-
Notifications
You must be signed in to change notification settings - Fork 33
/
AVXMoveOperationsx64.pas
617 lines (458 loc) · 23.1 KB
/
AVXMoveOperationsx64.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
// ###################################################################
// #### This file is part of the mathematics library project, and is
// #### offered under the licence agreement described on
// #### http://www.mrsoft.org/
// ####
// #### Copyright:(c) 2017, 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 AVXMoveOperationsx64;
// #################################################
// #### SSE optimized move oprationes
// #################################################
interface
{$I 'mrMath_CPU.inc'}
{$IFDEF x64}
procedure AVXMatrixCopyAligned(Dest : PDouble; const destLineWidth : NativeInt; src : PDouble; const srcLineWidth : NativeInt; {$ifdef UNIX}unixWidth{$ELSE}width{$endif}, {$ifdef UNIX}unixHeight{$ELSE}height{$endif} : NativeInt); {$IFDEF FPC}assembler;{$ENDIF}
procedure AVXMatrixCopyUnAligned(Dest : PDouble; const destLineWidth : NativeInt; src : PDouble; const srcLineWidth : NativeInt; {$ifdef UNIX}unixWidth{$ELSE}width{$endif}, {$ifdef UNIX}unixHeight{$ELSE}height{$endif} : NativeInt); {$IFDEF FPC}assembler;{$ENDIF}
procedure AVXRowSwapAligned(A, B : PDouble; width : NativeInt); {$IFDEF FPC}assembler;{$ENDIF}
procedure AVXRowSwapUnAligned(A, B : PDouble; width : NativeInt); {$IFDEF FPC}assembler;{$ENDIF}
procedure AVXInitMemAligned(A : PDouble; NumBytes : NativeInt; Value : double); {$IFDEF FPC}assembler;{$ENDIF}
procedure AVXMatrixInitAligned( dest : PDouble; const destLineWidth : NativeInt; Width, Height : NativeInt; {$IFDEF UNIX}unixValue{$ELSE}Value{$ENDIF} : double); {$IFDEF FPC} assembler; {$ELSE} register; {$ENDIF}
procedure AVXMatrixInitUnAligned( dest : PDouble; const destLineWidth : NativeInt; Width, Height : NativeInt; {$IFDEF UNIX}unixValue{$ELSE}Value{$ENDIF} : double); {$IFDEF FPC} assembler; {$ELSE} register; {$ENDIF}
{$ENDIF}
implementation
{$IFDEF x64}
// rcx = dest, rdx = destLineWidth; r8 = width, r9 = height
procedure AVXMatrixInitAligned( dest : PDouble; const destLineWidth : NativeInt; Width, Height : NativeInt; {$IFDEF UNIX}unixValue{$ELSE}Value{$ENDIF} : double); {$IFDEF FPC} assembler; {$ELSE} register; {$ENDIF}
{$IFDEF UNIX}
var value : double;
{$ENDIF}
asm
{$IFDEF UNIX}
// Linux uses a diffrent ABI -> copy over the registers so they meet with winABI
// (note that the 5th and 6th parameter are are on the stack)
// The parameters are passed in the following order:
// RDI, RSI, RDX, RCX -> mov to RCX, RDX, R8, R9
mov r8, rdx;
mov r9, rcx;
mov rcx, rdi;
