File: run_tests.py

package info (click to toggle)
lapackpp 2024.10.26-1
  • links: PTS, VCS
  • area: main
  • in suites: forky, sid, trixie
  • size: 6,500 kB
  • sloc: cpp: 80,181; ansic: 27,660; python: 4,838; xml: 182; perl: 99; makefile: 53; sh: 23
file content (896 lines) | stat: -rwxr-xr-x 38,816 bytes parent folder | download
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
#!/usr/bin/env python3
#
# Copyright (c) 2017-2023, University of Tennessee. All rights reserved.
# SPDX-License-Identifier: BSD-3-Clause
# This program is free software: you can redistribute it and/or modify it under
# the terms of the BSD 3-Clause license. See the accompanying LICENSE file.
#
# Example usage:
# help
#     ./run_tests.py -h
#
# run everything with default sizes
# output is redirected; summary information is printed on stderr
#     ./run_tests.py > output.txt
#
# run LU (gesv, getrf, getri, ...), Cholesky (posv, potrf, potri, ...)
# with single, double and default sizes
#     ./run_tests.py --lu --chol --type s,d
#
# run getrf, potrf with small, medium sizes
#     ./run_tests.py -s -m getrf potrf

from __future__ import print_function

import sys
import os
import re
import argparse
import subprocess
import xml.etree.ElementTree as ET
import io
import time

# ------------------------------------------------------------------------------
# command line arguments
parser = argparse.ArgumentParser()

group_test = parser.add_argument_group( 'test' )
group_test.add_argument( '-t', '--test', action='store',
    help='test command to run, e.g., --test "mpirun -np 4 ./test"; default "%(default)s"',
    default='./tester' )
group_test.add_argument( '--xml', help='generate report.xml for jenkins' )
group_test.add_argument( '--dry-run', action='store_true', help='print commands, but do not execute them' )
group_test.add_argument( '--start',   action='store', help='routine to start with, helpful for restarting', default='' )
group_test.add_argument( '-x', '--exclude', action='append', help='routines to exclude; repeatable', default=[] )

group_size = parser.add_argument_group( 'matrix dimensions (default is medium)' )
group_size.add_argument( '--quick',  action='store_true', help='run quick "sanity check" of few, small tests' )
group_size.add_argument( '--xsmall', action='store_true', help='run x-small tests' )
group_size.add_argument( '--small',  action='store_true', help='run small tests' )
group_size.add_argument( '--medium', action='store_true', help='run medium tests' )
group_size.add_argument( '--large',  action='store_true', help='run large tests' )
group_size.add_argument( '--square', action='store_true', help='run square (m = n = k) tests', default=False )
group_size.add_argument( '--tall',   action='store_true', help='run tall (m > n) tests', default=False )
group_size.add_argument( '--wide',   action='store_true', help='run wide (m < n) tests', default=False )
group_size.add_argument( '--mnk',    action='store_true', help='run tests with m, n, k all different', default=False )
group_size.add_argument( '--dim',    action='store',      help='explicitly specify size', default='' )

group_cat = parser.add_argument_group( 'category (default is all)' )
categories = [
    group_cat.add_argument( '--lu',            action='store_true', help='run LU tests' ),
    group_cat.add_argument( '--gb',            action='store_true', help='run GB tests' ),
    group_cat.add_argument( '--gt',            action='store_true', help='run GT tests' ),
    group_cat.add_argument( '--chol',          action='store_true', help='run Cholesky tests' ),
    group_cat.add_argument( '--sysv',          action='store_true', help='run symmetric indefinite (Bunch-Kaufman) tests' ),
    group_cat.add_argument( '--rook',          action='store_true', help='run symmetric indefinite (rook) tests' ),
    group_cat.add_argument( '--aasen',         action='store_true', help='run symmetric indefinite (Aasen) tests' ),
    group_cat.add_argument( '--hesv',          action='store_true', help='run hermetian tests (FIXME more informationhere)' ),
    group_cat.add_argument( '--least-squares', action='store_true', help='run least squares tests' ),
    group_cat.add_argument( '--qr',            action='store_true', help='run QR tests' ),
    group_cat.add_argument( '--lq',            action='store_true', help='run LQ tests' ),
    group_cat.add_argument( '--ql',            action='store_true', help='run QL tests' ),
    group_cat.add_argument( '--rq',            action='store_true', help='run RQ tests' ),
    group_cat.add_argument( '--syev',          action='store_true', help='run symmetric eigenvalues tests' ),
    group_cat.add_argument( '--sygv',          action='store_true', help='run generalized symmetric eigenvalues tests' ),
    group_cat.add_argument( '--geev',          action='store_true', help='run non-symmetric eigenvalues tests' ),
    group_cat.add_argument( '--svd',           action='store_true', help='run svd tests' ),
    group_cat.add_argument( '--aux',           action='store_true', help='run auxiliary tests' ),
    group_cat.add_argument( '--aux-house',     action='store_true', help='run auxiliary Householder reflector tests' ),
    group_cat.add_argument( '--aux-givens',    action='store_true', help='run auxiliary Givens rotations tests' ),
    group_cat.add_argument( '--aux-norm',      action='store_true', help='run auxiliary norm tests' ),
    group_cat.add_argument( '--blas',          action='store_true', help='run additional BLAS tests' ),
]
# map category objects to category names: ['lu', 'chol', ...]
categories = list( map( lambda x: x.dest, categories ) )

group_target = parser.add_argument_group( 'target' )
group_target.add_argument( '--host',          action='store_true', help='run all CPU host routines' ),
group_target.add_argument( '--device',        action='store_true', help='run all GPU device routines' ),

