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
|
# Tests that call 'compiler_stats_num_field' are skipped when debugging is on.
# See testsuite/driver/testlib.py.
def no_lint(name, opts):
opts.compiler_always_flags = \
[opt for opt in opts.compiler_always_flags if opt != '-dcore-lint' and opt != '-dcmm-lint']
setTestOpts(no_lint)
# Note [residency]
#
# Residency (peak_megabytes_allocated and max_bytes_used) is sensitive
# to when the major GC runs, which makes it inherently inaccurate.
# Sometime an innocuous change somewhere can shift things around such
# that the samples occur at a different time, and the residency
# appears to change (up or down) when the underlying profile hasn't
# really changed.
#
# However, please don't just ignore changes in residency. If you see
# a change in one of these figures, please check whether it is real or
# not as follows:
#
# * Run the test with old and new compilers, adding +RTS -h -i0.01
# (you don't need to compile anything for profiling or enable profiling
# libraries to get a heap profile).
# * view the heap profiles, read off the maximum residency. If it has
# really changed, then you know there's an issue.
test('T1969',
[compiler_stats_num_field('peak_megabytes_allocated', # Note [residency]
[(wordsize(32), 30, 15),
# 2010-05-17 14 (x86/Windows)
# 15 (x86/OS X)
# 19 (x86/OS X)
# 2013-02-10 13 (x86/Windows)
# 2013-02-10 14 (x86/OSX)
# 2013-11-13 17 (x86/Windows, 64bit machine)
# 2015-07-11 21 (x86/Linux, 64bit machine) use +RTS -G1
# 2016-04-06 30 (x86/Linux, 64bit machine)
(wordsize(64), 55, 20)]),
# 28 (amd64/Linux)
# 34 (amd64/Linux)
# 2012-09-20 23 (amd64/Linux)
# 2012-10-03 25 (amd64/Linux if .hi exists)
# 2013-02-13 23, but unstable so increased to 10% range
# 2013-02-13 27, very unstable!
# 2014-09-10 29 (amd64/Linux) post-AMP-cleanup
# 2013-09-11 30, 10 (amd64/Linux)
# 2013-09-11 30, 15 (adapt to Phab CI)
# 2015-06-03 41, (amd64/Linux) use +RTS -G1
# 2015-10-28 55, (amd64/Linux) emit Typeable at definition site
compiler_stats_num_field('max_bytes_used',
[(platform('i386-unknown-mingw32'), 5719436, 20),
# 2010-05-17 5717704 (x86/Windows)
# 2013-02-10 5159748 (x86/Windows)
# 2013-02-10 5030080 (x86/Windows)
# 2013-11-13 7295012 (x86/Windows, 64bit machine)
# 2014-04-24 5719436 (x86/Windows, 64bit machine)
(wordsize(32), 9093608, 1),
# 6707308 (x86/OS X)
# 2009-12-31 6149572 (x86/Linux)
# 2014-01-22 6429864 (x86/Linux)
# 2014-06-29 5949188 (x86/Linux)
# 2015-07-11 6241108 (x86/Linux, 64bit machine) use +RTS -G1
# 2016-04-06 9093608 (x86/Linux, 64bit machine)
(wordsize(64), 15017528, 15)]),
# 2014-09-10 10463640, 10 # post-AMP-update (somewhat stabelish)
# looks like the peak is around ~10M, but we're
# unlikely to GC exactly on the peak.
# varies quite a lot with CLEANUP and BINDIST,
# hence 10% range.
# See Note [residency] to get an accurate view.
# 2014-09-14 9684256, 10 # try to lower it a bit more to match Phab's CI
# 2014-11-03 10584344, # ghcspeed reports higher numbers consistently
# 2015-07-11 11670120 (amd64/Linux)
# 2015-10-28 15017528 (amd64/Linux) emit typeable at definition site
compiler_stats_num_field('bytes allocated',
[(platform('i386-unknown-mingw32'), 301784492, 5),
# 215582916 (x86/Windows)
# 2012-10-29 298921816 (x86/Windows)
# 2013-02-10 310633884 (x86/Windows)
# 2013-11-13 317975916 (x86/Windows, 64bit machine)
# 2014-04-04 301784492 (x86/Windows, 64bit machine)
(wordsize(32), 344730660, 1),
# 221667908 (x86/OS X)
# 274932264 (x86/Linux)
# 2012-10-08 303930948 (x86/Linux, new codegen)
# 2013-02-10 322937684 (x86/OSX)
# 2014-01-22 316103268 (x86/Linux)
# 2014-06-29 303300692 (x86/Linux)
# 2015-07-11 288699104 (x86/Linux, 64-bit machine) use +RTS -G1
# 2016-04-06 344730660 (x86/Linux, 64-bit machine)
(wordsize(64), 695430728, 5)]),