mov rdx, rsi;
movsd value, unixValue;
{$ENDIF}
lea rax, value;
{$IFDEF AVXSUP}vbroadcastsd ymm1, [rax]; {$ELSE}db $C4,$E2,$7D,$19,$08;{$ENDIF}
//iters := -width*sizeof(double);
imul r8, -8;
// helper registers for the src and dest pointers
sub rcx, r8;
// for y := 0 to height - 1:
@@addforyloop:
// for x := 0 to w - 1;
// prepare for reverse loop
mov rax, r8;
@addforxloop:
add rax, 128;
jg @loopEnd;
// move:
{$IFDEF AVXSUP}vmovapd [rcx + rax - 128], ymm1; {$ELSE}db $C5,$FD,$29,$4C,$01,$80;{$ENDIF}
{$IFDEF AVXSUP}vmovapd [rcx + rax - 96], ymm1; {$ELSE}db $C5,$FD,$29,$4C,$01,$A0;{$ENDIF}
{$IFDEF AVXSUP}vmovapd [rcx + rax - 64], ymm1; {$ELSE}db $C5,$FD,$29,$4C,$01,$C0;{$ENDIF}
{$IFDEF AVXSUP}vmovapd [rcx + rax - 32], ymm1; {$ELSE}db $C5,$FD,$29,$4C,$01,$E0;{$ENDIF}
jmp @addforxloop
@loopEnd:
sub rax, 128;
jz @nextLine;
@addforxloop2:
add rax, 16;
jg @loopEnd2;
{$IFDEF AVXSUP}vmovapd [rcx + rax - 16], xmm1; {$ELSE}db $C5,$F9,$29,$4C,$01,$F0;{$ENDIF}
jmp @addforxloop2;
@loopEnd2:
sub rax, 16;
jz @nextLine;
// last element
{$IFDEF AVXSUP}vmovsd [rcx + rax], xmm1; {$ELSE}db $C5,$FB,$11,$0C,$01;{$ENDIF}
@nextLine:
// next line:
add rcx, rdx;
// loop y end
dec r9;
jnz @@addforyloop;
{$IFDEF AVXSUP}vzeroupper; {$ELSE}db $C5,$F8,$77;{$ENDIF}
end;
// rcx = dest, rdx = destLineWidth; r8 = width, r9 = height
procedure AVXMatrixInitUnAligned( dest : PDouble; const destLineWidth : NativeInt; Width, Height : NativeInt; {$IFDEF UNIX}unixValue{$ELSE}Value{$ENDIF} : double); {$IFDEF FPC} assembler; {$ELSE} register; {$ENDIF}
{$IFDEF UNIX}
var value : double;
{$ENDIF}
asm
{$IFDEF UNIX}
// Linux uses a diffrent ABI -> copy over the registers so they meet with winABI
// (note that the 5th and 6th parameter are are on the stack)
// The parameters are passed in the following order:
// RDI, RSI, RDX, RCX -> mov to RCX, RDX, R8, R9
mov r8, rdx;
mov r9, rcx;
mov rcx, rdi;
mov rdx, rsi;
movsd value, unixValue;
{$ENDIF}
lea rax, value;
{$IFDEF AVXSUP}vbroadcastsd ymm1, [rax]; {$ELSE}db $C4,$E2,$7D,$19,$08;{$ENDIF}
//iters := -width*sizeof(double);
imul r8, -8;
// helper registers for the src and dest pointers
sub rcx, r8;
// for y := 0 to height - 1:
@@addforyloop:
// for x := 0 to w - 1;
// prepare for reverse loop
mov rax, r8;
@addforxloop:
add rax, 128;
jg @loopEnd;
// move:
{$IFDEF AVXSUP}vmovupd [rcx + rax - 128], ymm1; {$ELSE}db $C5,$FD,$11,$4C,$01,$80;{$ENDIF}
{$IFDEF AVXSUP}vmovupd [rcx + rax - 96], ymm1; {$ELSE}db $C5,$FD,$11,$4C,$01,$A0;{$ENDIF}
{$IFDEF AVXSUP}vmovupd [rcx + rax - 64], ymm1; {$ELSE}db $C5,$FD,$11,$4C,$01,$C0;{$ENDIF}
{$IFDEF AVXSUP}vmovupd [rcx + rax - 32], ymm1; {$ELSE}db $C5,$FD,$11,$4C,$01,$E0;{$ENDIF}
jmp @addforxloop
