forked from mikerabat/mrmath
-
Notifications
You must be signed in to change notification settings - Fork 1
/
AVXMatrixNormOperations.pas
874 lines (660 loc) · 30 KB
/
AVXMatrixNormOperations.pas
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
// ###################################################################
// #### This file is part of the mathematics library project, and is
// #### offered under the licence agreement described on
// #### http://www.mrsoft.org/
// ####
// #### Copyright:(c) 2018, Michael R. . All rights reserved.
// ####
// #### Unless required by applicable law or agreed to in writing, software
// #### distributed under the License is distributed on an "AS IS" BASIS,
// #### WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// #### See the License for the specific language governing permissions and
// #### limitations under the License.
// ###################################################################
unit AVXMatrixNormOperations;
interface
{$IFDEF CPUX64}
{$DEFINE x64}
{$ENDIF}
{$IFDEF cpux86_64}
{$DEFINE x64}
{$ENDIF}
{$IFNDEF x64}
uses MatrixConst;
function AVXMatrixElementwiseNorm2Aligned(dest : PDouble; const LineWidth : TASMNativeInt; Width, height : TASMNativeInt) : double; {$IFDEF FPC} assembler; {$ELSE} register; {$ENDIF}
function AVXMatrixElementwiseNorm2UnAligned(dest : PDouble; const LineWidth : TASMNativeInt; Width, height : TASMNativeInt) : double; {$IFDEF FPC} assembler; {$ELSE} register; {$ENDIF}
procedure AVXMatrixNormalizeRowAligned(dest : PDouble; const destLineWidth : TASMNativeInt; Src : PDouble; const srcLineWidth : TASMNativeInt; width, height : TASMNativeInt); {$IFDEF FPC} assembler; {$ELSE} register; {$ENDIF}
procedure AVXMatrixNormalizeRowUnAligned(dest : PDouble; const destLineWidth : TASMNativeInt; Src : PDouble; const srcLineWidth : TASMNativeInt; width, height : TASMNativeInt); {$IFDEF FPC} assembler; {$ELSE} register; {$ENDIF}
procedure AVXMatrixNormalizeColumnAligned(dest : PDouble; const destLineWidth : TASMNativeInt; Src : PDouble; const srcLineWidth : TASMNativeInt; width, height : TASMNativeInt); {$IFDEF FPC} assembler; {$ELSE} register; {$ENDIF}
procedure AVXMatrixNormalizeColumnUnAligned(dest : PDouble; const destLineWidth : TASMNativeInt; Src : PDouble; const srcLineWidth : TASMNativeInt; width, height : TASMNativeInt); {$IFDEF FPC} assembler; {$ELSE} register; {$ENDIF}
{$ENDIF}
implementation
{$IFNDEF x64}
{$WARNINGS OFF}
{$IFDEF FPC} {$ASMMODE intel} {$S-} {$ENDIF}
const cLocOne : double = 1;
function AVXMatrixElementwiseNorm2Aligned(dest : PDouble; const LineWidth : TASMNativeInt; Width, height : TASMNativeInt) : double;
// eax = dest, edx = lineWidth, ecx = width
asm
// prolog - simulate stack
push ebx;
push esi;
//iters := -width*sizeof(double);
imul ecx, -8;
// helper registers for the mt1, mt2 and dest pointers
sub eax, ecx;
{$IFDEF FPC}vxorpd xmm0, xmm0, xmm0;{$ELSE}db $C5,$F9,$57,$C0;{$ENDIF}
{$IFDEF FPC}vxorpd ymm1, ymm1, ymm1;{$ELSE}db $C5,$F5,$57,$C9;{$ENDIF}
mov ebx, height;
@@addforyloop:
// for x := 0 to w - 1;
// prepare for reverse loop
mov esi, ecx;
@addforxloop:
add esi, 128;
jg @loopEnd;
// prefetch data...
// prefetch [eax + rax];
// mul add:
{$IFDEF FPC}vmovapd ymm3, [eax + esi - 128];{$ELSE}db $C5,$FD,$28,$5C,$30,$80;{$ENDIF}
{$IFDEF FPC}vmulpd ymm3, ymm3, ymm3;{$ELSE}db $C5,$E5,$59,$DB;{$ENDIF}
{$IFDEF FPC}vaddpd ymm1, ymm1, ymm3;{$ELSE}db $C5,$F5,$58,$CB;{$ENDIF}
{$IFDEF FPC}vmovapd ymm2, [eax + esi - 96];{$ELSE}db $C5,$FD,$28,$54,$30,$A0;{$ENDIF}
{$IFDEF FPC}vmulpd ymm2, ymm2, ymm2;{$ELSE}db $C5,$ED,$59,$D2;{$ENDIF}
{$IFDEF FPC}vaddpd ymm1, ymm1, ymm2;{$ELSE}db $C5,$F5,$58,$CA;{$ENDIF}