group_opt = parser.add_argument_group( 'options' )
# BLAS and LAPACK
# Empty defaults (check, ref, etc.) use the default in test.cc.
group_opt.add_argument( '--type',   action='store', help='default=%(default)s', default='s,d,c,z' )
group_opt.add_argument( '--layout', action='store', help='default=%(default)s', default='c,r' )
group_opt.add_argument( '--transA', action='store', help='default=%(default)s', default='n,t,c' )
group_opt.add_argument( '--transB', action='store', help='default=%(default)s', default='n,t,c' )
group_opt.add_argument( '--trans',  action='store', help='default=%(default)s', default='n,t,c' )
group_opt.add_argument( '--uplo',   action='store', help='default=%(default)s', default='l,u' )
group_opt.add_argument( '--diag',   action='store', help='default=%(default)s', default='n,u' )
group_opt.add_argument( '--side',   action='store', help='default=%(default)s', default='l,r' )
group_opt.add_argument( '--alpha',  action='store', help='default=%(default)s', default='' )
group_opt.add_argument( '--beta',   action='store', help='default=%(default)s', default='' )
group_opt.add_argument( '--incx',   action='store', help='default=%(default)s', default='1,2,-1,-2' )
group_opt.add_argument( '--incy',   action='store', help='default=%(default)s', default='1,2,-1,-2' )
group_opt.add_argument( '--align',  action='store', help='default=%(default)s', default='32' )
group_opt.add_argument( '--check',  action='store', help='default=y', default='' )  # default in test.cc
group_opt.add_argument( '--ref',    action='store', help='default=y', default='' )  # default in test.cc
group_opt.add_argument( '--verbose', action='store', help='default=0', default='' )  # default in test.cc

# LAPACK only
group_opt.add_argument( '--itype',  action='store', help='default=%(default)s', default='1,2,3' )
group_opt.add_argument( '--factored', action='store', help='default=%(default)s', default='f,n,e' )
group_opt.add_argument( '--equed',  action='store', help='default=%(default)s', default='n,r,c,b' )
group_opt.add_argument( '--pivot',  action='store', help='default=%(default)s', default='v,t,b' )
group_opt.add_argument( '--direction', action='store', help='default=%(default)s', default='f,b' )
group_opt.add_argument( '--storev', action='store', help='default=%(default)s', default='c,r' )
group_opt.add_argument( '--norm',   action='store', help='default=%(default)s', default='max,1,inf,fro' )
group_opt.add_argument( '--ijob',   action='store', help='default=%(default)s', default='0:5:1' )
group_opt.add_argument( '--jobz',   action='store', help='default=%(default)s', default='n,v' )
group_opt.add_argument( '--jobvl',  action='store', help='default=%(default)s', default='n,v' )
group_opt.add_argument( '--jobvr',  action='store', help='default=%(default)s', default='n,v' )
group_opt.add_argument( '--jobvs',  action='store', help='default=%(default)s', default='n,v' )
group_opt.add_argument( '--jobu',   action='store', help='default=%(default)s', default='n,s,o,a' )
group_opt.add_argument( '--jobvt',  action='store', help='default=%(default)s', default='n,s,o,a' )
group_opt.add_argument( '--balanc', action='store', help='default=%(default)s', default='n,p,s,b' )
group_opt.add_argument( '--sort',   action='store', help='default=%(default)s', default='n,s' )
group_opt.add_argument( '--select', action='store', help='default=%(default)s', default='n,s' )
group_opt.add_argument( '--sense',  action='store', help='default=%(default)s', default='n,e,v,b' )
group_opt.add_argument( '--vect',   action='store', help='default=%(default)s', default='n,v' )
group_opt.add_argument( '--l',      action='store', help='default=%(default)s', default='0,100' )
group_opt.add_argument( '--ka',     action='store', help='default=%(default)s', default='20,100' )
group_opt.add_argument( '--kb',     action='store', help='default=%(default)s', default='20,100' )
group_opt.add_argument( '--kd',     action='store', help='default=%(default)s', default='20,100' )
group_opt.add_argument( '--kl',     action='store', help='default=%(default)s', default='20,100' )
group_opt.add_argument( '--ku',     action='store', help='default=%(default)s', default='20,100' )
group_opt.add_argument( '--vl',     action='store', help='default=%(default)s', default='-inf,0' )
group_opt.add_argument( '--vu',     action='store', help='default=%(default)s', default='inf' )
group_opt.add_argument( '--il',     action='store', help='default=%(default)s', default='10' )
group_opt.add_argument( '--iu',     action='store', help='default=%(default)s', default='-1,100' )
group_opt.add_argument( '--nb',     action='store', help='default=%(default)s', default='64' )
group_opt.add_argument( '--matrixtype', action='store', help='default=%(default)s', default='g,l,u' )

parser.add_argument( 'tests', nargs=argparse.REMAINDER )
opts = parser.parse_args()

for t in opts.tests:
    if (t.startswith('--')):
        print( 'Error: option', t, 'must come before any routine names' )
        print( 'usage:', sys.argv[0], '[options]', '[routines]' )
        print( '      ', sys.argv[0], '--help' )
        exit(1)

# by default, run medium sizes
if (not (opts.quick or opts.xsmall or opts.small or opts.medium or opts.large)):
    opts.medium = True

# by default, run all shapes
if (not (opts.square or opts.tall or opts.wide or opts.mnk)):
    opts.square = True
    opts.tall   = True
    opts.wide   = True
    opts.mnk    = True

# By default, run both host and device.
if (not opts.host and not opts.device):
    opts.host   = True
    opts.device = True