@loopEnd:
sub rax, 128;
jz @nextLine;
@addforxloop2:
add rax, 16;
jg @loopEnd2;
{$IFDEF AVXSUP}vmovupd [rcx + rax - 16], xmm1; {$ELSE}db $C5,$F9,$11,$4C,$01,$F0;{$ENDIF}
jmp @addforxloop2;
@loopEnd2:
sub rax, 16;
jz @nextLine;
// last element
{$IFDEF AVXSUP}vmovsd [rcx + rax], xmm1; {$ELSE}db $C5,$FB,$11,$0C,$01;{$ENDIF}
@nextLine:
// next line:
add rcx, rdx;
// loop y end
dec r9;
jnz @@addforyloop;
{$IFDEF AVXSUP}vzeroupper; {$ELSE}db $C5,$F8,$77;{$ENDIF}
end;
// uses non temporal moves so the cache is not poisned
// rcx = A, rdx = NumBytes;
procedure AVXInitMemAligned(A : PDouble; NumBytes : NativeInt; Value : double); {$IFDEF FPC}assembler;{$ENDIF}
var tmp : Double;
asm
{$IFDEF UNIX}
// Linux uses a diffrent ABI -> copy over the registers so they meet with winABI
// (note that the 5th and 6th parameter are are on the stack)
// The parameters are passed in the following order:
// RDI, RSI, RDX, RCX -> mov to RCX, RDX, R8, R9
mov r8, rdx;
mov r9, rcx;
mov rcx, rdi;
mov rdx, rsi;
{$ENDIF}
movsd tmp, value;
lea rax, tmp;
{$IFDEF AVXSUP}vbroadcastsd ymm1, [rax]; {$ELSE}db $C4,$E2,$7D,$19,$08;{$ENDIF}
imul rdx, -1;
sub rcx, rdx;
@@loopUnrolled:
add rdx, 128;
jg @@loopUnrolledEnd;
{$IFDEF AVXSUP}vmovapd [rcx + rdx - 128], ymm1; {$ELSE}db $C5,$FD,$29,$4C,$11,$80;{$ENDIF}
{$IFDEF AVXSUP}vmovapd [rcx + rdx - 96], ymm1; {$ELSE}db $C5,$FD,$29,$4C,$11,$A0;{$ENDIF}
{$IFDEF AVXSUP}vmovapd [rcx + rdx - 64], ymm1; {$ELSE}db $C5,$FD,$29,$4C,$11,$C0;{$ENDIF}
{$IFDEF AVXSUP}vmovapd [rcx + rdx - 32], ymm1; {$ELSE}db $C5,$FD,$29,$4C,$11,$E0;{$ENDIF}
jmp @@loopUnrolled;
@@loopUnrolledEnd:
sub rdx, 128;
jz @@exitProc;
add rdx, 32;
jg @@beginloop3;
@@loop2:
{$IFDEF AVXSUP}vmovapd [rcx + rdx - 32], xmm1; {$ELSE}db $C5,$F9,$29,$4C,$11,$E0;{$ENDIF}
{$IFDEF AVXSUP}vmovapd [rcx + rdx - 16], xmm1; {$ELSE}db $C5,$F9,$29,$4C,$11,$F0;{$ENDIF}
add rdx, 32;
jl @@loop2;
@@beginloop3:
sub rdx, 32;
jz @@exitProc;
@@loop3:
{$IFDEF AVXSUP}vmovsd [rcx + rdx], xmm1; {$ELSE}db $C5,$FB,$11,$0C,$11;{$ENDIF}
add rdx, 8;
jnz @@loop3;
@@exitProc:
{$IFDEF AVXSUP}vzeroupper; {$ELSE}db $C5,$F8,$77;{$ENDIF}
end;
procedure AVXRowSwapAligned(A, B : PDouble; width : NativeInt); {$IFDEF FPC}assembler;{$ENDIF}
asm
{$IFDEF UNIX}
// Linux uses a diffrent ABI -> copy over the registers so they meet with winABI
// (note that the 5th and 6th parameter are are on the stack)
// The parameters are passed in the following order:
// RDI, RSI, RDX, RCX -> mov to RCX, RDX, R8, R9
mov r8, rdx;
mov r9, rcx;
mov rcx, rdi;
mov rdx, rsi;
{$ENDIF}
// note: RCX = a, RDX = b, R8 = width
imul r8, -8;
sub rcx, r8;
sub rdx, r8;
@unrolloop:
add r8, 128;
jg @unrolloopend;
// prefetch data...