{$IFDEF FPC}vmovapd ymm3, [eax + esi - 64];{$ELSE}db $C5,$FD,$28,$5C,$30,$C0;{$ENDIF}
{$IFDEF FPC}vmulpd ymm3, ymm3, ymm3;{$ELSE}db $C5,$E5,$59,$DB;{$ENDIF}
{$IFDEF FPC}vaddpd ymm1, ymm1, ymm3;{$ELSE}db $C5,$F5,$58,$CB;{$ENDIF}
{$IFDEF FPC}vmovapd ymm2, [eax + esi - 32];{$ELSE}db $C5,$FD,$28,$54,$30,$E0;{$ENDIF}
{$IFDEF FPC}vmulpd ymm2, ymm2, ymm2;{$ELSE}db $C5,$ED,$59,$D2;{$ENDIF}
{$IFDEF FPC}vaddpd ymm1, ymm1, ymm2;{$ELSE}db $C5,$F5,$58,$CA;{$ENDIF}
jmp @addforxloop
@loopEnd:
sub esi, 128;
jz @nextLine;
@addforxloop2:
add esi, 16;
jg @@addforxloop2end;
{$IFDEF FPC}vmovapd xmm3, [eax + esi - 16];{$ELSE}db $C5,$F9,$28,$5C,$30,$F0;{$ENDIF}
{$IFDEF FPC}vmulpd xmm3, xmm3, xmm3;{$ELSE}db $C5,$E1,$59,$DB;{$ENDIF}
{$IFDEF FPC}vaddpd xmm0, xmm0, xmm3;{$ELSE}db $C5,$F9,$58,$C3;{$ENDIF}
jmp @addforxloop2;
@@addforxloop2end:
sub esi, 16;
jz @nextLine;
{$IFDEF FPC}vmovsd xmm3, [eax + esi];{$ELSE}db $C5,$FB,$10,$1C,$30;{$ENDIF}
{$IFDEF FPC}vmulsd xmm3, xmm3, xmm3;{$ELSE}db $C5,$E3,$59,$DB;{$ENDIF}
{$IFDEF FPC}vaddsd xmm0, xmm0, xmm3;{$ELSE}db $C5,$FB,$58,$C3;{$ENDIF}
@nextLine:
// next line:
add eax, edx;
// loop y end
dec ebx;
jnz @@addforyloop;
// build result
{$IFDEF FPC}vextractf128 xmm2, ymm1, 1;{$ELSE}db $C4,$E3,$7D,$19,$CA,$01;{$ENDIF}
{$IFDEF FPC}vhaddpd xmm1, xmm1, xmm2;{$ELSE}db $C5,$F1,$7C,$CA;{$ENDIF}
{$IFDEF FPC}vaddpd xmm0, xmm0, xmm1;{$ELSE}db $C5,$F9,$58,$C1;{$ENDIF}
{$IFDEF FPC}vhaddpd xmm0, xmm0, xmm0;{$ELSE}db $C5,$F9,$7C,$C0;{$ENDIF}
// epilog claenup stack
{$IFDEF FPC}vzeroupper;{$ELSE}db $C5,$F8,$77;{$ENDIF}
movsd Result, xmm0;
pop esi;
pop ebx;
end;
function AVXMatrixElementwiseNorm2UnAligned(dest : PDouble; const LineWidth : TASMNativeInt; Width, height : TASMNativeInt) : double;
asm
// prolog - simulate stack
push ebx;
push esi;
//iters := -width*sizeof(double);
imul ecx, -8;
// helper registers for the mt1, mt2 and dest pointers
sub eax, ecx;
{$IFDEF FPC}vxorpd xmm0, xmm0, xmm0;{$ELSE}db $C5,$F9,$57,$C0;{$ENDIF}
{$IFDEF FPC}vxorpd ymm1, ymm1, ymm1;{$ELSE}db $C5,$F5,$57,$C9;{$ENDIF}
mov ebx, height;
@@addforyloop:
// for x := 0 to w - 1;
// prepare for reverse loop
mov esi, ecx;
@addforxloop:
add esi, 128;
jg @loopEnd;
// prefetch data...
// prefetch [eax + esi];
// mul add:
{$IFDEF FPC}vmovupd ymm3, [eax + esi - 128];{$ELSE}db $C5,$FD,$10,$5C,$30,$80;{$ENDIF}
{$IFDEF FPC}vmulpd ymm3, ymm3, ymm3;{$ELSE}db $C5,$E5,$59,$DB;{$ENDIF}
{$IFDEF FPC}vaddpd ymm1, ymm1, ymm3;{$ELSE}db $C5,$F5,$58,$CB;{$ENDIF}
{$IFDEF FPC}vmovupd ymm2, [eax + esi - 96];{$ELSE}db $C5,$FD,$10,$54,$30,$A0;{$ENDIF}
{$IFDEF FPC}vmulpd ymm2, ymm2, ymm2;{$ELSE}db $C5,$ED,$59,$D2;{$ENDIF}
{$IFDEF FPC}vaddpd ymm1, ymm1, ymm2;{$ELSE}db $C5,$F5,$58,$CA;{$ENDIF}
{$IFDEF FPC}vmovupd ymm3, [eax + esi - 64];{$ELSE}db $C5,$FD,$10,$5C,$30,$C0;{$ENDIF}
{$IFDEF FPC}vmulpd ymm3, ymm3, ymm3;{$ELSE}db $C5,$E5,$59,$DB;{$ENDIF}
{$IFDEF FPC}vaddpd ymm1, ymm1, ymm3;{$ELSE}db $C5,$F5,$58,$CB;{$ENDIF}
{$IFDEF FPC}vmovupd ymm2, [eax + esi - 32];{$ELSE}db $C5,$FD,$10,$54,$30,$E0;{$ENDIF}
{$IFDEF FPC}vmulpd ymm2, ymm2, ymm2;{$ELSE}db $C5,$ED,$59,$D2;{$ENDIF}
{$IFDEF FPC}vaddpd ymm1, ymm1, ymm2;{$ELSE}db $C5,$F5,$58,$CA;{$ENDIF}
jmp @addforxloop
@loopEnd:
sub esi, 128;
jz @nextLine;
@addforxloop2:
add esi, 16;
jg @@addforxloop2end;
{$IFDEF FPC}vmovupd xmm3, [eax + esi - 16];{$ELSE}db $C5,$F9,$10,$5C,$30,$F0;{$ENDIF}
{$IFDEF FPC}vmulpd xmm3, xmm3, xmm3;{$ELSE}db $C5,$E1,$59,$DB;{$ENDIF}
{$IFDEF FPC}vaddpd xmm0, xmm0, xmm3;{$ELSE}db $C5,$F9,$58,$C3;{$ENDIF}
jmp @addforxloop2;
@@addforxloop2end:
sub esi, 16;
jz @nextLine;
{$IFDEF FPC}vmovsd xmm3, [eax + esi];{$ELSE}db $C5,$FB,$10,$1C,$30;{$ENDIF}
{$IFDEF FPC}vmulsd xmm3, xmm3, xmm3;{$ELSE}db $C5,$E3,$59,$DB;{$ENDIF}
