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
|
/*!
***************************************************************************
* \file transform.c
*
* \brief
* Transform functions
*
* \author
* Main contributors (see contributors.h for copyright, address and affiliation details)
* - Alexis Michael Tourapis
* \date
* 01. July 2007
**************************************************************************
*/
#include "global.h"
#include "transform.h"
#include <emmintrin.h>
void forward4x4(int **block, int **tblock, int pos_y, int pos_x)
{
int i, ii;
int tmp[16];
int *pTmp = tmp, *pblock;
int p0,p1,p2,p3;
int t0,t1,t2,t3;
// Horizontal
for (i=pos_y; i < pos_y + BLOCK_SIZE; i++)
{
pblock = &block[i][pos_x];
p0 = *(pblock++);
p1 = *(pblock++);
p2 = *(pblock++);
p3 = *(pblock );
t0 = p0 + p3;
t1 = p1 + p2;
t2 = p1 - p2;
t3 = p0 - p3;
*(pTmp++) = t0 + t1;
*(pTmp++) = (t3 << 1) + t2;
*(pTmp++) = t0 - t1;
*(pTmp++) = t3 - (t2 << 1);
}
// Vertical
for (i=0; i < BLOCK_SIZE; i++)
{
pTmp = tmp + i;
p0 = *pTmp;
p1 = *(pTmp += BLOCK_SIZE);
p2 = *(pTmp += BLOCK_SIZE);
p3 = *(pTmp += BLOCK_SIZE);
t0 = p0 + p3;
t1 = p1 + p2;
t2 = p1 - p2;
t3 = p0 - p3;
ii = pos_x + i;
tblock[pos_y ][ii] = t0 + t1;
tblock[pos_y + 1][ii] = t2 + (t3 << 1);
tblock[pos_y + 2][ii] = t0 - t1;
tblock[pos_y + 3][ii] = t3 - (t2 << 1);
}
}
static void inverse4x4(const h264_short_block_t tblock, h264_short_block_t block, int pos_y, int pos_x)
{
int i;
short tmp[16];
short *pTmp = tmp;
int p0,p1,p2,p3;
int t0,t1,t2,t3;
// Horizontal
for (i = 0; i < BLOCK_SIZE; i++)
{
t0 = tblock[i][0];
t1 = tblock[i][1];
t2 = tblock[i][2];
t3 = tblock[i][3];
p0 = t0 + t2;
p1 = t0 - t2;
p2 = (t1 >> 1) - t3;
p3 = t1 + (t3 >> 1);
*(pTmp++) = p0 + p3;
*(pTmp++) = p1 + p2;
*(pTmp++) = p1 - p2;
*(pTmp++) = p0 - p3;
}
// Vertical
for (i = 0; i < BLOCK_SIZE; i++)
{
pTmp = tmp + i;
t0 = *pTmp;
t1 = *(pTmp += BLOCK_SIZE);
t2 = *(pTmp += BLOCK_SIZE);
t3 = *(pTmp += BLOCK_SIZE);
p0 = t0 + t2;
p1 = t0 - t2;
p2 =(t1 >> 1) - t3;
p3 = t1 + (t3 >> 1);
block[0][i] = p0 + p3;
block[1][i] = p1 + p2;
block[2][i] = p1 - p2;
block[3][i] = p0 - p3;
}
}
#ifdef _M_IX86
// benski> this exists just for conformance testing. not used in production code
static void inverse4x4_sse2_x86(const h264_short_macroblock_t tblock, h264_short_macroblock_t block, int pos_y, int pos_x)
{
__asm
{
mov edx, pos_y
shl edx, 4 // 16 step stride