# 17/11/2009 434845560 (amd64/Linux)
# 08/12/2009 459776680 (amd64/Linux)
# 17/05/2010 519377728 (amd64/Linux)
# 05/08/2011 561382568 (amd64/OS X)
# 16/07/2012 589168872 (amd64/Linux)
# 20/07/2012 595936240 (amd64/Linux)
# 23/08/2012 606230880 (amd64/Linux)
# 29/08/2012 633334184 (amd64/Linux) new codegen
# 18/09/2012 641959976 (amd64/Linux)
# 19/10/2012 661832592 (amd64/Linux) -fPIC turned on
# 23/10/2012 642594312 (amd64/Linux) -fPIC turned off again
# 12/11/2012 658786936 (amd64/Linux) UNKNOWN REASON
# 17/1/13: 667160192 (x86_64/Linux) new demand analyser
# 18/10/2013 698612512 (x86_64/Linux) fix for #8456
# 10/02/2014 660922376 (x86_64/Linux) call arity analysis
# 17/07/2014 651626680 (x86_64/Linux) roundabout update
# 10/09/2014 630299456 (x86_64/Linux) post-AMP-cleanup
# 03/06/2015 581460896 (x86_64/Linux) use +RTS -G1
# 28/10/2015 695430728 (x86_64/Linux) emit Typeable at definition site
only_ways(['normal']),
extra_hc_opts('-dcore-lint -static'),
# Leave -dcore-lint on for this one test, so that we have something
# that will catch a regression in -dcore-lint performance.
# Use `+RTS -G1` for more stable residency measurements. Note [residency].
extra_hc_opts('+RTS -G1 -RTS')
],
compile,
[''])
# This one tests for the space leak in the native code generator,
# where it holds onto the entire asm output until the end. The space
# leak reappears from time to time, so it's a good idea to have a test
# for it. The residency of this test will jump by 10MB to 40MB or so
# on x86-64 if the space leak appears.
# Only run this one if we have an NCG:
if 'optasm' in config.compile_ways:
conf_3294 = only_ways(['normal'])
else:
conf_3294 = skip
test('T3294',
[
compiler_stats_num_field('max_bytes_used', # Note [residency]
[(wordsize(32), 28686588, 15),
# 17725476 (x86/OS X)
# 14593500 (Windows)
# 2013-02-10 20651576 (x86/Windows)
# 2013-02-10 20772984 (x86/OSX)
# 2013-11-13 24009436 (x86/Windows, 64bit machine)
# 2014-04-24 19882188 (x86/Windows, 64bit machine)
# 2014-12-22 26525384 (x86/Windows) Increase due to silent superclasses?
# 2015-07-11 43196344 (x86/Linux, 64-bit machine) use +RTS -G1
# 2016-04-06 28686588 (x86/Linux, 64-bit machine)
(wordsize(64), 50367248, 20)]),
# prev: 25753192 (amd64/Linux)
# 29/08/2012: 37724352 (amd64/Linux)
# (increase due to new codegen, see #7198)
# 13/13/2012: 44894544 (amd64/Linux)
# (reason for increase unknown)
# 15/5/2013: 36904752 (amd64/Linux)
# (reason for decrease unknown)
# 29/5/2013: 43224080 (amd64/Linux)
# (reason for increase back to earlier value unknown)
# 2014-07-14: 36670800 (amd64/Linux)
# (reason unknown, setting expected value somewhere in between)
# 2015-01-22: 45000000 (amd64/Linux)
# varies between 40959592 and 52914488... increasing to +-20%
# 2015-10-28: 50367248 (amd64/Linux)
# D757: emit Typeable instances at site of type definition
compiler_stats_num_field('bytes allocated',
[(wordsize(32), 1377050640, 5),
# previous: 815479800 (x86/Linux)
# (^ increase due to new codegen, see #7198)
# 2012-10-08: 1373514844 (x86/Linux)
# 2013-11-13: 1478325844 (x86/Windows, 64bit machine)
# 2014-01-12: 1565185140 (x86/Linux)
# 2013-04-04: 1377050640 (x86/Windows, 64bit machine)
(wordsize(64), 2709595808, 5)]),
# old: 1357587088 (amd64/Linux)
# 29/08/2012: 2961778696 (amd64/Linux)
# (^ increase due to new codegen, see #7198)
# 18/09/2012: 2717327208 (amd64/Linux)
# 08/06/2013: 2901451552 (amd64/Linux) (reason unknown)
# 12/12/2013: 3083825616 (amd64/Linux) (reason unknown)
# 18/02/2014: 2897630040 (amd64/Linux) (call arity improvements)
# 12/03/2014: 2705289664 (amd64/Linux) (more call arity improvements)
# 2014-17-07: 2671595512 (amd64/Linux) (round-about update)
# 2014-09-10: 2709595808 (amd64/Linux) post-AMP cleanup
conf_3294,
# Use `+RTS -G1` for more stable residency measurements. Note [residency].
extra_hc_opts('+RTS -G1 -RTS')
],
compile,
[''])