// prefetchw [rcx + r10];
// prefetchw [rdx + r10];
{$IFDEF AVXSUP}vmovdqa ymm0, [rcx + r8 - 128]; {$ELSE}db $C4,$A1,$7D,$6F,$44,$01,$80;{$ENDIF}
{$IFDEF AVXSUP}vmovdqa ymm1, [rdx + r8 - 128]; {$ELSE}db $C4,$A1,$7D,$6F,$4C,$02,$80;{$ENDIF}
{$IFDEF AVXSUP}vmovdqa [rcx + r8 - 128], ymm1; {$ELSE}db $C4,$A1,$7D,$7F,$4C,$01,$80;{$ENDIF}
{$IFDEF AVXSUP}vmovdqa [rdx + r8 - 128], ymm0; {$ELSE}db $C4,$A1,$7D,$7F,$44,$02,$80;{$ENDIF}
{$IFDEF AVXSUP}vmovdqa ymm2, [rcx + r8 - 96]; {$ELSE}db $C4,$A1,$7D,$6F,$54,$01,$A0;{$ENDIF}
{$IFDEF AVXSUP}vmovdqa ymm3, [rdx + r8 - 96]; {$ELSE}db $C4,$A1,$7D,$6F,$5C,$02,$A0;{$ENDIF}
{$IFDEF AVXSUP}vmovdqa [rcx + r8 - 96], ymm3; {$ELSE}db $C4,$A1,$7D,$7F,$5C,$01,$A0;{$ENDIF}
{$IFDEF AVXSUP}vmovdqa [rdx + r8 - 96], ymm2; {$ELSE}db $C4,$A1,$7D,$7F,$54,$02,$A0;{$ENDIF}
{$IFDEF AVXSUP}vmovdqa ymm0, [rcx + r8 - 64]; {$ELSE}db $C4,$A1,$7D,$6F,$44,$01,$C0;{$ENDIF}
{$IFDEF AVXSUP}vmovdqa ymm1, [rdx + r8 - 64]; {$ELSE}db $C4,$A1,$7D,$6F,$4C,$02,$C0;{$ENDIF}
{$IFDEF AVXSUP}vmovdqa [rcx + r8 - 64], ymm1; {$ELSE}db $C4,$A1,$7D,$7F,$4C,$01,$C0;{$ENDIF}
{$IFDEF AVXSUP}vmovdqa [rdx + r8 - 64], ymm0; {$ELSE}db $C4,$A1,$7D,$7F,$44,$02,$C0;{$ENDIF}
{$IFDEF AVXSUP}vmovdqa ymm2, [rcx + r8 - 32]; {$ELSE}db $C4,$A1,$7D,$6F,$54,$01,$E0;{$ENDIF}
{$IFDEF AVXSUP}vmovdqa ymm3, [rdx + r8 - 32]; {$ELSE}db $C4,$A1,$7D,$6F,$5C,$02,$E0;{$ENDIF}
{$IFDEF AVXSUP}vmovdqa [rcx + r8 - 32], ymm3; {$ELSE}db $C4,$A1,$7D,$7F,$5C,$01,$E0;{$ENDIF}
{$IFDEF AVXSUP}vmovdqa [rdx + r8 - 32], ymm2; {$ELSE}db $C4,$A1,$7D,$7F,$54,$02,$E0;{$ENDIF}
jmp @unrolloop;
@unrolloopend:
sub r8, 128;
jz @endfunc;
@loop2:
add r8, 16;
jg @loop2End;
{$IFDEF AVXSUP}vmovdqa xmm0, [rcx + r8 - 16]; {$ELSE}db $C4,$A1,$79,$6F,$44,$01,$F0;{$ENDIF}
{$IFDEF AVXSUP}vmovdqa xmm1, [rdx + r8 - 16]; {$ELSE}db $C4,$A1,$79,$6F,$4C,$02,$F0;{$ENDIF}
{$IFDEF AVXSUP}vmovdqa [rcx + r8 - 16], xmm1; {$ELSE}db $C4,$A1,$79,$7F,$4C,$01,$F0;{$ENDIF}
{$IFDEF AVXSUP}vmovdqa [rdx + r8 - 16], xmm0; {$ELSE}db $C4,$A1,$79,$7F,$44,$02,$F0;{$ENDIF}
jmp @loop2;
@loop2End:
sub r8, 16;
jz @endfunc;
{$IFDEF AVXSUP}vmovsd xmm0, [rcx + r8]; {$ELSE}db $C4,$A1,$7B,$10,$04,$01;{$ENDIF}
{$IFDEF AVXSUP}vmovsd xmm1, [rdx + r8]; {$ELSE}db $C4,$A1,$7B,$10,$0C,$02;{$ENDIF}