# By default, or if specific test routines given, enable all categories
# to get whichever has the routines.
if (opts.tests or not any( map( lambda c: opts.__dict__[ c ], categories ))):
    for c in categories:
        opts.__dict__[ c ] = True

start_routine = opts.start

# ------------------------------------------------------------------------------
# parameters
# begin with space to ease concatenation

# if given, use explicit dim
dim = ' --dim ' + opts.dim if (opts.dim) else ''
n        = dim
tall     = dim
wide     = dim
mn       = dim
mnk      = dim
nk_tall  = dim
nk_wide  = dim
nk       = dim

if (not opts.dim):
    if (opts.quick):
        n        = ' --dim 100'
        tall     = ' --dim 100x50'  # 2:1
        wide     = ' --dim 50x100'  # 1:2
        mnk      = ' --dim 25x50x75'
        nk_tall  = ' --dim 1x100x50'  # 2:1
        nk_wide  = ' --dim 1x50x100'  # 1:2
        opts.incx  = '1,-1'
        opts.incy  = '1,-1'
        opts.batch = '10'
        opts.l     = '0,20,50'
        opts.nb    = '16'

    if (opts.xsmall):
        n       += ' --dim 10'
        tall    += ' --dim 20x10'
        wide    += ' --dim 10x20'
        mnk     += ' --dim 10x15x20 --dim 15x10x20' \
                +  ' --dim 10x20x15 --dim 15x20x10' \
                +  ' --dim 20x10x15 --dim 20x15x10'
        nk_tall += ' --dim 1x20x10'
        nk_wide += ' --dim 1x10x20'
        # tpqrt, tplqt needs small l, nb <= min( m, n )
        if (opts.l == parser.get_default('l')):
            opts.l = '0,5,100'
        if (opts.nb == parser.get_default('nb')):
            opts.nb = '8,64'
        if (opts.ka == parser.get_default('ka')):
            opts.ka = '5'
        if (opts.kb == parser.get_default('kb')):
            opts.kb = '5'
        if (opts.kd == parser.get_default('kd')):
            opts.kd = '5'

    if (opts.small):
        n       += ' --dim 25:100:25'
        tall    += ' --dim 50:200:50x25:100:25'  # 2:1
        wide    += ' --dim 25:100:25x50:200:50'  # 1:2
        mnk     += ' --dim 25x50x75 --dim 50x25x75' \
                +  ' --dim 25x75x50 --dim 50x75x25' \
                +  ' --dim 75x25x50 --dim 75x50x25'
        nk_tall += ' --dim 1x50:200:50x25:100:25'
        nk_wide += ' --dim 1x25:100:25x50:200:50'

    if (opts.medium):
        n       += ' --dim 100:500:100'
        tall    += ' --dim 200:1000:200x100:500:100'  # 2:1
        wide    += ' --dim 100:500:100x200:1000:200'  # 1:2
        mnk     += ' --dim 100x300x600 --dim 300x100x600' \
                +  ' --dim 100x600x300 --dim 300x600x100' \
                +  ' --dim 600x100x300 --dim 600x300x100'
        nk_tall += ' --dim 1x200:1000:200x100:500:100'
        nk_wide += ' --dim 1x100:500:100x200:1000:200'

    if (opts.large):
        n       += ' --dim 1000:5000:1000'
        tall    += ' --dim 2000:10000:2000x1000:5000:1000'  # 2:1
        wide    += ' --dim 1000:5000:1000x2000:10000:2000'  # 1:2
        mnk     += ' --dim 1000x3000x6000 --dim 3000x1000x6000' \
                +  ' --dim 1000x6000x3000 --dim 3000x6000x1000' \
                +  ' --dim 6000x1000x3000 --dim 6000x3000x1000'
        nk_tall += ' --dim 1x2000:10000:2000x1000:5000:1000'
        nk_wide += ' --dim 1x1000:5000:1000x2000:10000:2000'

    mn  = ''
    nk  = ''
    if (opts.square):
        mn = n
        nk = n
    if (opts.tall):
        mn += tall
        nk += nk_tall
    if (opts.wide):
        mn += wide
        nk += nk_wide
    if (opts.mnk):
        mnk = mn + mnk
    else:
        mnk = mn
# end

# BLAS and LAPACK
dtype  = ' --type '   + opts.type   if (opts.type)   else ''
layout = ' --layout ' + opts.layout if (opts.layout) else ''
transA = ' --transA ' + opts.transA if (opts.transA) else ''
transB = ' --transB ' + opts.transB if (opts.transB) else ''
trans  = ' --trans '  + opts.trans  if (opts.trans)  else ''
uplo   = ' --uplo '   + opts.uplo   if (opts.uplo)   else ''
diag   = ' --diag '   + opts.diag   if (opts.diag)   else ''
side   = ' --side '   + opts.side   if (opts.side)   else ''
a      = ' --alpha '  + opts.alpha  if (opts.alpha)  else ''
ab     = a+' --beta ' + opts.beta   if (opts.beta)   else a
incx   = ' --incx '   + opts.incx   if (opts.incx)   else ''
incy   = ' --incy '   + opts.incy   if (opts.incy)   else ''
align  = ' --align '  + opts.align  if (opts.align)  else ''
check  = ' --check '  + opts.check  if (opts.check)  else ''
ref    = ' --ref '    + opts.ref    if (opts.ref)    else ''
verbose = ' --verbose ' + opts.verbose if (opts.verbose) else ''