test('T4801',
[ # expect_broken(5224),
# temporarily unbroken (#5227)
###################################
# deactivated for now, as this metric became too volatile recently
# compiler_stats_num_field('peak_megabytes_allocated',# Note [residency]
# [(platform('x86_64-apple-darwin'), 70, 1),
# # expected value: 58 (amd64/OS X)
# # 13/01/2014 - 70
# (wordsize(32), 30, 20),
# (wordsize(64), 48, 20)]),
# # prev: 50 (amd64/Linux)
# # 19/10/2012: 64 (amd64/Linux)
# # (^ REASON UNKNOWN!)
# # 12/11/2012: 49 (amd64/Linux)
# # (^ REASON UNKNOWN!)
# # 28/8/13: 60 (amd64/Linux)
# # (^ REASON UNKNOWN!)
# # 2014-09-10: 55 post-AMP-cleanup
# # 2014-10-08: 62 (jumps between 55 and 71 observed -- GC tipping point?)
# # 2014-10-13: 48 stricter seqDmdType
compiler_stats_num_field('bytes allocated',
[(platform('x86_64-apple-darwin'), 465653312, 10),
# prev: 510938976 (amd64/OS X):
# 2015-12-11: 465653312 (amd64/OS X) Update, bump tolerance to +/-10%
(wordsize(32), 239556572, 10),
# prev: 185669232 (x86/OSX)
# 2014-01-22: 211198056 (x86/Linux)
# 2014-09-03: 185242032 (Windows laptop)
# 2014-12-01: 203962148 (Windows laptop)
# 2016-04-06: 239556572 (x86/Linux)
(wordsize(64), 434278248, 10)]),
# prev: 360243576 (amd64/Linux)
# 19/10/2012: 447190832 (amd64/Linux) (-fPIC turned on)
# 19/10/2012: 392409984 (amd64/Linux) (-fPIC turned off)
# 2014-04-08: 362939272 (amd64/Linux) cumulation of various smaller improvements over recent commits
# 2014-10-08: 382056344 (amd64/Linux) stricter foldr2 488e95b
# 2015-10-28: 434278248 (amd64/Linux) emit Typeable at definition site
###################################
# deactivated for now, as this metric became too volatile recently
#
# compiler_stats_num_field('max_bytes_used',
# [(platform('x86_64-apple-darwin'), 25145320, 5),
# (wordsize(32), 11829000, 15),
# # 9651948 (x86/OSX)
# # 10290952 (windows)
# # 2013-02-10 11071060 (x86/Windows)
# # 2013-02-10: 11207828 (x86/OSX)
# # (some date): 11139444
# # 2013-11-13: 11829000 (x86/Windows, 64bit machine)
# (wordsize(64), 19296544, 15)]),
# # prev: 20486256 (amd64/OS X)
# # 30/08/2012: 17305600--20391920 (varies a lot)
# # 19/10/2012: 26882576 (-fPIC turned on)
# # 19/10/2012: 18619912 (-fPIC turned off)
# # 24/12/2012: 21657520 (perhaps gc sampling time wibbles?)
# # 10/01/2014: 25166280
# # 13/01/2014: 22646000 (mostly due to #8647)
# # 18/02/2014: 25002136 (call arity analysis changes)
# # 12/05/2014: 25002136 (specialisation and inlining changes)
# # 10/09/2014: 19296544, 10 (post-AMP-cleanup)
# # 14/09/2014: 19585456, 15 (adapt to Phab CI env)
only_ways(['normal']),
extra_hc_opts('-static'),
# Use `+RTS -G1` for more stable residency measurements. Note [residency].
extra_hc_opts('+RTS -G1 -RTS')
],
compile,
[''])
test('T3064',
[compiler_stats_num_field('peak_megabytes_allocated',# Note [residency]