{$IFDEF AVXSUP}vmovsd [rcx + r8], xmm1; {$ELSE}db $C4,$A1,$7B,$11,$0C,$01;{$ENDIF}
{$IFDEF AVXSUP}vmovsd [rdx + r8], xmm0; {$ELSE}db $C4,$A1,$7B,$11,$04,$02;{$ENDIF}
@endfunc:
{$IFDEF AVXSUP}vzeroupper; {$ELSE}db $C5,$F8,$77;{$ENDIF}
end;
procedure AVXRowSwapUnAligned(A, B : PDouble; width : NativeInt); {$IFDEF FPC}assembler;{$ENDIF}
asm
{$IFDEF UNIX}
// Linux uses a diffrent ABI -> copy over the registers so they meet with winABI
// (note that the 5th and 6th parameter are are on the stack)
// The parameters are passed in the following order:
// RDI, RSI, RDX, RCX -> mov to RCX, RDX, R8, R9
mov r8, rdx;
mov r9, rcx;
mov rcx, rdi;
mov rdx, rsi;
{$ENDIF}
// note: RCX = a, RDX = b, R8 = width
imul r8, -8;
sub rcx, r8;
sub rdx, r8;
@unrolloop:
add r8, 128;
jg @unrolloopend;
// prefetch data...
// prefetchw [rcx + r10];
// prefetchw [rdx + r10];
{$IFDEF AVXSUP}vmovdqu ymm0, [rcx + r8 - 128]; {$ELSE}db $C4,$A1,$7E,$6F,$44,$01,$80;{$ENDIF}
{$IFDEF AVXSUP}vmovdqu ymm1, [rdx + r8 - 128]; {$ELSE}db $C4,$A1,$7E,$6F,$4C,$02,$80;{$ENDIF}
{$IFDEF AVXSUP}vmovdqu [rcx + r8 - 128], ymm1; {$ELSE}db $C4,$A1,$7E,$7F,$4C,$01,$80;{$ENDIF}
{$IFDEF AVXSUP}vmovdqu [rdx + r8 - 128], ymm0; {$ELSE}db $C4,$A1,$7E,$7F,$44,$02,$80;{$ENDIF}
{$IFDEF AVXSUP}vmovdqu ymm2, [rcx + r8 - 96]; {$ELSE}db $C4,$A1,$7E,$6F,$54,$01,$A0;{$ENDIF}
{$IFDEF AVXSUP}vmovdqu ymm3, [rdx + r8 - 96]; {$ELSE}db $C4,$A1,$7E,$6F,$5C,$02,$A0;{$ENDIF}
{$IFDEF AVXSUP}vmovdqu [rcx + r8 - 96], ymm3; {$ELSE}db $C4,$A1,$7E,$7F,$5C,$01,$A0;{$ENDIF}
{$IFDEF AVXSUP}vmovdqu [rdx + r8 - 96], ymm2; {$ELSE}db $C4,$A1,$7E,$7F,$54,$02,$A0;{$ENDIF}
{$IFDEF AVXSUP}vmovdqu ymm0, [rcx + r8 - 64]; {$ELSE}db $C4,$A1,$7E,$6F,$44,$01,$C0;{$ENDIF}
{$IFDEF AVXSUP}vmovdqu ymm1, [rdx + r8 - 64]; {$ELSE}db $C4,$A1,$7E,$6F,$4C,$02,$C0;{$ENDIF}
{$IFDEF AVXSUP}vmovdqu [rcx + r8 - 64], ymm1; {$ELSE}db $C4,$A1,$7E,$7F,$4C,$01,$C0;{$ENDIF}
{$IFDEF AVXSUP}vmovdqu [rdx + r8 - 64], ymm0; {$ELSE}db $C4,$A1,$7E,$7F,$44,$02,$C0;{$ENDIF}
{$IFDEF AVXSUP}vmovdqu ymm2, [rcx + r8 - 32]; {$ELSE}db $C4,$A1,$7E,$6F,$54,$01,$E0;{$ENDIF}
{$IFDEF AVXSUP}vmovdqu ymm3, [rdx + r8 - 32]; {$ELSE}db $C4,$A1,$7E,$6F,$5C,$02,$E0;{$ENDIF}