{$IFDEF FPC}vaddsd xmm0, xmm0, xmm3;{$ELSE}db $C5,$FB,$58,$C3;{$ENDIF}
@nextLine:
// next line:
add eax, edx;
// loop y end
dec ebx;
jnz @@addforyloop;
// build result
{$IFDEF FPC}vextractf128 xmm2, ymm1, 1;{$ELSE}db $C4,$E3,$7D,$19,$CA,$01;{$ENDIF}
{$IFDEF FPC}vhaddpd xmm1, xmm1, xmm2;{$ELSE}db $C5,$F1,$7C,$CA;{$ENDIF}
{$IFDEF FPC}vaddpd xmm0, xmm0, xmm1;{$ELSE}db $C5,$F9,$58,$C1;{$ENDIF}
{$IFDEF FPC}vhaddpd xmm0, xmm0, xmm0;{$ELSE}db $C5,$F9,$7C,$C0;{$ENDIF}
// epilog claenup stack
{$IFDEF FPC}vzeroupper;{$ELSE}db $C5,$F8,$77;{$ENDIF}
movsd Result, xmm0;
pop esi;
pop ebx;
end;
procedure AVXMatrixNormalizeRowAligned(dest : PDouble; const destLineWidth : TASMNativeInt; Src : PDouble; const srcLineWidth : TASMNativeInt; width, height : TASMNativeInt);
var tmp : double;
asm
// prolog - simulate stack
push ebx;
push edi;
push esi;
//iters := -width*sizeof(double);
mov edi, width;
imul edi, -8;
// helper registers for the mt1, mt2 and dest pointers
sub eax, edi;
sub ecx, edi;
mov ebx, height;
@@addforyloop:
{$IFDEF FPC}vxorpd xmm0, xmm0, xmm0;{$ELSE}db $C5,$F9,$57,$C0;{$ENDIF}
{$IFDEF FPC}vxorpd ymm1, ymm1, ymm1;{$ELSE}db $C5,$F5,$57,$C9;{$ENDIF}
// for x := 0 to w - 1;
// prepare for reverse loop
mov esi, edi;
@addforxloop:
add esi, 128;
jg @loopEnd;
// prefetch data...
// prefetch [eax + esi];
// mul add:
{$IFDEF FPC}vmovapd ymm3, [ecx + esi - 128];{$ELSE}db $C5,$FD,$28,$5C,$31,$80;{$ENDIF}
{$IFDEF FPC}vmulpd ymm3, ymm3, ymm3;{$ELSE}db $C5,$E5,$59,$DB;{$ENDIF}
{$IFDEF FPC}vaddpd ymm1, ymm1, ymm3;{$ELSE}db $C5,$F5,$58,$CB;{$ENDIF}
{$IFDEF FPC}vmovapd ymm2, [ecx + esi - 96];{$ELSE}db $C5,$FD,$28,$54,$31,$A0;{$ENDIF}
{$IFDEF FPC}vmulpd ymm2, ymm2, ymm2;{$ELSE}db $C5,$ED,$59,$D2;{$ENDIF}
{$IFDEF FPC}vaddpd ymm1, ymm1, ymm2;{$ELSE}db $C5,$F5,$58,$CA;{$ENDIF}
{$IFDEF FPC}vmovapd ymm3, [ecx + esi - 64];{$ELSE}db $C5,$FD,$28,$5C,$31,$C0;{$ENDIF}
{$IFDEF FPC}vmulpd ymm3, ymm3, ymm3;{$ELSE}db $C5,$E5,$59,$DB;{$ENDIF}
{$IFDEF FPC}vaddpd ymm1, ymm1, ymm3;{$ELSE}db $C5,$F5,$58,$CB;{$ENDIF}
{$IFDEF FPC}vmovapd ymm2, [ecx + esi - 32];{$ELSE}db $C5,$FD,$28,$54,$31,$E0;{$ENDIF}
{$IFDEF FPC}vmulpd ymm2, ymm2, ymm2;{$ELSE}db $C5,$ED,$59,$D2;{$ENDIF}
{$IFDEF FPC}vaddpd ymm1, ymm1, ymm2;{$ELSE}db $C5,$F5,$58,$CA;{$ENDIF}
jmp @addforxloop
@loopEnd:
sub esi, 128;
jz @nextLine;
@addforxloop2:
add esi, 16;
jg @@addforxloop2end;
{$IFDEF FPC}vmovapd xmm3, [ecx + esi - 16];{$ELSE}db $C5,$F9,$28,$5C,$31,$F0;{$ENDIF}
{$IFDEF FPC}vmulpd xmm3, xmm3, xmm3;{$ELSE}db $C5,$E1,$59,$DB;{$ENDIF}
{$IFDEF FPC}vaddpd xmm0, xmm0, xmm3;{$ELSE}db $C5,$F9,$58,$C3;{$ENDIF}
jmp @addforxloop2;
@@addforxloop2end:
sub esi, 16;
jz @nextLine;
{$IFDEF FPC}vmovsd xmm3, [ecx + esi];{$ELSE}db $C5,$FB,$10,$1C,$31;{$ENDIF}
{$IFDEF FPC}vmulsd xmm3, xmm3, xmm3;{$ELSE}db $C5,$E3,$59,$DB;{$ENDIF}
{$IFDEF FPC}vaddsd xmm0, xmm0, xmm3;{$ELSE}db $C5,$FB,$58,$C3;{$ENDIF}
@nextLine:
// build result
{$IFDEF FPC}vextractf128 xmm2, ymm1, 1;{$ELSE}db $C4,$E3,$7D,$19,$CA,$01;{$ENDIF}
{$IFDEF FPC}vhaddpd xmm1, xmm1, xmm2;{$ELSE}db $C5,$F1,$7C,$CA;{$ENDIF}
{$IFDEF FPC}vaddpd xmm0, xmm0, xmm1;{$ELSE}db $C5,$F9,$58,$C1;{$ENDIF}
{$IFDEF FPC}vhaddpd xmm0, xmm0, xmm0;{$ELSE}db $C5,$F9,$7C,$C0;{$ENDIF}
// build result
{$IFDEF FPC}vsqrtsd xmm0, xmm0, xmm0;{$ELSE}db $C5,$FB,$51,$C0;{$ENDIF}
//1/sqrt(norm)
lea esi, cLocOne;
{$IFDEF FPC}vmovsd xmm1, [esi];{$ELSE}db $C5,$FB,$10,$0E;{$ENDIF}
{$IFDEF FPC}vdivsd xmm1, xmm1, xmm0;{$ELSE}db $C5,$F3,$5E,$C8;{$ENDIF}
lea esi, tmp;
{$IFDEF FPC}vmovsd [esi], xmm1;{$ELSE}db $C5,$FB,$11,$0E;{$ENDIF}
{$IFDEF FPC}vbroadcastsd ymm2, [esi]; {$ELSE}db $C4,$E2,$7D,$19,$16;{$ENDIF} // need avx2
// normalize
mov esi, edi;
@addforxloop3:
add esi, 128;
jg @loopEnd2;
// prefetch data...