[(wordsize(32), 28, 20),
# expected value: 14 (x86/Linux 28-06-2012):
# 2013-11-13: 18 (x86/Windows, 64bit machine)
# 2014-01-22: 23 (x86/Linux)
# 2014-12-22: 23 (x86/Linux) death to silent superclasses
# 2015-07-11 28 (x86/Linux, 64-bit machine) use +RTS -G1
(wordsize(64), 54, 20)]),
# (amd64/Linux): 18
# (amd64/Linux) 2012-02-07: 26
# (amd64/Linux) 2013-02-12: 23; increased range to 10%
# (amd64/Linux) 2013-04-03: 26
# (amd64/Linux) 2013-09-11: 30; result of AMP patch
# Increased range to 20%. peak-usage varies from 22 to 26,
# depending on whether the old .hi file exists
# (amd64/Linux) 2013-09-11: 37; better arity analysis (weird)
# (amd64/Linux) (09/09/2014): 42, AMP changes (larger interfaces, more loading)
# (amd64/Linux) 2014-10-13: 38: Stricter seqDmdType
# (amd64/Linux) 2014-12-22: 27: death to silent superclasses
# (amd64/Linux) 2015-01-22: 32: Varies from 30 to 34, at least here.
# (amd64/Linux) 2015-06-03: 54: use +RTS -G1
compiler_stats_num_field('bytes allocated',
[(wordsize(32), 153261024, 10),
# 2011-06-28: 56380288 (x86/Linux)
# 2012-10-30: 111189536 (x86/Windows)
# 2013-11-13: 146626504 (x86/Windows, 64bit machine)
# 2014-01-22: 162457940 (x86/Linux)
# 2014-12-01: 162457940 (Windows)
# 2014-12-22: 122836340 (Windows) Death to silent superclasses
# 2016-04-06: 153261024 (x86/Linux) probably wildcard refactor
(wordsize(64), 304344936, 5)]),
# (amd64/Linux) (28/06/2011): 73259544
# (amd64/Linux) (07/02/2013): 224798696
# (amd64/Linux) (02/08/2013): 236404384, increase from roles
# (amd64/Linux) (11/09/2013): 290165632, increase from AMP warnings
# (amd64/Linux) (22/11/2013): 308300448, GND via Coercible and counters for constraints solving
# (amd64/Linux) (02/12/2013): 329795912, Coercible refactor
# (amd64/Linux) (11/02/2014): 308422280, optimize Coercions in simpleOptExpr
# (amd64/Linux) (23/05/2014): 324022680, unknown cause
# (amd64/Linux) (2014-07-17): 332702112, general round of updates
# (amd64/Linux) (2014-08-29): 313638592, w/w for INLINABLE things
# (amd64/Linux) (09/09/2014): 407416464, AMP changes (larger interfaces, more loading)
# (amd64/Linux) (14/09/2014): 385145080, BPP changes (more NoImplicitPrelude in base)
# (amd64/Linux) (10/12/2014): 363103840, improvements in constraint solver
# (Mac) (18/12/2014): 350418600, improvements to flattener
# (amd64/Linux) (22/12/2014): 243670824, Ha! Death to superclass constraints, makes
# much less code for Monad instances
# (amd64/Linux) (01/12/2015): 264952256, Regression due to Simon's wildcard refactor
# Tracked as #11151.
# (amd64/Linux) (11/12/2015): 304344936, Regression due to TypeInType
# Tracked as #11196
###################################
# deactivated for now, as this metric became too volatile recently
#
# compiler_stats_num_field('max_bytes_used',
# [(wordsize(32), 11202304, 20),
# # 2011-06-28: 2247016 (x86/Linux) (28/6/2011):
# #(some date): 5511604
# # 2013-11-13: 7218200 (x86/Windows, 64bit machine)
# # 2014-04-04: 11202304 (x86/Windows, 64bit machine)
# (wordsize(64), 13251728, 20)]),
# # (amd64/Linux, intree) (28/06/2011): 4032024
# # (amd64/Linux, intree) (07/02/2013): 9819288
# # (amd64/Linux) (14/02/2013): 8687360
# # (amd64/Linux) (18/02/2013): 9397488
# # (amd64/Linux) (02/08/2013): 10742536, increase