{$IFDEF AVXSUP}vmovdqu [rcx + r8 - 32], ymm3; {$ELSE}db $C4,$A1,$7E,$7F,$5C,$01,$E0;{$ENDIF}
{$IFDEF AVXSUP}vmovdqu [rdx + r8 - 32], ymm2; {$ELSE}db $C4,$A1,$7E,$7F,$54,$02,$E0;{$ENDIF}
jmp @unrolloop;
@unrolloopend:
sub r8, 128;
jz @endfunc;
@loop2:
add r8, 16;
jg @loop2End;
{$IFDEF AVXSUP}vmovdqu xmm0, [rcx + r8 - 16]; {$ELSE}db $C4,$A1,$7A,$6F,$44,$01,$F0;{$ENDIF}
{$IFDEF AVXSUP}vmovdqu xmm1, [rdx + r8 - 16]; {$ELSE}db $C4,$A1,$7A,$6F,$4C,$02,$F0;{$ENDIF}
{$IFDEF AVXSUP}vmovdqu [rcx + r8 - 16], xmm1; {$ELSE}db $C4,$A1,$7A,$7F,$4C,$01,$F0;{$ENDIF}
{$IFDEF AVXSUP}vmovdqu [rdx + r8 - 16], xmm0; {$ELSE}db $C4,$A1,$7A,$7F,$44,$02,$F0;{$ENDIF}
jmp @loop2;
@loop2End:
sub r8, 16;
jz @endfunc;
{$IFDEF AVXSUP}vmovsd xmm0, [rcx + r8]; {$ELSE}db $C4,$A1,$7B,$10,$04,$01;{$ENDIF}
{$IFDEF AVXSUP}vmovsd xmm1, [rdx + r8]; {$ELSE}db $C4,$A1,$7B,$10,$0C,$02;{$ENDIF}
{$IFDEF AVXSUP}vmovsd [rcx + r8], xmm1; {$ELSE}db $C4,$A1,$7B,$11,$0C,$01;{$ENDIF}
{$IFDEF AVXSUP}vmovsd [rdx + r8], xmm0; {$ELSE}db $C4,$A1,$7B,$11,$04,$02;{$ENDIF}
@endfunc:
{$IFDEF AVXSUP}vzeroupper; {$ELSE}db $C5,$F8,$77;{$ENDIF}
end;
procedure AVXMatrixCopyAligned(Dest : PDouble; const destLineWidth : NativeInt; src : PDouble; const srcLineWidth : NativeInt; {$ifdef UNIX}unixWidth{$ELSE}width{$endif}, {$ifdef UNIX}unixHeight{$ELSE}height{$endif} : NativeInt); {$IFDEF FPC}assembler;{$ENDIF}
{$ifdef UNIX}
var width : NativeInt;
height : NativeInt;
{$ENDIF}
asm
{$IFDEF UNIX}
// Linux uses a diffrent ABI -> copy over the registers so they meet with winABI
// (note that the 5th and 6th parameter are are on the stack)
// The parameters are passed in the following order:
// RDI, RSI, RDX, RCX -> mov to RCX, RDX, R8, R9
mov width, r8;
mov height, r9;
mov r8, rdx;
mov r9, rcx;
mov rcx, rdi;
mov rdx, rsi;
{$ENDIF}
// note: RCX = dest, RDX = destLineWidth, R8 = src, R9 = srcLineWidth
//iters := -width*sizeof(double);
mov r10, width;
imul r10, -8;
// helper registers for the mt1, mt2 and dest pointers
sub r8, r10;
sub rcx, r10;
// for y := 0 to height - 1:
mov r11, Height;
@@addforyloop:
// for x := 0 to w - 1;
// prepare for reverse loop
mov rax, r10;
@addforxloop:
add rax, 128;
jg @loopEnd;
// prefetch data...