// prefetch [ecx + esi];
// prefetchw [eax + esi];
// mult:
{$IFDEF FPC}vmovapd ymm0, [ecx + esi - 128];{$ELSE}db $C5,$FD,$28,$44,$31,$80;{$ENDIF}
{$IFDEF FPC}vmulpd ymm0, ymm0, ymm2;{$ELSE}db $C5,$FD,$59,$C2;{$ENDIF}
{$IFDEF FPC}vmovapd [eax + esi - 128], ymm0;{$ELSE}db $C5,$FD,$29,$44,$30,$80;{$ENDIF}
{$IFDEF FPC}vmovapd ymm1, [ecx + esi - 96];{$ELSE}db $C5,$FD,$28,$4C,$31,$A0;{$ENDIF}
{$IFDEF FPC}vmulpd ymm1, ymm1, ymm2;{$ELSE}db $C5,$F5,$59,$CA;{$ENDIF}
{$IFDEF FPC}vmovapd [eax + esi - 96], ymm1;{$ELSE}db $C5,$FD,$29,$4C,$30,$A0;{$ENDIF}
{$IFDEF FPC}vmovapd ymm0, [ecx + esi - 64];{$ELSE}db $C5,$FD,$28,$44,$31,$C0;{$ENDIF}
{$IFDEF FPC}vmulpd ymm0, ymm0, ymm2;{$ELSE}db $C5,$FD,$59,$C2;{$ENDIF}
{$IFDEF FPC}vmovapd [eax + esi - 64], ymm0;{$ELSE}db $C5,$FD,$29,$44,$30,$C0;{$ENDIF}
{$IFDEF FPC}vmovapd ymm1, [ecx + esi - 32];{$ELSE}db $C5,$FD,$28,$4C,$31,$E0;{$ENDIF}
{$IFDEF FPC}vmulpd ymm1, ymm1, ymm2;{$ELSE}db $C5,$F5,$59,$CA;{$ENDIF}
{$IFDEF FPC}vmovapd [eax + esi - 32], ymm1;{$ELSE}db $C5,$FD,$29,$4C,$30,$E0;{$ENDIF}
jmp @addforxloop3
@loopEnd2:
sub esi, 128;
jz @nextLine2;
@addforxloop4:
add esi, 16;
jg @addforxloop4end;
{$IFDEF FPC}vmovapd xmm0, [ecx + esi - 16];{$ELSE}db $C5,$F9,$28,$44,$31,$F0;{$ENDIF}
{$IFDEF FPC}vmulpd xmm0, xmm0, xmm2;{$ELSE}db $C5,$F9,$59,$C2;{$ENDIF}
{$IFDEF FPC}vmovapd [eax + esi - 16], xmm0;{$ELSE}db $C5,$F9,$29,$44,$30,$F0;{$ENDIF}
jmp @addforxloop4;
@addforxloop4end:
sub esi, 16;
jz @nextLine2;
{$IFDEF FPC}vmovsd xmm0, [ecx + esi];{$ELSE}db $C5,$FB,$10,$04,$31;{$ENDIF}
{$IFDEF FPC}vmulsd xmm0, xmm0, xmm2;{$ELSE}db $C5,$FB,$59,$C2;{$ENDIF}
{$IFDEF FPC}vmovsd [eax + esi], xmm0;{$ELSE}db $C5,$FB,$11,$04,$30;{$ENDIF}
@nextLine2:
// next line:
add eax, edx;
add ecx, srcLineWidth;
// loop y end
dec ebx;
jnz @@addforyloop;
// epilog claenup stack
{$IFDEF FPC}vzeroupper;{$ELSE}db $C5,$F8,$77;{$ENDIF}
pop esi;
pop edi;
pop ebx;
end;
procedure AVXMatrixNormalizeRowUnAligned(dest : PDouble; const destLineWidth : TASMNativeInt; Src : PDouble; const srcLineWidth : TASMNativeInt; width, height : TASMNativeInt);
var tmp : double;
asm
// prolog - simulate stack
push ebx;
push edi;
push esi;
//iters := -width*sizeof(double);
mov edi, width;
imul edi, -8;
// helper registers for the mt1, mt2 and dest pointers
sub eax, edi;
sub ecx, edi;
mov ebx, height;
@@addforyloop:
{$IFDEF FPC}vxorpd xmm0, xmm0, xmm0;{$ELSE}db $C5,$F9,$57,$C0;{$ENDIF}
{$IFDEF FPC}vxorpd ymm1, ymm1, ymm1;{$ELSE}db $C5,$F5,$57,$C9;{$ENDIF}
// for x := 0 to w - 1;
// prepare for reverse loop
mov esi, edi;
@addforxloop:
add esi, 128;
jg @loopEnd;
// prefetch data...