from roles
# # (amd64/Linux) (19/08/2013): 9211816, decrease apparently from better eta reduction
# # (amd64/Linux) (11/09/2013): 12000480, increase from AMP warnings
# # 933cdf15a2d85229d3df04b437da31fdfbf4961f
# # (amd64/Linux) (22/11/2013): 16266992, GND via Coercible and counters for constraints solving
# # (amd64/Linux) (12/12/2013): 19821544, better One shot analysis
# # (amd64/Linux) (09/09/2014): 24357392, AMP changes (larger interfaces, more loading)
# # (amd64/Linux) (14/09/2014): 16053888, BPP changes (more NoImplicitPrelude in base)
# # (amd64/Linux) (19/09/2014): 18744992, unknown
# # (amd64/Linux) 2014-10-13: 13251728, Stricter seqDmdType
only_ways(['normal']),
# Use `+RTS -G1` for more stable residency measurements. Note [residency].
extra_hc_opts('+RTS -G1 -RTS')
],
compile,
[''])
test('T4007',
normal,
run_command,
['$MAKE -s --no-print-directory T4007'])
test('T5030',
[compiler_stats_num_field('bytes allocated',
[(wordsize(32), 345668088, 10),
# previous: 196457520
# 2012-10-08: 259547660 (x86/Linux, new codegen)
# 2013-11-21: 198573456 (x86 Windows, 64 bit machine)
# 2014-12-10: 227205560 constraint solver got worse again; more aggressive solving
# of family-applications leads to less sharing, I think
# 2015-07-11: 201882912 reason unknown
# 2016-04-06: 345668088 likely TypeInType
(wordsize(64), 653710960, 10)]),
# Previously 530000000 (+/- 10%)
# 17/1/13: 602993184 (x86_64/Linux)
# (new demand analyser)
# 2013-06-08 538467496 (x86_64/Linux)
# ^ reason unknown
# 2013-08-02 454498592 (amd64/Linux)
# decrease from more aggressive coercion optimisations from roles
# 2013-11-12 397672152 (amd64/Linux)
# big decrease following better CSE and arity
# 2014-07-17 409314320 (amd64/Linux)
# general round of updates
# 2014-09-10 385152728 post-AMP-cleanup
# 2014-12-08 340969128 constraint solver perf improvements (esp kick-out)
# 2014-12-10 449042120 constraint solver got worse again; more aggressive solving
# of family-applications leads to less sharing, I think
# 2015-03-17 403932600 tweak to solver algorithm
# 2015-12-11 653710960 TypeInType (see #11196)
only_ways(['normal'])
],
compile,
['-freduction-depth=300'])
test('T5631',
[compiler_stats_num_field('bytes allocated',
[(wordsize(32), 570137436, 10),
# expected value: 392904228 (x86/Linux)
# 2014-04-04: 346389856 (x86 Windows, 64 bit machine)
# 2014-12-01: 390199244 (Windows laptop)
# 2016-04-06: 570137436 (amd64/Linux) many reasons
(wordsize(64), 1124068664, 5)]),
# expected value: 774595008 (amd64/Linux):
# expected value: 735486328 (amd64/Linux) 2012/12/12:
# expected value: 690742040 (amd64/Linux) Call Arity improvements
# 2014-09-09: 739704712 (amd64/Linux) AMP changes
# 2014-11-04: 776121120 (amd64/Linux) new-flatten-skolems
# 2015-06-01: 812288344 (amd64/Linux) unknown cause
# 2015-12-11: 1128828928 (amd64/Linux) TypeInType (see #11196)
# 2015-12-21: 1198327544 (Mac) TypeApplications (will fix with #11196)
# 2015-03-18: 1124068664 (Mac) optimize Unify & zonking
only_ways(['normal'])
],
compile,
[''])
test('parsing001',
[compiler_stats_num_field('bytes allocated',
[(wordsize(32), 274000576, 10),
(wordsize(64), 587079016, 5)]),
# expected value: 587079016 (amd64/Linux)
only_ways(['normal']),
],
compile_fail, [''])
