bd14b
parent: 1796:695e799bdc3f
user: Patrik Jonsson <[email protected]>
date: Wed Jun 22 08:09:58 2011 -0700
description:
Merge.
changeset: 1807:67c9701bd14b
user: Patrik Jonsson <[email protected]>
date: Wed Jun 22 08:08:57 2011 -0700
files: blitz/array-impl.h blitz/array/methods.cc configure.ac testsuite/contiguous.cpp testsuite/ctors.cpp testsuite/peter-bienstman-3.cpp testsuite/preexisting.cpp testsuite/transpose.cpp
description:
Added configure option --enable-array-length-padding as a switch for padding the minor rank of arrays to even SIMD widths. Added back test of uneven array sizes in contiguous test, and removed the warning printout from the tests that assumed they knew what the strides should be unless this option is enabled.
changeset: 1806:2be82135f511
user: Patrik Jonsson <[email protected]>
date: Wed Jun 22 07:58:17 2011 -0700
files: testsuite/expression-slicing.cpp testsuite/safeToReturn.cpp
description:
Fixed safeToReturn and expression-slicing tests to not compare floating-point expressions for equality.
changeset: 1805:76b227678361
parent: 1804:c5164c57d20b
parent: 1799:51e03b388db3
user: Patrik Jonsson <[email protected]>
date: Wed Jun 22 09:54:42 2011 -0400
description:
Merged from governator.
changeset: 1804:c5164c57d20b
parent: 1803:0759f0ead3b9
parent: 1798:f6f34b144649
user: Patrik Jonsson <[email protected]>
date: Wed Jun 22 09:50:55 2011 -0400
files: blitz/applics.h blitz/array-old.h blitz/array/eval.cc blitz/benchext.cc blitz/extremum.h blitz/matdiag.h blitz/matexpr.h blitz/matgen.h blitz/mathf2.h blitz/matltri.h blitz/matref.h blitz/matrix.cc blitz/matrix.h blitz/matsymm.h blitz/mattoep.h blitz/matutri.h blitz/mstruct.h blitz/rand-dunif.h blitz/rand-mt.h blitz/rand-normal.h blitz/rand-tt800.h blitz/rand-uniform.h blitz/random.h blitz/randref.h blitz/tiny.h blitz/tinymat.h blitz/tinymatexpr.h blitz/tinymatio.cc blitz/tinyvec-et.h blitz/tinyvec.h blitz/tinyvecio.cc blitz/tinyveciter.h blitz/vecaccum.cc blitz/vecall.cc blitz/vecany.cc blitz/veccount.cc blitz/vecdelta.cc blitz/vecdot.cc blitz/vecexpr.h blitz/vecexprwrap.h blitz/vecglobs.cc blitz/vecglobs.h blitz/vecio.cc blitz/veciter.h blitz/vecmax.cc blitz/vecmin.cc blitz/vecnorm.cc blitz/vecnorm1.cc blitz/vecpick.cc blitz/vecpick.h blitz/vecpickio.cc blitz/vecpickiter.h blitz/vecproduct.cc blitz/vecsum.cc blitz/vector-et.h blitz/vector.cc blitz/vector.h blitz/vecwhere.h blitz/zero.cc blitz/zero.h test.cc
description:
Merged from governator.
changeset: 1803:0759f0ead3b9
user: Patrik Jonsson <[email protected]>
date: Wed Jun 22 09:50:16 2011 -0400
files: blitz/benchext.cc blitz/benchext.h
description:
Benchext now uses long long integers to measure cycles instead of doubles.
changeset: 1802:dd0c79809d3c
parent: 1801:41a98caf5177
parent: 1763:0532e5ff7ab5
user: Patrik Jonsson <[email protected]>
date: Wed Jun 15 13:08:51 2011 -0400
description:
Merged makeloops.
changeset: 1801:41a98caf5177
user: Patrik Jonsson <[email protected]>
date: Wed Jun 15 13:08:15 2011 -0400
files: blitz/benchext.cc
description:
Removed some debugging output from benchext.
changeset: 1800:1934801ac8af
parent: 1762:33cf8f76121c
user: Patrik Jonsson <[email protected]>
date: Wed Jun 15 12:55:55 2011 -0400
files: blitz/benchext.cc blitz/benchext.h blitz/timer.h
description:
Changed PAPI timer to use PAPI_read_counters, because stopping and starting them is very expensive. Also changed so benchext only uses one timer class, since they never run concurrently and it complicates PAPI use.
changeset: 1799:51e03b388db3
user: Patrik Jonsson <[email protected]>
date: Wed Jun 22 06:54:14 2011 -0700
files: blitz/globeval.cc
description:
Fixed a typo in globeval.
changeset: 1798:f6f34b144649
user: Patrik Jonsson <[email protected]>
date: Tue Jun 21 20:59:58 2011 -0700
files: blitz/benchext.cc
description:
Tweaked pylab graph.
changeset: 1797:1521c60b11a6
parent: 1795:924b46cce6ab
user: Patrik Jonsson <[email protected]>
date: Tue Jun 21 20:59:24 2011 -0700
files: blitz/globeval.cc
description:
Tweaked unit stride evaluation after looking at float outputs.
changeset: 1796:695e799bdc3f
user: Paul P. Hilscher <[email protected]>
date: Wed Jun 22 01:38:27 2011 -0400
files: blitz/types.h
description:
Added types.h with typedefs for most famous array types (later also vectors)
e.g. typedef Array<double, 3> Array3d;
thus instead of writing
void Func(Array<double, 3> A) { Array<std::complex<double>, 4> B; ... }