// prefetch [eax + esi];
// mul add:
{$IFDEF FPC}vmovupd ymm3, [ecx + esi - 128];{$ELSE}db $C5,$FD,$10,$5C,$31,$80;{$ENDIF}
{$IFDEF FPC}vmulpd ymm3, ymm3, ymm3;{$ELSE}db $C5,$E5,$59,$DB;{$ENDIF}
{$IFDEF FPC}vaddpd ymm1, ymm1, ymm3;{$ELSE}db $C5,$F5,$58,$CB;{$ENDIF}
{$IFDEF FPC}vmovupd ymm2, [ecx + esi - 96];{$ELSE}db $C5,$FD,$10,$54,$31,$A0;{$ENDIF}
{$IFDEF FPC}vmulpd ymm2, ymm2, ymm2;{$ELSE}db $C5,$ED,$59,$D2;{$ENDIF}
{$IFDEF FPC}vaddpd ymm1, ymm1, ymm2;{$ELSE}db $C5,$F5,$58,$CA;{$ENDIF}
{$IFDEF FPC}vmovupd ymm3, [ecx + esi - 64];{$ELSE}db $C5,$FD,$10,$5C,$31,$C0;{$ENDIF}
{$IFDEF FPC}vmulpd ymm3, ymm3, ymm3;{$ELSE}db $C5,$E5,$59,$DB;{$ENDIF}
{$IFDEF FPC}vaddpd ymm1, ymm1, ymm3;{$ELSE}db $C5,$F5,$58,$CB;{$ENDIF}
{$IFDEF FPC}vmovupd ymm2, [ecx + esi - 32];{$ELSE}db $C5,$FD,$10,$54,$31,$E0;{$ENDIF}
{$IFDEF FPC}vmulpd ymm2, ymm2, ymm2;{$ELSE}db $C5,$ED,$59,$D2;{$ENDIF}
{$IFDEF FPC}vaddpd ymm1, ymm1, ymm2;{$ELSE}db $C5,$F5,$58,$CA;{$ENDIF}
jmp @addforxloop
@loopEnd:
sub esi, 128;
jz @nextLine;
@addforxloop2:
add esi, 16;
jg @@addforxloop2end;
{$IFDEF FPC}vmovupd xmm3, [ecx + esi - 16];{$ELSE}db $C5,$F9,$10,$5C,$31,$F0;{$ENDIF}
{$IFDEF FPC}vmulpd xmm3, xmm3, xmm3;{$ELSE}db $C5,$E1,$59,$DB;{$ENDIF}
{$IFDEF FPC}vaddpd xmm0, xmm0, xmm3;{$ELSE}db $C5,$F9,$58,$C3;{$ENDIF}
jmp @addforxloop2;
@@addforxloop2end:
sub esi, 16;
jz @nextLine;
{$IFDEF FPC}vmovsd xmm3, [ecx + esi];{$ELSE}db $C5,$FB,$10,$1C,$31;{$ENDIF}
{$IFDEF FPC}vmulsd xmm3, xmm3, xmm3;{$ELSE}db $C5,$E3,$59,$DB;{$ENDIF}
{$IFDEF FPC}vaddsd xmm0, xmm0, xmm3;{$ELSE}db $C5,$FB,$58,$C3;{$ENDIF}
@nextLine:
// build result
{$IFDEF FPC}vextractf128 xmm2, ymm1, 1;{$ELSE}db $C4,$E3,$7D,$19,$CA,$01;{$ENDIF}
{$IFDEF FPC}vhaddpd xmm1, xmm1, xmm2;{$ELSE}db $C5,$F1,$7C,$CA;{$ENDIF}
{$IFDEF FPC}vaddpd xmm0, xmm0, xmm1;{$ELSE}db $C5,$F9,$58,$C1;{$ENDIF}
{$IFDEF FPC}vhaddpd xmm0, xmm0, xmm0;{$ELSE}db $C5,$F9,$7C,$C0;{$ENDIF}
// build result
{$IFDEF FPC}vsqrtsd xmm0, xmm0, xmm0;{$ELSE}db $C5,$FB,$51,$C0;{$ENDIF}
//1/sqrt(norm)
lea esi, cLocOne;
{$IFDEF FPC}vmovsd xmm1, [esi];{$ELSE}db $C5,$FB,$10,$0E;{$ENDIF}
{$IFDEF FPC}vdivsd xmm1, xmm1, xmm0;{$ELSE}db $C5,$F3,$5E,$C8;{$ENDIF}
lea esi, tmp;
{$IFDEF FPC}vmovsd [esi], xmm1;{$ELSE}db $C5,$FB,$11,$0E;{$ENDIF}
{$IFDEF FPC}vbroadcastsd ymm2, [esi]; {$ELSE}db $C4,$E2,$7D,$19,$16;{$ENDIF} // need avx2
// normalize
mov esi, edi;
@addforxloop3:
add esi, 128;
jg @loopEnd2;
// prefetch data...