test('T783',
[ only_ways(['normal']), # no optimisation for this one
# expected value: 175,569,928 (x86/Linux)
compiler_stats_num_field('bytes allocated',
[(wordsize(32), 249332816, 5),
# 2012-10-08: 226907420 (x86/Linux)
# 2013-02-10: 329202116 (x86/Windows)
# 2013-02-10: 338465200 (x86/OSX)
# 2014-04-04: 319179104 (x86 Windows, 64 bit machine)
# 2014-09-03: 223377364 (Windows) better specialisation, raft of core-to-core optimisations
# 2014-12-22: 235002220 (Windows) not sure why
# 2016-04-06: 249332816 (x86/Linux, 64-bit machine)
(wordsize(64), 488592288, 10)]),
# prev: 349263216 (amd64/Linux)
# 07/08/2012: 384479856 (amd64/Linux)
# 29/08/2012: 436927840 (amd64/Linux)
# 12/11/2012: 640324528 (amd64/Linux)
# (OldCmm removed: not sure why this got worse, the
# other perf tests remained about the same)
# 18/10/2013: 734038080 (amd64/Linux)
# (fix for #8456)
# 24/10/2013: 654804144 (amd64/Linux)
# (fix previous fix for #8456)
# 2014-07-17: 640031840 (amd64/Linux)
# (general round of updates)
# 2014-08-29: 441932632 (amd64/Linux)
# (better specialisation, raft of core-to-core optimisations)
# 2014-08-29: 719814352 (amd64/Linux)
# (changed order of cmm block causes analyses to allocate much more,
# but the changed order is slighly better in terms of runtime, and
# this test seems to be an extreme outlier.)
# 2015-05-16: 548288760 (amd64/Linux)
# (improved sequenceBlocks in nativeCodeGen, #10422)
# 2015-08-07: 470738808 (amd64/Linux)
# (simplifying the switch plan code path for simple checks, #10677)
# 2015-08-28: 526230456 (amd64/Linux)
# (D757: Emit Typeable instances at site of type definition)
# 2015-12-04: 1134085384 (amd64/Linux)
# (D1535: Major overhaul of pattern match checker, #11162)
# 2016-02-03: 488592288 (amd64/Linux)
# (D1795: Another overhaul of pattern match checker, #11374)
extra_hc_opts('-static')
],
compile,[''])
test('T5321Fun',
[ only_ways(['normal']), # no optimisation for this one
compiler_stats_num_field('bytes allocated',
[(wordsize(32), 279922360, 10),
# prev: 300000000
# 2012-10-08: 344416344 x86/Linux
# (increase due to new codegen)
# 2014-09-03: 299656164 (specialisation and inlining)
# 10/12/2014: 206406188 # Improvements in constraint solver
# 2016-04-06: 279922360 x86/Linux
(wordsize(64), 565883176, 10)])
# prev: 585521080
# 29/08/2012: 713385808 # (increase due to new codegen)
# 15/05/2013: 628341952 # (reason for decrease unknown)
# 24/06/2013: 694019152 # (reason for re-increase unknown)
# 12/05/2014: 614409344 # (specialisation and inlining changes)
# 10/09/2014: 601629032 # post-AMP-cleanup
# 06/11/2014: 541287000 # Simon's flat-skol changes to the constraint solver
# 10/12/2014: 408110888 # Improvements in constraint solver
# 16/12/2014: 429921312 # Flattener parameterized over roles
# 10/08/2015: 509921312
# (undefined now takes an implicit parameter and GHC -O0 does
# not recognize that the application is bottom)
# 11/12/2015: 565883176 # TypeInType (see #11196)
],
compile,[''])
test('T5321FD',
[ only_ways(['normal']), # no optimisation for this one
compiler_stats_num_field('bytes allocated',
[(wordsize(32), 250757460, 10),
# prev: 213380256
# 2012-10-08: 240302920 (x86/Linux)
# (increase due to new codegen)
# 2014-07-31: 211699816 (Windows) (-11%)
# (due to better optCoercion, 5e7406d9, #9233)
# 2016-04-06: 250757460 (x86/Linux)
(wordsize(64), 532365376, 10)])
# prev: 418306336
# 29/08/2012: 492905640
# (increase due to new codegen)
# 15/05/2013: 406039584
# (reason for decrease unknown)
# 08/06/2013: 476497048
# (reason for increase unknown)
# before 2014-07-17: 441997096
# (with -8%, still in range, hence cause not known)
# 2014-07-17: 426960992 (-11% of previous value)
# (due to better optCoercion, 5e7406d9, #9233)
# 2014-10-08 410895536
# (various changes; biggest improvements due to 949ad67 and FastString package ids)
# 2015-08-10: 470895536
# (undefined now takes an implicit parameter and GHC -O0 does
# not recognize that the application is bottom)
# 2015-10-28: 532365376
# D757: emit Typeable instances at site of type definition
],
compile,[''])
test('T5642',
[ only_ways(['normal']),
normal,
compiler_stats_num_field('bytes allocated',
[(wordsize(32), 462677300, 10),
# sample from x86/Linux
# prev: 650000000
# 2014-09-03: 753045568
# 2014-12-10: 641085256 Improvements in constraints solver
# 2016-04-06: 462677300
(wordsize(64), 950004816, 10)])
# prev: 1300000000
# 2014-07-17: 1358833928 (general round of updates)
# 2014-08-07: 1402242360 (caused by 1fc60ea)
# Watch out for:
# 23/05/2014: 1452688392 (More aggressive specialisation means we get
# specialised copies of imported functions that
# are ultimately discarded by trimAutoRules
# It's a bizarre program with LOTS of data types)
# 2014-09-10: 1536924976 post-AMP-cleanup
# 2014-12-10: 1282916024 Improvements in constraints solver
# 2015-10-28: 1412808976 Emit Typeable at definition site
# 2015-11-22: 1071915072 Use TypeLits in the metadata encoding
# 2016-02-08: 950004816 Pattern match checker re-rework
],