add edx, pos_x
shl edx, 1 // * sizeof(short)
// eax: pointer to the start of tblock (offset by passed pos_y, pos_x)
mov eax, edx
add eax, tblock
// esi: results
mov esi, edx
add esi, block
// load 4x4 matrix
movq mm0, MMWORD PTR 0[eax]
movq mm1, MMWORD PTR 32[eax]
movq mm2, MMWORD PTR 64[eax]
movq mm3, MMWORD PTR 96[eax]
// rotate 4x4 matrix
movq mm4, mm0 // p0 = mm4 (copy)
punpcklwd mm0, mm2 // r0 = mm0
punpckhwd mm4, mm2 // r2 = mm4
movq mm5, mm1 // p1 = mm5 (copy)
punpcklwd mm1, mm3 // r1 = mm1
punpckhwd mm5, mm3 // r3 = mm5
movq mm6, mm0 // r0 = mm6 (copy)
punpcklwd mm0, mm1 // t0 = mm0
punpckhwd mm6, mm1 // t1 = mm6
movq mm1, mm4 // r2 = mm1 (copy)
punpcklwd mm1, mm5 // t2 = mm1
punpckhwd mm4, mm5 // t3 = mm4
/* register state:
mm0: t0
mm1: t2
mm2:
mm3:
mm4: t3
mm5:
mm6: t1
mm7:
*/
/*
p0 = t0 + t2;
p1 = t0 - t2;
p2 = (t1 >> 1) - t3;
p3 = t1 + (t3 >> 1);
*/
movq mm2, mm0 // mm2 = t0 (copy)
paddw mm0, mm1 // mm0 = p0
psubw mm2, mm1 // mm2 = p1, mm1 available
movq mm5, mm6 // mm5 = t1 (copy)
psraw mm5, 1 // mm5 = (t1 >> 1)
psubw mm5, mm4 // mm5 = p2
psraw mm4, 1 // mm4 = (t3 >> 1)
paddw mm6, mm4 // mm6 = p3
/* register state:
mm0: p0
mm1:
mm2: p1
mm3:
mm4:
mm5: p2
mm6: p3
mm7:
*/
/*
*(pTmp++) = p0 + p3;
*(pTmp++) = p1 + p2;
*(pTmp++) = p1 - p2;
*(pTmp++) = p0 - p3;
*/
movq mm3, mm0 // mm3 = p0 (copy)
paddw mm0, mm6 // mm0 = r0
movq mm1, mm2 // mm1 = p1 (copy)
paddw mm1, mm5 // mm1 = r1
psubw mm2, mm5 // mm2 = r2, mm5 available
psubw mm3, mm6 // mm3 = r3
/* register state:
mm0: r0
mm1: r1
mm2: r2
mm3: r3
mm4:
mm5:
mm6:
mm7:
*/
// rotate 4x4 matrix to set up for vertical
movq mm4, mm0 // r0 = mm4 (copy)
punpcklwd mm0, mm2 // p0 = mm0
punpckhwd mm4, mm2 // p2 = mm4
movq mm5, mm1 // r1 = mm5 (copy)
punpcklwd mm1, mm3 // p1 = mm1
punpckhwd mm5, mm3 // p3 = mm5
movq mm6, mm0 // p0 = mm6 (copy)
punpcklwd mm0, mm1 // t0 = mm0
punpckhwd mm6, mm1 // t1 = mm6
movq mm1, mm4 // p2 = mm1 (copy)
punpcklwd mm1, mm5 // t2 = mm1
punpckhwd mm4, mm5 // t3 = mm4
/* register state:
mm0: t0
mm1: t2
mm2:
mm3:
mm4: t3
mm5:
mm6: t1
mm7:
*/
/*
p0 = t0 + t2;
p1 = t0 - t2;
p2 = (t1 >> 1) - t3;
p3 = t1 + (t3 >> 1);
*/
movq mm2, mm0 // mm2 = t0 (copy)
paddw mm0, mm1 // mm0 = p0
psubw mm2, mm1 // mm2 = p1, mm1 available
movq mm5, mm6 // mm5 = t1 (copy)
psraw mm5, 1 // mm5 = (t1 >> 1)
psubw mm5, mm4 // mm5 = p2
psraw mm4, 1 // mm4 = (t3 >> 1)
paddw mm6, mm4 // mm6 = p3