// prefetch [ecx + esi];
// prefetchw [eax + esi];
// mult:
{$IFDEF FPC}vmovupd ymm0, [ecx + esi - 128];{$ELSE}db $C5,$FD,$10,$44,$31,$80;{$ENDIF}
{$IFDEF FPC}vmulpd ymm0, ymm0, ymm2;{$ELSE}db $C5,$FD,$59,$C2;{$ENDIF}
{$IFDEF FPC}vmovupd [eax + esi - 128], ymm0;{$ELSE}db $C5,$FD,$11,$44,$30,$80;{$ENDIF}
{$IFDEF FPC}vmovupd ymm1, [ecx + esi - 96];{$ELSE}db $C5,$FD,$10,$4C,$31,$A0;{$ENDIF}
{$IFDEF FPC}vmulpd ymm1, ymm1, ymm2;{$ELSE}db $C5,$F5,$59,$CA;{$ENDIF}
{$IFDEF FPC}vmovupd [eax + esi - 96], ymm1;{$ELSE}db $C5,$FD,$11,$4C,$30,$A0;{$ENDIF}
{$IFDEF FPC}vmovupd ymm0, [ecx + esi - 64];{$ELSE}db $C5,$FD,$10,$44,$31,$C0;{$ENDIF}
{$IFDEF FPC}vmulpd ymm0, ymm0, ymm2;{$ELSE}db $C5,$FD,$59,$C2;{$ENDIF}
{$IFDEF FPC}vmovupd [eax + esi - 64], ymm0;{$ELSE}db $C5,$FD,$11,$44,$30,$C0;{$ENDIF}
{$IFDEF FPC}vmovupd ymm1, [ecx + esi - 32];{$ELSE}db $C5,$FD,$10,$4C,$31,$E0;{$ENDIF}
{$IFDEF FPC}vmulpd ymm1, ymm1, ymm2;{$ELSE}db $C5,$F5,$59,$CA;{$ENDIF}
{$IFDEF FPC}vmovupd [eax + esi - 32], ymm1;{$ELSE}db $C5,$FD,$11,$4C,$30,$E0;{$ENDIF}
jmp @addforxloop3
@loopEnd2:
sub esi, 128;
jz @nextLine2;
@addforxloop4:
add esi, 16;
jg @addforxloop4end;
{$IFDEF FPC}vmovupd xmm0, [ecx + esi - 16];{$ELSE}db $C5,$F9,$10,$44,$31,$F0;{$ENDIF}
{$IFDEF FPC}vmulpd xmm0, xmm0, xmm2;{$ELSE}db $C5,$F9,$59,$C2;{$ENDIF}
{$IFDEF FPC}vmovupd [eax + esi - 16], xmm0;{$ELSE}db $C5,$F9,$11,$44,$30,$F0;{$ENDIF}
jmp @addforxloop4;
@addforxloop4end:
sub esi, 16;
jz @nextLine2;
{$IFDEF FPC}vmovsd xmm0, [ecx + esi];{$ELSE}db $C5,$FB,$10,$04,$31;{$ENDIF}
{$IFDEF FPC}vmulsd xmm0, xmm0, xmm2;{$ELSE}db $C5,$FB,$59,$C2;{$ENDIF}
{$IFDEF FPC}vmovsd [eax + esi], xmm0;{$ELSE}db $C5,$FB,$11,$04,$30;{$ENDIF}
@nextLine2:
// next line:
add eax, edx;
add ecx, srcLineWidth;
// loop y end
dec ebx;
jnz @@addforyloop;
// epilog claenup stack
{$IFDEF FPC}vzeroupper;{$ELSE}db $C5,$F8,$77;{$ENDIF}
pop esi;
pop edi;
pop ebx;
end;
procedure AVXMatrixNormalizeColumnAligned(dest : PDouble; const destLineWidth : TASMNativeInt; Src : PDouble; const srcLineWidth : TASMNativeInt; width, height : TASMNativeInt);
var aDestLineWidth : TASMNativeInt;
asm
// prolog
push ebx;
push edi;
push esi;
// init
mov aDestLineWidth, edx;
// note: eax = dest, RDX = destLineWidth, edx = src
xor edi, edi;
sub edi, height;
imul edi, srcLineWidth;
// helper registers for the mt1 and dest pointers
sub ecx, edi;
mov esi, srcLineWidth;
// for x := 0 to width - 1:
mov ebx, Width;
sub ebx, 4;
jl @@addforxloop4end;
// 4 columns at once
@@addforxloop4:
{$IFDEF FPC}vxorpd ymm1, ymm1, ymm1;{$ELSE}db $C5,$F5,$57,$C9;{$ENDIF}
// for y := 0 to height - 1;
// prepare for reverse loop
mov edx, edi;
@addforyloop4:
{$IFDEF FPC}vmovapd ymm3, [ecx + edx];{$ELSE}db $C5,$FD,$28,$1C,$11;{$ENDIF}
{$IFDEF FPC}vmulpd ymm3, ymm3, ymm3;{$ELSE}db $C5,$E5,$59,$DB;{$ENDIF}
{$IFDEF FPC}vaddpd ymm1, ymm1, ymm3;{$ELSE}db $C5,$F5,$58,$CB;{$ENDIF}
add edx, esi;
jnz @addforyloop4;
// normalization factor
lea edx, cLocOne;
{$IFDEF FPC}vbroadcastsd ymm2, [edx];{$ELSE}db $C4,$E2,$7D,$19,$12;{$ENDIF}
{$IFDEF FPC}vsqrtpd ymm1, ymm1;{$ELSE}db $C5,$FD,$51,$C9;{$ENDIF}
{$IFDEF FPC}vdivpd ymm0, ymm2, ymm1;{$ELSE}db $C5,$ED,$5E,$C1;{$ENDIF}
push eax;
// normalize
mov edx, edi;
@addforyloop4_2:
{$IFDEF FPC}vmovapd ymm1, [ecx + edx];{$ELSE}db $C5,$FD,$28,$0C,$11;{$ENDIF}