compile,['-O'])
test('T5837',
[ only_ways(['normal']),
compiler_stats_num_field('bytes allocated',
[(wordsize(32), 24199320, 10),
# 40000000 (x86/Linux)
# 2013-11-13: 45520936 (x86/Windows, 64bit machine)
# 2014-09-03: 37096484 (Windows laptop, w/w for INLINABLE things
# 2014-12-01: 135914136 (Windows laptop, regression see below)
# 2014-12-08: 115905208 Constraint solver perf improvements (esp kick-out)
# 2016-04-06: 24199320 (x86/Linux, 64-bit machine) TypeInType
(wordsize(64), 48507272, 10)])
# sample: 3926235424 (amd64/Linux, 15/2/2012)
# 2012-10-02 81879216
# 2012-09-20 87254264 amd64/Linux
# 2013-09-18 90587232 amd64/Linux
# 2013-11-21 86795752 amd64/Linux, GND via Coercible and counters
# for constraints solving
# 2014-08-29 73639840 amd64/Linux, w/w for INLINABLE things
# 2014-10-08 73639840 amd64/Linux, Burning Bridges and other small changes
# 2014-11-06 271028976 Linux, Accept big regression;
# See Note [An alternative story for the inert substitution] in TcFlatten
# 2014-12-08 234790312 Constraint solver perf improvements (esp kick-out)
# 2014-12-16 231155640 Mac Flattener parameterized over roles;
# some optimization
# 2015-03-17 53424304 Mac Better depth checking; fails earlier
# 2015-06-09 38834096 Better "improvement"; I'm not sure whey it improves things
# 2015-12-11 43877520 amd64/Linux, TypeInType (see #11196)
# 2016-03-18 48507272 Mac, accept small regression in exchange
# for other optimisations
],
compile_fail,['-freduction-depth=50'])
test('T6048',
[ only_ways(['optasm']),
compiler_stats_num_field('bytes allocated',
[(wordsize(32), 55701280, 10),
# prev: 38000000 (x86/Linux)
# 2012-10-08: 48887164 (x86/Linux)
# 2014-04-04: 62618072 (x86 Windows, 64 bit machine)
# 2014-09-03: 56315812 (x86 Windows, w/w for INLINEABLE)
# 2014-12-01: 49987836 (x86 Windows)
# 2016-04-06: 55701280 (x86/Linux, 64-bit machine)
(wordsize(64), 108225624, 12)])
# 18/09/2012 97247032 amd64/Linux
# 16/01/2014 108578664 amd64/Linux (unknown, likely foldl-via-foldr)
# 18/01/2014 95960720 amd64/Linux Call Arity improvements
# 28/02/2014 105556793 amd64/Linux (unknown, tweak in base/4d9e7c9e3 resulted in change)
# 05/03/2014 110646312 amd64/Linux Call Arity became more elaborate
# 14/07/2014 125431448 amd64/Linux unknown reason. Even worse in GHC-7.8.3. *shurg*
# 29/08/2014 108354472 amd64/Linux w/w for INLINABLE things
# 14/09/2014 88186056 amd64/Linux BPP part1 change (more NoImplicitPreludes in base)
# 08/01/2014 95946688 amd64/Linux Mostly 4c834fd. Occasional spikes to 103822120!
# 11/03/2016 108225624 amd64/Linux unknown reason sadly; likely gradual creep.
],
compile,[''])
test('T9020',
[ only_ways(['optasm']),
compiler_stats_num_field('bytes allocated',
[(wordsize(32), 343005716, 10),
# Original: 381360728
# 2014-07-31: 343005716 (Windows) (general round of updates)
(wordsize(64), 698401736, 10)])
# prev: 795469104
# 2014-07-17: 728263536 (general round of updates)
# 2014-09-10: 785871680 post-AMP-cleanup
# 2014-11-03: 680162056 Further Applicative and Monad adjustments
# 2015-10-21: 786189008 Make stronglyConnCompFromEdgedVertices deterministic
# 2016-01-26: 698401736 improvement from using ExpTypes instead of ReturnTvs
# 2016-04-06: 852298336 Refactoring of CSE #11781
# 2016-04-06: 698401736 Use thenIO in Applicative IO
],
compile,[''])
test('T9675',
[ only_ways(['optasm']),
compiler_stats_num_field('max_bytes_used', # Note [residency]
[(wordsize(64), 38776008, 15),
# 2014-10-13 29596552
# 2014-10-13 26570896 seq the DmdEnv in seqDmdType as well
# 2014-10-13 18582472 different machines giving different results..
# 2014-10-13 22220552 use the mean
# 2015-06-21 28056344 switch to `+RTS -G1`, tighten bound to 15%
# 2015-10-28 23776640 emit Typeable at definition site
# 2015-12-11 30837312 TypeInType (see #11196)
# 2016-04-14 38776008 Final demand analyzer run
(wordsize(32), 18043224, 15)
# 2015-07-11 15341228 (x86/Linux, 64-bit machine) use +RTS -G1
# 2016-04-06 18043224 (x86/Linux, 64-bit machine)
]),
compiler_stats_num_field('peak_megabytes_allocated', # Note [residency]
[(wordsize(64), 144, 15),
# 2014-10-13 66
# 2014-10-13 58 seq the DmdEnv in seqDmdType as well
# 2014-10-13 49 different machines giving different results...
# 2014-10-13 53 use the mean
# 2015-06-15 44 reduced for some reason
# 2015-06-21 105 switch to `+RTS -G1`
# 2015-12-04 88 new pattern checker (D1535)
# 2015-12-11 113 TypeInType (see #11196)
# 2016-04-14 144 Final demand analyzer run
(wordsize(32), 56, 15)
# 2015-07-11 56 (x86/Linux, 64-bit machine) use +RTS -G1
]),
compiler_stats_num_field('bytes allocated',
[(wordsize(64), 608284152, 10)
# 2014-10-13 544489040
# 2015-10-28 608284152 emit Typeable at definition site
,(wordsize(32), 279480696, 10)