# LAPACK only
itype  = ' --itype '  + opts.itype  if (opts.itype)  else ''
factored = ' --factored ' + opts.factored if (opts.factored)  else ''
equed  = ' --equed '  + opts.equed  if (opts.equed)  else ''
pivot  = ' --pivot '  + opts.pivot  if (opts.pivot)  else ''
direction = ' --direction ' + opts.direction if (opts.direction) else ''
storev = ' --storev ' + opts.storev if (opts.storev) else ''
norm   = ' --norm '   + opts.norm   if (opts.norm)   else ''
ijob   = ' --ijob '   + opts.ijob   if (opts.ijob)   else ''
jobz   = ' --jobz '   + opts.jobz   if (opts.jobz)   else ''
jobu   = ' --jobu '   + opts.jobu   if (opts.jobu)   else ''
jobvt  = ' --jobvt '  + opts.jobvt  if (opts.jobvt)  else ''
jobvl  = ' --jobvl '  + opts.jobvl  if (opts.jobvl)  else ''
jobvr  = ' --jobvr '  + opts.jobvr  if (opts.jobvr)  else ''
jobvs  = ' --jobvs '  + opts.jobvs  if (opts.jobvs)  else ''
balanc = ' --balanc ' + opts.balanc if (opts.balanc) else ''
sort   = ' --sort '   + opts.sort   if (opts.sort)   else ''
sense  = ' --sense '  + opts.sense  if (opts.sense)  else ''
vect   = ' --vect '   + opts.vect   if (opts.vect)   else ''
l      = ' --l '      + opts.l      if (opts.l)      else ''
nb     = ' --nb '     + opts.nb     if (opts.nb)     else ''
ka     = ' --ka '     + opts.ka     if (opts.ka)     else ''
kb     = ' --kb '     + opts.kb     if (opts.kb)     else ''
kd     = ' --kd '     + opts.kd     if (opts.kd)     else ''
kl     = ' --kl '     + opts.kl     if (opts.kl)     else ''
ku     = ' --ku '     + opts.ku     if (opts.ku)     else ''
vl     = ' --vl '     + opts.vl     if (opts.vl)     else ''
vu     = ' --vu '     + opts.vu     if (opts.vu)     else ''
il     = ' --il '     + opts.il     if (opts.il)     else ''
iu     = ' --iu '     + opts.iu     if (opts.iu)     else ''
mtype  = ' --matrixtype ' + opts.matrixtype if (opts.matrixtype) else ''

# general options for all routines
gen = check + ref + verbose

# ------------------------------------------------------------------------------
# filters a comma separated list csv based on items in list values.
# if no items from csv are in values, returns first item in values.
def filter_csv( values, csv ):
    f = list( filter( lambda x: x in values, csv.split( ',' ) ) )
    if (not f):
        return values[0]
    return ','.join( f )
# end

# ------------------------------------------------------------------------------
# limit options to specific values
dtype_real    = ' --type ' + filter_csv( ('s', 'd'), opts.type )
dtype_complex = ' --type ' + filter_csv( ('c', 'z'), opts.type )
dtype_double  = ' --type ' + filter_csv( ('d', 'z'), opts.type )

trans_nt = ' --trans ' + filter_csv( ('n', 't'), opts.trans )
trans_nc = ' --trans ' + filter_csv( ('n', 'c'), opts.trans )

# positive inc
incx_pos = ' --incx ' + filter_csv( ('1', '2'), opts.incx )
incy_pos = ' --incy ' + filter_csv( ('1', '2'), opts.incy )

# ------------------------------------------------------------------------------
cmds = []

# LU
if (opts.lu and opts.host):
    cmds += [
    [ 'gesv',  gen + dtype + align + n ],
    # todo: equed
    [ 'gesvx', gen + dtype + align + n + factored + trans ],
    [ 'getrf', gen + dtype + align + mn ],
    [ 'getrs', gen + dtype + align + n + trans ],
    [ 'getri', gen + dtype + align + n ],
    [ 'gecon', gen + dtype + align + n ],
    [ 'gerfs', gen + dtype + align + n + trans ],
    [ 'geequ', gen + dtype + align + n ],
    ]

if (opts.lu and opts.device):
    # GPU
    cmds += [
    [ 'dev-getrf', gen + dtype + align + n ],
    ]

# General Banded
if (opts.gb and opts.host):
    cmds += [
    [ 'gbsv',  gen + dtype + align + n  + kl + ku ],
    [ 'gbtrf', gen + dtype + align + mn + kl + ku ],
    [ 'gbtrs', gen + dtype + align + n  + kl + ku + trans ],
    [ 'gbcon', gen + dtype + align + n  + kl + ku ],
    [ 'gbrfs', gen + dtype + align + n  + kl + ku + trans ],
    [ 'gbequ', gen + dtype + align + n  + kl + ku ],
    ]

# General Tri-Diagonal
if (opts.gt and opts.host):
    cmds += [
    [ 'gtsv',  gen + dtype + align + n ],
    [ 'gttrf', gen + dtype +         n ],
    [ 'gttrs', gen + dtype + align + n + trans ],
    [ 'gtcon', gen + dtype +         n ],
    [ 'gtrfs', gen + dtype + align + n + trans ],
    ]

# Cholesky
if (opts.chol and opts.host):
    cmds += [
    [ 'posv',  gen + dtype + align + n + uplo ],
    [ 'potrf', gen + dtype + align + n + uplo ],
    [ 'potrs', gen + dtype + align + n + uplo ],
    [ 'potri', gen + dtype + align + n + uplo ],
    [ 'pocon', gen + dtype + align + n + uplo ],
    [ 'porfs', gen + dtype + align + n + uplo ],
    [ 'poequ', gen + dtype + align + n ],  # only diagonal elements (no uplo)

    # Packed
    [ 'ppsv',  gen + dtype + align + n + uplo ],
    [ 'pptrf', gen + dtype +         n + uplo ],
    [ 'pptrs', gen + dtype + align + n + uplo ],
    [ 'pptri', gen + dtype +         n + uplo ],
    [ 'ppcon', gen + dtype +         n + uplo ],
    [ 'pprfs', gen + dtype + align + n + uplo ],
    [ 'ppequ', gen + dtype +         n + uplo ],