{$IFDEF FPC}vmulpd ymm1, ymm1, ymm0;{$ELSE}db $C5,$F5,$59,$C8;{$ENDIF}
{$IFDEF FPC}vmovapd [eax], ymm1;{$ELSE}db $C5,$FD,$29,$08;{$ENDIF}
add eax, adestLineWidth;
add edx, esi;
jnz @addforyloop4_2;
// next columns:
pop eax;
add eax, 32;
add ecx, 32;
// loop x end
sub ebx, 4;
jge @@addforxloop4;
@@addforxloop4end:
add ebx, 4;
jz @@endProc;
sub ebx, 2;
jl @@lastcolumn;
{$IFDEF FPC}vxorpd xmm1, xmm1, xmm1;{$ELSE}db $C5,$F1,$57,$C9;{$ENDIF}
// for y := 0 to height - 1;
// prepare for reverse loop
mov edx, edi;
@addforyloop2:
{$IFDEF FPC}vmovapd xmm0, [ecx + edx];{$ELSE}db $C5,$F9,$28,$04,$11;{$ENDIF}
{$IFDEF FPC}vmulpd xmm0, xmm0, xmm0;{$ELSE}db $C5,$F9,$59,$C0;{$ENDIF}
{$IFDEF FPC}vaddpd xmm1, xmm1, xmm0;{$ELSE}db $C5,$F1,$58,$C8;{$ENDIF}
add edx, esi;
jnz @addforyloop2;
// normalization factor
lea edx, cLocOne;
{$IFDEF FPC}vmovddup xmm2, [edx];{$ELSE}db $C5,$FB,$12,$12;{$ENDIF}
{$IFDEF FPC}vsqrtpd xmm1, xmm1;{$ELSE}db $C5,$F9,$51,$C9;{$ENDIF}
{$IFDEF FPC}vdivpd xmm0, xmm2, xmm1;{$ELSE}db $C5,$E9,$5E,$C1;{$ENDIF}
// normalize
push eax;
mov edx, edi;
@addforyloop2_2:
{$IFDEF FPC}vmovapd xmm1, [ecx + edx];{$ELSE}db $C5,$F9,$28,$0C,$11;{$ENDIF}
{$IFDEF FPC}vmulpd xmm1, xmm1, xmm0;{$ELSE}db $C5,$F1,$59,$C8;{$ENDIF}
{$IFDEF FPC}vmovapd [eax], xmm1;{$ELSE}db $C5,$F9,$29,$08;{$ENDIF}
add eax, adestLineWidth;
add edx, esi;
jnz @addforyloop2_2;
// next columns:
pop eax;
add eax, 16;
add ecx, 16;
dec ebx;
jnz @@endProc;
@@lastcolumn:
{$IFDEF FPC}vxorpd xmm1, xmm1, xmm1;{$ELSE}db $C5,$F1,$57,$C9;{$ENDIF}
// for y := 0 to height - 1;
// prepare for reverse loop
mov edx, edi;
@addforyloop_1:
{$IFDEF FPC}vmovsd xmm0, [ecx + edx];{$ELSE}db $C5,$FB,$10,$04,$11;{$ENDIF}
{$IFDEF FPC}vmulsd xmm0, xmm0, xmm0;{$ELSE}db $C5,$FB,$59,$C0;{$ENDIF}
{$IFDEF FPC}vaddsd xmm1, xmm1, xmm0;{$ELSE}db $C5,$F3,$58,$C8;{$ENDIF}
add edx, esi;
jnz @addforyloop_1;
// build result
lea edx, cLocOne;
{$IFDEF FPC}vmovsd xmm2, [edx];{$ELSE}db $C5,$FB,$10,$12;{$ENDIF}
{$IFDEF FPC}vsqrtsd xmm1, xmm1, xmm1;{$ELSE}db $C5,$F3,$51,$C9;{$ENDIF}
{$IFDEF FPC}vdivsd xmm0, xmm2, xmm1;{$ELSE}db $C5,$EB,$5E,$C1;{$ENDIF}
// normalize last column
mov edx, edi;
@addforyloop2_1:
{$IFDEF FPC}vmovsd xmm1, [ecx + edx];{$ELSE}db $C5,$FB,$10,$0C,$11;{$ENDIF}
{$IFDEF FPC}vmulsd xmm1, xmm1, xmm0;{$ELSE}db $C5,$F3,$59,$C8;{$ENDIF}
{$IFDEF FPC}vmovsd [eax], xmm1;{$ELSE}db $C5,$FB,$11,$08;{$ENDIF}
add eax, adestLineWidth;
add edx, esi;
jnz @addforyloop2_1;
@@endProc:
// epilog
{$IFDEF FPC}vzeroupper;{$ELSE}db $C5,$F8,$77;{$ENDIF}
pop esi;
pop edi;
pop ebx;
end;
procedure AVXMatrixNormalizeColumnUnAligned(dest : PDouble; const destLineWidth : TASMNativeInt; Src : PDouble; const srcLineWidth : TASMNativeInt; width, height : TASMNativeInt);
var aDestLineWidth : TASMNativeInt;
asm
// prolog
push ebx;
push edi;
push esi;
// init
mov aDestLineWidth, edx;
// note: eax = dest, RDX = destLineWidth, edx = src
xor edi, edi;
sub edi, height;
imul edi, srcLineWidth;
// helper registers for the mt1 and dest pointers
sub ecx, edi;
mov esi, srcLineWidth;
// for x := 0 to width - 1:
mov ebx, Width;
sub ebx, 4;
jl @@addforxloop4end;
// 4 columns at once
@@addforxloop4:
{$IFDEF FPC}vxorpd ymm1, ymm1, ymm1;{$ELSE}db $C5,$F5,$57,$C9;{$ENDIF}
// for y := 0 to height - 1;