// prefetch [r8 + rax];
// prefetchw [rcx + rax];
// move:
{$IFDEF AVXSUP}vmovdqa ymm0, [r8 + rax - 128]; {$ELSE}db $C4,$C1,$7D,$6F,$44,$00,$80;{$ENDIF}
{$IFDEF AVXSUP}vmovdqa [rcx + rax - 128], ymm0; {$ELSE}db $C5,$FD,$7F,$44,$01,$80;{$ENDIF}
{$IFDEF AVXSUP}vmovdqa ymm1, [r8 + rax - 96]; {$ELSE}db $C4,$C1,$7D,$6F,$4C,$00,$A0;{$ENDIF}
{$IFDEF AVXSUP}vmovdqa [rcx + rax - 96], ymm1; {$ELSE}db $C5,$FD,$7F,$4C,$01,$A0;{$ENDIF}
{$IFDEF AVXSUP}vmovdqa ymm2, [r8 + rax - 64]; {$ELSE}db $C4,$C1,$7D,$6F,$54,$00,$C0;{$ENDIF}
{$IFDEF AVXSUP}vmovdqa [rcx + rax - 64], ymm2; {$ELSE}db $C5,$FD,$7F,$54,$01,$C0;{$ENDIF}
{$IFDEF AVXSUP}vmovdqa ymm3, [r8 + rax - 32]; {$ELSE}db $C4,$C1,$7D,$6F,$5C,$00,$E0;{$ENDIF}
{$IFDEF AVXSUP}vmovdqa [rcx + rax - 32], ymm3; {$ELSE}db $C5,$FD,$7F,$5C,$01,$E0;{$ENDIF}
jmp @addforxloop
@loopEnd:
sub rax, 128;
jz @nextLine;
@addforxloop2:
add rax, 16;
jg @loopEnd2;
{$IFDEF AVXSUP}vmovdqa xmm0, [r8 + rax - 16]; {$ELSE}db $C4,$C1,$79,$6F,$44,$00,$F0;{$ENDIF}
{$IFDEF AVXSUP}vmovdqa [rcx + rax - 16], xmm0; {$ELSE}db $C5,$F9,$7F,$44,$01,$F0;{$ENDIF}
jmp @addforxloop2;
@loopEnd2:
sub rax, 16;
jz @nextLine;
// last element
{$IFDEF AVXSUP}vmovsd xmm0, [r8 + rax]; {$ELSE}db $C4,$C1,$7B,$10,$04,$00;{$ENDIF}
{$IFDEF AVXSUP}vmovsd [rcx + rax], xmm0; {$ELSE}db $C5,$FB,$11,$04,$01;{$ENDIF}
@nextLine:
// next line:
add r8, r9;
add rcx, rdx;
// loop y end
dec r11;
jnz @@addforyloop;
{$IFDEF AVXSUP}vzeroupper; {$ELSE}db $C5,$F8,$77;{$ENDIF}
end;
procedure AVXMatrixCopyUnAligned(Dest : PDouble; const destLineWidth : NativeInt; src : PDouble; const srcLineWidth : NativeInt; {$ifdef UNIX}unixWidth{$ELSE}width{$endif}, {$ifdef UNIX}unixHeight{$ELSE}height{$endif} : NativeInt); {$IFDEF FPC}assembler;{$ENDIF}
{$ifdef UNIX}
var width : NativeInt;
height : NativeInt;
{$ENDIF}
asm
{$IFDEF UNIX}
// Linux uses a diffrent ABI -> copy over the registers so they meet with winABI
// (note that the 5th and 6th parameter are are on the stack)
// The parameters are passed in the following order:
// RDI, RSI, RDX, RCX -> mov to RCX, RDX, R8, R9
mov width, r8;
mov height, r9;
mov r8, rdx;
mov r9, rcx;
mov rcx, rdi;
mov rdx, rsi;
{$ENDIF}
// note: RCX = dest, RDX = destLineWidth, R8 = src, R9 = srcLineWidth
//iters := -width*sizeof(double);
mov r10, width;
imul r10, -8;
// helper registers for the mt1, mt2 and dest pointers
sub r8, r10;
sub rcx, r10;
// for y := 0 to height - 1:
mov r11, Height;
@@addforyloop:
// for x := 0 to w - 1;
// prepare for reverse loop
mov rax, r10;
@addforxloop:
add rax, 128;
jg @loopEnd;
// prefetch data...