    # Banded
    [ 'pbsv',  gen + dtype + align + n + kd + uplo ],
    [ 'pbtrf', gen + dtype + align + n + kd + uplo ],
    [ 'pbtrs', gen + dtype + align + n + kd + uplo ],
    [ 'pbcon', gen + dtype + align + n + kd + uplo ],
    [ 'pbrfs', gen + dtype + align + n + kd + uplo ],
    [ 'pbequ', gen + dtype + align + n + kd + uplo ],

    # Tri-diagonal
    [ 'ptsv',  gen + dtype + align + n ],
    [ 'pttrf', gen + dtype         + n ],
    [ 'pttrs', gen + dtype + align + n + uplo ],
    [ 'ptcon', gen + dtype         + n ],
    [ 'ptrfs', gen + dtype + align + n + uplo ],
    ]

if (opts.chol and opts.device):
    # GPU
    cmds += [
    [ 'dev-potrf', gen + dtype + align + n + uplo ],
    ]

# symmetric indefinite, Bunch-Kaufman
if (opts.sysv and opts.host):
    cmds += [
    [ 'sysv',  gen + dtype + align + n + uplo ],
    [ 'sytrf', gen + dtype + align + n + uplo ],
    [ 'sytrs', gen + dtype + align + n + uplo ],
    [ 'sytri', gen + dtype + align + n + uplo ],
    [ 'sycon', gen + dtype + align + n + uplo ],
    [ 'syrfs', gen + dtype + align + n + uplo ],

    # Packed
    [ 'spsv',  gen + dtype + align + n + uplo ],
    [ 'sptrf', gen + dtype         + n + uplo ],
    [ 'sptrs', gen + dtype + align + n + uplo ],
    [ 'sptri', gen + dtype         + n + uplo ],
    [ 'spcon', gen + dtype         + n + uplo ],
    [ 'sprfs', gen + dtype + align + n + uplo ],
    ]

# symmetric indefinite, rook
if (opts.rook and opts.host):
    cmds += [
    # original Rook
    [ 'sysv_rook',  gen + dtype + align + n + uplo ],
    [ 'sytrf_rook', gen + dtype + align + n + uplo ],
    [ 'sytrs_rook', gen + dtype + align + n + uplo ],
    #[ 'sytri_rook', gen + dtype + align + n + uplo ],

    # new Rook
    [ 'sysv_rk',    gen + dtype + align + n + uplo ],
    [ 'sytrf_rk',   gen + dtype + align + n + uplo ],
    #[ 'sytrs_rk',   gen + dtype + align + n + uplo ],
    #[ 'sytri_rk',   gen + dtype + align + n + uplo ],
    ]

# symmetric indefinite, Aasen
if (opts.aasen and opts.host):
    cmds += [
    [ 'sysv_aa',  gen + dtype + align + n + uplo ],
    [ 'sytrf_aa', gen + dtype + align + n + uplo ],
    [ 'sytrs_aa', gen + dtype + align + n + uplo ],
    #[ 'sytri_aa', gen + dtype + align + n + uplo ],

    #[ 'sysv_aasen_2stage',  gen + dtype + align + n + uplo ],
    #[ 'sytrf_aasen_2stage', gen + dtype + align + n + uplo ],
    #[ 'sytrs_aasen_2stage', gen + dtype + align + n + uplo ],
    #[ 'sytri_aasen_2stage', gen + dtype + align + n + uplo ],
    ]

# Hermitian indefinite
if (opts.hesv and opts.host):
    cmds += [
    [ 'hesv',  gen + dtype + align + n + uplo ],
    [ 'hetrf', gen + dtype + align + n + uplo ],
    [ 'hetrs', gen + dtype + align + n + uplo ],
    [ 'hetri', gen + dtype + align + n + uplo ],
    [ 'hecon', gen + dtype + align + n + uplo ],
    [ 'herfs', gen + dtype + align + n + uplo ],

    # Packed
    [ 'hpsv',  gen + dtype + align + n + uplo ],
    [ 'hptrf', gen + dtype + n + uplo ],
    [ 'hptrs', gen + dtype + align + n + uplo ],
    [ 'hptri', gen + dtype + n + uplo ],
    [ 'hpcon', gen + dtype + n + uplo ],
    [ 'hprfs', gen + dtype + align + n + uplo ],
    ]

# least squares
if (opts.least_squares and opts.host):
    cmds += [
    [ 'gels',   gen + dtype + align + mn + trans_nc ],
    [ 'gelsy',  gen + dtype + align + mn ],
    # todo: gelsd is failing
    #[ 'gelsd',  gen + dtype + align + mn ],
    [ 'gelss',  gen + dtype + align + mn ],
    [ 'getsls', gen + dtype + align + mn + trans_nc ],

    # Generalized
    [ 'gglse', gen + dtype + align + mnk ],
    # todo: ggglm is failing
    #[ 'ggglm', gen + dtype + align + mnk ],
    ]

# QR
if (opts.qr and opts.host):
    cmds += [
    [ 'geqr',  gen + dtype + align + n + wide + tall ],
    [ 'geqrf', gen + dtype + align + n + wide + tall ],
    # todo: ggqrf is failing
    #[ 'ggqrf', gen + dtype + align + mnk ],
    [ 'ungqr', gen + dtype + align + mn ],  # m >= n
    #[ 'unmqr', gen + dtype_real    + align + mnk + side + trans    ],  # real does trans = N, T, C
    #[ 'unmqr', gen + dtype_complex + align + mnk + side + trans_nc ],  # complex does trans = N, C, not T