// prepare for reverse loop
mov edx, edi;
@addforyloop4:
{$IFDEF FPC}vmovupd ymm3, [ecx + edx];{$ELSE}db $C5,$FD,$10,$1C,$11;{$ENDIF}
{$IFDEF FPC}vmulpd ymm3, ymm3, ymm3;{$ELSE}db $C5,$E5,$59,$DB;{$ENDIF}
{$IFDEF FPC}vaddpd ymm1, ymm1, ymm3;{$ELSE}db $C5,$F5,$58,$CB;{$ENDIF}
add edx, esi;
jnz @addforyloop4;
// normalization factor
lea edx, cLocOne;
{$IFDEF FPC}vbroadcastsd ymm2, [edx];{$ELSE}db $C4,$E2,$7D,$19,$12;{$ENDIF}
{$IFDEF FPC}vsqrtpd ymm1, ymm1;{$ELSE}db $C5,$FD,$51,$C9;{$ENDIF}
{$IFDEF FPC}vdivpd ymm0, ymm2, ymm1;{$ELSE}db $C5,$ED,$5E,$C1;{$ENDIF}
push eax;
// normalize
mov edx, edi;
@addforyloop4_2:
{$IFDEF FPC}vmovupd ymm1, [ecx + edx];{$ELSE}db $C5,$FD,$10,$0C,$11;{$ENDIF}
{$IFDEF FPC}vmulpd ymm1, ymm1, ymm0;{$ELSE}db $C5,$F5,$59,$C8;{$ENDIF}
{$IFDEF FPC}vmovupd [eax], ymm1;{$ELSE}db $C5,$FD,$11,$08;{$ENDIF}
add eax, adestLineWidth;
add edx, esi;
jnz @addforyloop4_2;
// next columns:
pop eax;
add eax, 32;
add ecx, 32;
// loop x end
sub ebx, 4;
jge @@addforxloop4;
@@addforxloop4end:
add ebx, 4;
jz @@endProc;
sub ebx, 2;
jl @@lastcolumn;
{$IFDEF FPC}vxorpd xmm1, xmm1, xmm1;{$ELSE}db $C5,$F1,$57,$C9;{$ENDIF}
// for y := 0 to height - 1;
// prepare for reverse loop
mov edx, edi;
@addforyloop2:
{$IFDEF FPC}vmovupd xmm0, [ecx + edx];{$ELSE}db $C5,$F9,$10,$04,$11;{$ENDIF}
{$IFDEF FPC}vmulpd xmm0, xmm0, xmm0;{$ELSE}db $C5,$F9,$59,$C0;{$ENDIF}
{$IFDEF FPC}vaddpd xmm1, xmm1, xmm0;{$ELSE}db $C5,$F1,$58,$C8;{$ENDIF}
add edx, esi;
jnz @addforyloop2;
// normalization factor
lea edx, cLocOne;
{$IFDEF FPC}vmovddup xmm2, [edx];{$ELSE}db $C5,$FB,$12,$12;{$ENDIF}
{$IFDEF FPC}vsqrtpd xmm1, xmm1;{$ELSE}db $C5,$F9,$51,$C9;{$ENDIF}
{$IFDEF FPC}vdivpd xmm0, xmm2, xmm1;{$ELSE}db $C5,$E9,$5E,$C1;{$ENDIF}
// normalize
push eax;
mov edx, edi;
@addforyloop2_2:
{$IFDEF FPC}vmovupd xmm1, [ecx + edx];{$ELSE}db $C5,$F9,$10,$0C,$11;{$ENDIF}
{$IFDEF FPC}vmulpd xmm1, xmm1, xmm0;{$ELSE}db $C5,$F1,$59,$C8;{$ENDIF}
{$IFDEF FPC}vmovupd [eax], xmm1;{$ELSE}db $C5,$F9,$11,$08;{$ENDIF}
add eax, adestLineWidth;
add edx, esi;
jnz @addforyloop2_2;
// next columns:
pop eax;
add eax, 16;
add ecx, 16;
dec ebx;
jnz @@endProc;
@@lastcolumn:
{$IFDEF FPC}vxorpd xmm1, xmm1, xmm1;{$ELSE}db $C5,$F1,$57,$C9;{$ENDIF}
// for y := 0 to height - 1;
// prepare for reverse loop
mov edx, edi;
@addforyloop_1:
{$IFDEF FPC}vmovsd xmm0, [ecx + edx];{$ELSE}db $C5,$FB,$10,$04,$11;{$ENDIF}
{$IFDEF FPC}vmulsd xmm0, xmm0, xmm0;{$ELSE}db $C5,$FB,$59,$C0;{$ENDIF}
{$IFDEF FPC}vaddsd xmm1, xmm1, xmm0;{$ELSE}db $C5,$F3,$58,$C8;{$ENDIF}
add edx, esi;
jnz @addforyloop_1;
// build result
lea edx, cLocOne;
{$IFDEF FPC}vmovsd xmm2, [edx];{$ELSE}db $C5,$FB,$10,$12;{$ENDIF}
{$IFDEF FPC}vsqrtsd xmm1, xmm1, xmm1;{$ELSE}db $C5,$F3,$51,$C9;{$ENDIF}
{$IFDEF FPC}vdivsd xmm0, xmm2, xmm1;{$ELSE}db $C5,$EB,$5E,$C1;{$ENDIF}
// normalize last column
mov edx, edi;
@addforyloop2_1:
{$IFDEF FPC}vmovsd xmm1, [ecx + edx];{$ELSE}db $C5,$FB,$10,$0C,$11;{$ENDIF}
{$IFDEF FPC}vmulsd xmm1, xmm1, xmm0;{$ELSE}db $C5,$F3,$59,$C8;{$ENDIF}
{$IFDEF FPC}vmovsd [eax], xmm1;{$ELSE}db $C5,$FB,$11,$08;{$ENDIF}
add eax, adestLineWidth;
add edx, esi;
jnz @addforyloop2_1;
@@endProc:
// epilog
{$IFDEF FPC}vzeroupper;{$ELSE}db $C5,$F8,$77;{$ENDIF}
pop esi;
pop edi;
pop ebx;
end;
{$ENDIF}
end.