// prefetch [r8 + rax];
// prefetchw [rcx + rax];
// move:
{$IFDEF AVXSUP}vmovdqu ymm0, [r8 + rax - 128]; {$ELSE}db $C4,$C1,$7E,$6F,$44,$00,$80;{$ENDIF}
{$IFDEF AVXSUP}vmovdqu [rcx + rax - 128], ymm0; {$ELSE}db $C5,$FE,$7F,$44,$01,$80;{$ENDIF}
{$IFDEF AVXSUP}vmovdqu ymm1, [r8 + rax - 96]; {$ELSE}db $C4,$C1,$7E,$6F,$4C,$00,$A0;{$ENDIF}
{$IFDEF AVXSUP}vmovdqu [rcx + rax - 96], ymm1; {$ELSE}db $C5,$FE,$7F,$4C,$01,$A0;{$ENDIF}
{$IFDEF AVXSUP}vmovdqu ymm2, [r8 + rax - 64]; {$ELSE}db $C4,$C1,$7E,$6F,$54,$00,$C0;{$ENDIF}
{$IFDEF AVXSUP}vmovdqu [rcx + rax - 64], ymm2; {$ELSE}db $C5,$FE,$7F,$54,$01,$C0;{$ENDIF}
{$IFDEF AVXSUP}vmovdqu ymm3, [r8 + rax - 32]; {$ELSE}db $C4,$C1,$7E,$6F,$5C,$00,$E0;{$ENDIF}
{$IFDEF AVXSUP}vmovdqu [rcx + rax - 32], ymm3; {$ELSE}db $C5,$FE,$7F,$5C,$01,$E0;{$ENDIF}
jmp @addforxloop
@loopEnd:
sub rax, 128;
jz @nextLine;
@addforxloop2:
add rax, 16;
jg @loopEnd2;
{$IFDEF AVXSUP}vmovdqu xmm0, [r8 + rax - 16]; {$ELSE}db $C4,$C1,$7A,$6F,$44,$00,$F0;{$ENDIF}
{$IFDEF AVXSUP}vmovdqu [rcx + rax - 16], xmm0; {$ELSE}db $C5,$FA,$7F,$44,$01,$F0;{$ENDIF}
jmp @addforxloop2;
@loopEnd2:
sub rax, 16;
jz @nextLine;
// last element
{$IFDEF AVXSUP}vmovsd xmm0, [r8 + rax]; {$ELSE}db $C4,$C1,$7B,$10,$04,$00;{$ENDIF}
{$IFDEF AVXSUP}vmovsd [rcx + rax], xmm0; {$ELSE}db $C5,$FB,$11,$04,$01;{$ENDIF}
@nextLine:
// next line:
add r8, r9;
add rcx, rdx;
// loop y end
dec r11;
jnz @@addforyloop;
{$IFDEF AVXSUP}vzeroupper; {$ELSE}db $C5,$F8,$77;{$ENDIF}
end;
{$ENDIF}
end.