    [ 'orhr_col', gen + dtype_real + align + n + tall ],
    [ 'unhr_col', gen + dtype      + align + n + tall ],

    [ 'gemqrt', gen + dtype_real    + align + n + nb + side + trans    ],  # real does trans = N, T, C
    [ 'gemqrt', gen + dtype_complex + align + n + nb + side + trans_nc ],  # complex does trans = N, C, not T

    # Triangle-pentagon
    [ 'tpqrt',  gen + dtype + align + mn + l + nb ],
    [ 'tpqrt2', gen + dtype + align + mn + l ],
    [ 'tpmqrt', gen + dtype_real    + align + mn + l + nb + side + trans    ],  # real does trans = N, T, C
    [ 'tpmqrt', gen + dtype_complex + align + mn + l + nb + side + trans_nc ],  # complex does trans = N, C, not T
    #[ 'tprfb',  gen + dtype + align + mn + l ],  # TODO: bug in LAPACKE crashes tester
    ]

if (opts.qr and opts.device):
    # GPU
    cmds += [
    [ 'dev-geqrf', gen + dtype + align + n + wide + tall ],
    ]

# LQ
if (opts.lq and opts.host):
    cmds += [
    [ 'gelqf', gen + dtype + align + mn ],
    #[ 'gglqf', gen + dtype + align + mn ],
    [ 'unglq', gen + dtype + align + mn ],  # m <= n, k <= m  TODO Fix the input sizes to match constraints
    #[ 'unmlq', gen + dtype_real    + align + mnk + side + trans    ],  # real does trans = N, T, C
    #[ 'unmlq', gen + dtype_complex + align + mnk + side + trans_nc ],  # complex does trans = N, C, not T

    # Triangle-pentagon
    [ 'tplqt',  gen + dtype + align + mn + l + nb ],
    [ 'tplqt2', gen + dtype + align + mn + l ],
    [ 'tpmlqt', gen + dtype_real    + align + mn + l + nb + side + trans    ],  # real does trans = N, T, C
    [ 'tpmlqt', gen + dtype_complex + align + mn + l + nb + side + trans_nc ],  # complex does trans = N, C, not T
    ]

# QL
if (opts.ql and opts.host):
    cmds += [
    [ 'geqlf', gen + dtype + align + mn ],
    #[ 'ggqlf', gen + dtype + align + mn ],
    [ 'ungql', gen + dtype + align + mn ],
    #[ 'unmql', gen + dtype_real    + align + mnk + side + trans    ],  # real does trans = N, T, C
    #[ 'unmql', gen + dtype_complex + align + mnk + side + trans_nc ],  # complex does trans = N, C, not T
    ]

# RQ
if (opts.rq and opts.host):
    cmds += [
    [ 'gerqf', gen + dtype + align + mn ],
    # todo: ggrqf is failing
    #[ 'ggrqf', gen + dtype + align + mnk ],
    [ 'ungrq', gen + dtype + align + mnk ],
    #[ 'unmrq', gen + dtype_real    + align + mnk + side + trans    ],  # real does trans = N, T, C
    #[ 'unmrq', gen + dtype_complex + align + mnk + side + trans_nc ],  # complex does trans = N, C, not T
    ]

# symmetric eigenvalues
if (opts.syev and opts.host):
    cmds += [
    [ 'heev',  gen + dtype + align + n + jobz + uplo ],
    [ 'heevx', gen + dtype + align + n + jobz + uplo + vl + vu ],
    [ 'heevx', gen + dtype + align + n + jobz + uplo + il + iu ],
    [ 'heevd', gen + dtype + align + n + jobz + uplo ],
    [ 'heevr', gen + dtype + align + n + jobz + uplo + vl + vu ],
    [ 'heevr', gen + dtype + align + n + jobz + uplo + il + iu ],
    [ 'hetrd', gen + dtype + align + n + uplo ],
    [ 'lae2',  gen + dtype_real ],  # 2x2, eigvals only
    [ 'laev2', gen + dtype ],  # 2x2
    [ 'ungtr', gen + dtype + align + n + uplo ],
    [ 'unmtr', gen + dtype_real    + align + mn + uplo + side + trans    ],  # real does trans = N, T, C
    [ 'unmtr', gen + dtype_complex + align + mn + uplo + side + trans_nc ],  # complex does trans = N, C, not T

    # Packed
    [ 'hpev',  gen + dtype + align + n + jobz + uplo ],
    [ 'hpevx', gen + dtype + align + n + jobz + uplo + vl + vu ],
    [ 'hpevx', gen + dtype + align + n + jobz + uplo + il + iu ],
    [ 'hpevd', gen + dtype + align + n + jobz + uplo ],
    #[ 'hpevr', gen + dtype + align + n + jobz + uplo + vl + vu ],
    #[ 'hpevr', gen + dtype + align + n + jobz + uplo + il + iu ],
    [ 'hptrd', gen + dtype + n + uplo ],
    [ 'upgtr', gen + dtype + align + n + uplo ],
    [ 'upmtr', gen + dtype + align + mn + side + uplo + trans_nc ],

    # Banded
    [ 'hbev',  gen + dtype + align + n + jobz + uplo ],
    [ 'hbevx', gen + dtype + align + n + jobz + uplo + vl + vu ],
    [ 'hbevx', gen + dtype + align + n + jobz + uplo + il + iu ],
    [ 'hbevd', gen + dtype + align + n + jobz + uplo ],
    #[ 'hbevr', gen + dtype + align + n + jobz + uplo + vl + vu ],
    #[ 'hbevr', gen + dtype + align + n + jobz + uplo + il + iu ],
    #[ 'hbtrd', gen + dtype + align + n + uplo ],
    #[ 'ubgtr', gen + dtype + align + n + uplo ],
    #[ 'ubmtr', gen + dtype_real    + la + mn + uplo + side + trans    ],
    #[ 'ubmtr', gen + dtype_complex + la + mn + uplo + side + trans_nc ],
    ]

if (opts.syev and opts.device):
    # GPU
    cmds += [
    [ 'dev-heevd', gen + dtype + align + n + jobz + uplo ],
    ]