# 2015-07-11 279480696 (x86/Linux, 64-bit machine) use +RTS -G1
]),
# Use `+RTS -G1` for more stable residency measurements. Note [residency].
extra_hc_opts('+RTS -G1 -RTS')
],
compile,
[''])
test('T9872a',
[ only_ways(['normal']),
compiler_stats_num_field('bytes allocated',
[(wordsize(64), 3352882080, 5),
# 2014-12-10 5521332656 Initally created
# 2014-12-16 5848657456 Flattener parameterized over roles
# 2014-12-18 2680733672 Reduce type families even more eagerly
# 2015-12-11 3581500440 TypeInType (see #11196)
# 2016-04-07 3352882080 CSE improvements
(wordsize(32), 1740903516, 5)
# was 1325592896
# 2016-04-06 1740903516 x86/Linux
]),
],
compile_fail,
[''])
test('T9872b',
[ only_ways(['normal']),
compiler_stats_num_field('bytes allocated',
[(wordsize(64), 4600233488, 5),
# 2014-12-10 6483306280 Initally created
# 2014-12-16 6892251912 Flattener parameterized over roles
# 2014-12-18 3480212048 Reduce type families even more eagerly
# 2015-12-11 5199926080 TypeInType (see #11196)
# 2016-02-08 4918990352 Improved a bit by tyConRolesRepresentational
# 2016-04-06: 4600233488 Refactoring of CSE #11781
(wordsize(32), 2422750696, 5)
# was 1700000000
# 2016-04-06 2422750696 x86/Linux
]),
],
compile_fail,
[''])
test('T9872c',
[ only_ways(['normal']),
compiler_stats_num_field('bytes allocated',
[(wordsize(64), 4306667256, 5),
# 2014-12-10 5495850096 Initally created
# 2014-12-16 5842024784 Flattener parameterized over roles
# 2014-12-18 2963554096 Reduce type families even more eagerly
# 2015-12-11 4723613784 TypeInType (see #11196)
# 2016-02-08 4454071184 Improved a bit by tyConRolesRepresentational
# 2016-04-06: 4306667256 Refactoring of CSE #11781
(wordsize(32), 2257242896, 5)
# was 1500000000
# 2016-04-06 2257242896
]),
],
compile_fail,
[''])
test('T9872d',
[ only_ways(['normal']),
compiler_stats_num_field('bytes allocated',
[(wordsize(64), 506691240, 5),
# 2014-12-18 796071864 Initally created
# 2014-12-18 739189056 Reduce type families even more eagerly
# 2015-01-07 687562440 TrieMap leaf compression
# 2015-03-17 726679784 tweak to solver; probably flattens more
# 2016-02-08 534693648 Improved a bit by tyConRolesRepresentational
# 2016-03-18 506691240 optimize Unify & zonking
(wordsize(32), 264566040, 5)
# some date 328810212
# 2015-07-11 350369584
# 2015-12-11 566134504 TypeInType; see #11196
# 2016-04-06 264566040 x86/Linux, no idea, opened #11800
]),
],
compile,
[''])
test('T9961',
[ only_ways(['normal']),
compiler_stats_num_field('bytes allocated',
[(wordsize(64), 568526784, 5),
# 2015-01-12 807117816 Initally created
# 2015-spring 772510192 Got better
# 2015-05-22 663978160 Fix for #10370 improves it more
# 2015-10-28 708680480 x86_64/Linux Emit Typeable at definition site
# 2015-12-17 745044392 x86_64/Darwin Creep upwards
# 2016-03-20 519436672 x64_64/Linux Don't use build desugaring for large lists (#11707)
# 2016-03-24 568526784 x64_64/Linux Add eqInt* variants (#11688)
(wordsize(32), 275264188, 5)
# was 375647160
# 2016-04-06 275264188 x86/Linux
]),
],
compile,
['-O'])
test('T9233',
[ only_ways(['normal']),
compiler_stats_num_field('bytes allocated',
[(wordsize(64), 1066246248, 5),
# 2015-08-04 999826288 initial value
# 2016-04-14 1066246248 Final demand analyzer run
(wordsize(32), 515672240, 5) # Put in your value here if you hit this
# 2016-04-06 515672240 (x86/Linux) initial value
]),
extra_clean(['T9233a.hi', 'T9233a.o'])
],
multimod_compile,
['T9233', '-v0 -O2 -fno-spec-constr'])
test('T10370',
[ only_ways(['optasm']),
compiler_stats_num_field('max_bytes_used', # Note [residency]
[(wordsize(64), 28256896, 15),
# 2015-10-22 19548720
# 2016-02-24 22823976 Changing Levity to RuntimeRep; not sure why this regresses though, even after some analysis
# 2016-04-14 28256896 final demand analyzer run
(wordsize(32), 11371496, 15),
# 2015-10-22 11371496
]),
compiler_stats_num_field('peak_megabytes_allocated', # Note [residency]
[(wordsize(64), 101, 15),
# 2015-10-22 76
# 2016-04-14 101 final demand analyzer run
(wordsize(32), 39, 15),
# 2015-10-22 39
]),
# Use `+RTS -G1` for more stable residency measurements. Note [residency].
extra_hc_opts('+RTS -G1 -RTS')
],
compile,
[''])
|