# generalized symmetric eigenvalues
if (opts.sygv and opts.host):
    cmds += [
    [ 'hegv',  gen + dtype + align + n + itype + jobz + uplo ],
    [ 'hegvx', gen + dtype + align + n + itype + jobz + uplo + vl + vu ],
    [ 'hegvx', gen + dtype + align + n + itype + jobz + uplo + il + iu ],
    [ 'hegvd', gen + dtype + align + n + itype + jobz + uplo ],
    #[ 'hegvr', gen + dtype + align + n + uplo ],
    [ 'hegst', gen + dtype + align + n + itype + uplo ],

    # Packed
    [ 'hpgv',  gen + dtype + align + n + itype + jobz + uplo ],
    [ 'hpgvx', gen + dtype + align + n + itype + jobz + uplo + vl + vu ],
    [ 'hpgvx', gen + dtype + align + n + itype + jobz + uplo + il + iu ],
    [ 'hpgvd', gen + dtype + align + n + itype + jobz + uplo ],
    #[ 'hpgvr', gen + dtype + align + n + uplo ],
    [ 'hpgst', gen + dtype + n + itype + uplo ],

    # Banded
    [ 'hbgv',  gen + dtype + align + n + jobz + uplo + ka + kb ],
    [ 'hbgvx', gen + dtype + align + n + jobz + uplo + ka + kb + vl + vu ],
    [ 'hbgvx', gen + dtype + align + n + jobz + uplo + ka + kb + il + iu ],
    [ 'hbgvd', gen + dtype + align + n + jobz + uplo + ka + kb ],
    #[ 'hbgvr', gen + dtype + align + n + uplo + ka + kb ],
    #[ 'hbgst', gen + dtype + align + n + vect + uplo + ka + kb ],
    ]

# non-symmetric eigenvalues
if (opts.geev and opts.host):
    cmds += [
    [ 'geev',  gen + dtype + align + n + jobvl + jobvr ],
    # todo: ggev is failing
    #[ 'ggev',  gen + dtype + align + n + jobvl + jobvr ],
    #[ 'geevx', gen + dtype + align + n + balanc + jobvl + jobvr + sense ],
    [ 'gehrd', gen + dtype + align + n ],
    [ 'unghr', gen + dtype + align + n ],
    [ 'unmhr', gen + dtype_real    + align + mn + side + trans    ],  # real does trans = N, T, C
    [ 'unmhr', gen + dtype_complex + align + mn + side + trans_nc ],  # complex does trans = N, C, not T
    #[ 'trevc', gen + dtype + align + n + side + howmany + select ],
    #[ 'geesx', gen + dtype + align + n + jobvs + sort + select + sense ],
    [ 'tgexc', gen + dtype + align + n + jobvl + jobvr ],
    [ 'tgsen', gen + dtype + align + n + jobvl + jobvr + ijob ],
    ]

# svd
if (opts.svd and opts.host):
    cmds += [
    # todo: MKL seems to have a bug with jobu=o,s and jobvt=o,s,a
    # for tall matrices, e.g., dim=100x50. Skip failing combinations for now.
    #[ 'gesvd',         gen + dtype + align + mn + jobu + jobvt ],
    [ 'gesvd',         gen + dtype + align + mn + " --jobu n,a" + jobvt ],
    [ 'gesvd',         gen + dtype + align + mn + " --jobu o,s --jobvt n" ],
    [ 'gesdd',         gen + dtype + align + mn + jobu ],
    # todo: gesvdx is failing
    #[ 'gesvdx',        gen + dtype + align + mn + jobz + jobvr + vl + vu ],
    #[ 'gesvdx',        gen + dtype + align + mn + jobz + jobvr + il + iu ],
    #[ 'gesvd_2stage',  gen + dtype + align + mn ],
    #[ 'gesdd_2stage',  gen + dtype + align + mn ],
    #[ 'gesvdx_2stage', gen + dtype + align + mn ],
    #[ 'gejsv',         gen + dtype + align + mn ],
    #[ 'gesvj',         gen + dtype + align + mn + joba + jobu + jobv ],
    ]

# auxilary
if (opts.aux and opts.host):
    cmds += [
    [ 'lacpy', gen + dtype + align + mn + mtype ],
    [ 'laed4', gen + dtype_real + n ],
    [ 'laset', gen + dtype + align + mn + mtype ],
    [ 'laswp', gen + dtype + align + mn ],
    ]

# auxilary - householder
if (opts.aux_house and opts.host):
    cmds += [
    [ 'larfg', dtype         + n   + incx_pos ],
    [ 'larfgp', dtype        + n   + incx_pos ],
    [ 'larf',  gen + dtype + align + mn  + incx + side ],
    [ 'larfx', gen + dtype + align + mn  + side ],
    [ 'larfy', gen + dtype + align + n   + incx ],
    [ 'larfb', gen + dtype + align + mnk + side + trans + direction + storev ],
    [ 'larft', gen + dtype + align + nk  + direction + storev ],
    ]

# auxilary - Givens rotations
if (opts.aux_givens and opts.host):
    cmds += [
    [ 'lasr', gen + dtype + align + mn + side + pivot + direction ],
    ]

# auxilary - norms
if (opts.aux_norm and opts.host):
    cmds += [
    [ 'lange', gen + dtype + align + mn + norm ],
    # todo: lanhe is failing
    #[ 'lanhe', gen + dtype + align + n  + norm + uplo ],
    [ 'lansy', gen + dtype + align + n  + norm + uplo ],
    [ 'lantr', gen + dtype + align + mn + norm + uplo + diag ],
    [ 'lanhs', gen + dtype + align + n  + norm ],

    # Packed
    [ 'lanhp', gen + dtype + n + norm + uplo ],
    [ 'lansp', gen + dtype + n + norm + uplo ],
    [ 'lantp', gen + dtype + n + norm + uplo + diag ],

    # Banded
    [ 'langb', gen + dtype + align + mn + kl + ku + norm ],
    [ 'lanhb', gen + dtype + align + n + kd + norm + uplo ],
    [ 'lansb', gen + dtype + align + n + kd + norm + uplo ],
    [ 'lantb', gen + dtype + align + n + kd + norm + uplo + diag ],

    # Tri-diagonal
    [ 'langt', gen + dtype + n + norm ],
    [ 'lanht', gen + dtype + n + norm ],
    [ 'lanst', gen + dtype + n + norm ],
    ]

# additional blas
if (opts.blas and opts.host):
    cmds += [
    [ 'syr',   gen + dtype + align + n + uplo ],
    [ 'symv',  gen + dtype + layout + align + uplo + n + incx + incy ],
    ]

# ------------------------------------------------------------------------------
# When stdout is redirected to file instead of TTY console,
# and  stderr is still going to a TTY console,
# print extra summary messages to stderr.
output_redirected = sys.stderr.isatty() and not sys.stdout.isatty()

# ------------------------------------------------------------------------------
# if output is redirected, prints to both stderr and stdout;
# otherwise prints to just stdout.
def print_tee( *args ):
    global output_redirected
    print( *args )
    if (output_redirected):
        print( *args, file=sys.stderr )
# end

# ------------------------------------------------------------------------------
# cmd is a pair of strings: (function, args)

def run_test( cmd ):
    cmd = opts.test +' '+ cmd[1] +' '+ cmd[0]
    print_tee( cmd )
    if (opts.dry_run):
        return (None, None)

    output = ''
    p = subprocess.Popen( cmd.split(), stdout=subprocess.PIPE,
                                       stderr=subprocess.STDOUT )
    p_out = p.stdout
    if (sys.version_info.major >= 3):
        p_out = io.TextIOWrapper(p.stdout, encoding='utf-8')
    # Read unbuffered ("for line in p.stdout" will buffer).
    for line in iter(p_out.readline, ''):
        print( line, end='' )
        output += line
    err = p.wait()
    if (err != 0):
        print_tee( 'FAILED: exit code', err )
    else:
        print_tee( 'pass' )
    return (err, output)
# end

# ------------------------------------------------------------------------------
# Utility to pretty print XML.
# See https://stackoverflow.com/a/33956544/1655607
#
def indent_xml( elem, level=0 ):
    i = "\n" + level*"  "
    if len(elem):
        if not elem.text or not elem.text.strip():
            elem.text = i + "  "
        if not elem.tail or not elem.tail.strip():
            elem.tail = i
        for elem in elem:
            indent_xml( elem, level+1 )
        if not elem.tail or not elem.tail.strip():
            elem.tail = i
    else:
        if level and (not elem.tail or not elem.tail.strip()):
            elem.tail = i
# end

# ------------------------------------------------------------------------------
# run each test

start = time.time()
print_tee( time.ctime() )

failed_tests = []
passed_tests = []
ntests = len(opts.tests)
run_all = (ntests == 0)

seen = set()
for cmd in cmds:
    if ((run_all or cmd[0] in opts.tests) and cmd[0] not in opts.exclude):
        if (start_routine and cmd[0] != start_routine):
            print_tee( 'skipping', cmd[0] )
            continue
        start_routine = None

        seen.add( cmd[0] )
        (err, output) = run_test( cmd )
        if (err):
            failed_tests.append( (cmd[0], err, output) )
        else:
            passed_tests.append( cmd[0] )

not_seen = list( filter( lambda x: x not in seen, opts.tests ) )
if (not_seen):
    print_tee( 'Warning: unknown routines:', ' '.join( not_seen ))

# print summary of failures
nfailed = len( failed_tests )
if (nfailed > 0):
    print_tee( '\n' + str(nfailed) + ' routines FAILED:',
               ', '.join( [x[0] for x in failed_tests] ) )
else:
    print_tee( '\n' + 'All routines passed.' )

# generate jUnit compatible test report
if opts.xml:
    print( 'writing XML file', opts.xml )
    root = ET.Element("testsuites")
    doc = ET.SubElement(root, "testsuite",
                        name="lapackpp_suite",
                        tests=str(ntests),
                        errors="0",
                        failures=str(nfailed))

    for (test, err, output) in failed_tests:
        testcase = ET.SubElement(doc, "testcase", name=test)

        failure = ET.SubElement(testcase, "failure")
        if (err < 0):
            failure.text = "exit with signal " + str(-err)
        else:
            failure.text = str(err) + " tests failed"

        system_out = ET.SubElement(testcase, "system-out")
        system_out.text = output
    # end

    for test in passed_tests:
        testcase = ET.SubElement(doc, 'testcase', name=test)
        testcase.text = 'PASSED'

    tree = ET.ElementTree(root)
    indent_xml( root )
    tree.write( opts.xml )
# end

elapsed = time.time() - start
print_tee( 'Elapsed %.2f sec' % elapsed )
print_tee( time.ctime() )

exit( nfailed )