/* register state:
mm0: p0
mm1:
mm2: p1
mm3:
mm4:
mm5: p2
mm6: p3
mm7:
*/
/*
*(pTmp++) = p0 + p3;
*(pTmp++) = p1 + p2;
*(pTmp++) = p1 - p2;
*(pTmp++) = p0 - p3;
*/
movq mm3, mm0 // mm3 = p0 (copy)
paddw mm0, mm6 // mm0 = r0
movq mm1, mm2 // mm1 = p1 (copy)
paddw mm1, mm5 // mm1 = r1
psubw mm2, mm5 // mm2 = r2, mm5 available
psubw mm3, mm6 // mm3 = r3
/* register state:
mm0: r0
mm1: r1
mm2: r2
mm3: r3
mm4:
mm5:
mm6:
mm7:
*/
movq XMMWORD PTR 0[esi], mm0
movq XMMWORD PTR 32[esi], mm1
movq XMMWORD PTR 64[esi], mm2
movq XMMWORD PTR 96[esi], mm3
}
}
#endif
static void sample_reconstruct(h264_imgpel_macroblock_t curImg, const h264_imgpel_macroblock_t mpr, const h264_short_block_t tblock, int joff, int mb_x, int max_imgpel_value)
{
#ifdef _M_IX86
__asm
{
// mm0 : constant value 32
mov edx, 0x00200020
movd mm0, edx
punpckldq mm0, mm0
// ecx: y offset
mov ecx, joff
shl ecx, 4 // imgpel stuff is going to be 16 byte stride
add ecx, mb_x
// eax: curImg
mov eax, curImg
add eax, ecx
// edx: mpr
mov edx, mpr
add edx, ecx
// ecx: tblock (which is short, not byte)
mov ecx, tblock
// mm7: zero
pxor mm7, mm7
// load coefficients
movq mm1, MMWORD PTR 0[ecx]
movq mm2, MMWORD PTR 8[ecx]
movq mm3, MMWORD PTR 16[ecx]
movq mm4, MMWORD PTR 24[ecx]
paddw mm1, mm0 // rres + 32
paddw mm2, mm0 // rres + 32
paddw mm3, mm0 // rres + 32
paddw mm0, mm4 // rres + 32
psraw mm1, 6 // (rres + 32) >> 6
psraw mm2, 6 // (rres + 32) >> 6
psraw mm3, 6 // (rres + 32) >> 6
psraw mm0, 6 // (rres + 32) >> 6
// mm1-mm3: tblock[0] - tblock[2], mm0: tblock[3]
// convert mpr from unsigned char to short
movd mm4, DWORD PTR 0[edx]
movd mm5, DWORD PTR 16[edx]
movd mm6, DWORD PTR 32[edx]
punpcklbw mm4, mm7
punpcklbw mm5, mm7
punpcklbw mm6, mm7
paddsw mm4, mm1 // pred_row + rres_row
movd mm1, DWORD PTR 48[edx] // reuse mm1 for mpr[3]
paddsw mm5, mm2 // pred_row + rres_row
punpcklbw mm1, mm7
paddsw mm6, mm3 // pred_row + rres_row
paddsw mm1, mm0 // pred_row + rres_row
// results in mm4, mm5, mm6, mm1
// move back to 8 bit
packuswb mm4, mm7
packuswb mm5, mm7
packuswb mm6, mm7
packuswb mm1, mm7
movd DWORD PTR 0[eax], mm4
movd DWORD PTR 16[eax], mm5
movd DWORD PTR 32[eax], mm6
movd DWORD PTR 48[eax], mm1
}
#else
int i, j;
for (j = 0; j < BLOCK_SIZE; j++)
{
for (i=0;i<BLOCK_SIZE;i++)
curImg[j+joff][mb_x+i] = (imgpel) iClip1( max_imgpel_value, rshift_rnd_sf(tblock[j][i], DQ_BITS) + mpr[j+joff][mb_x+i]);
}
#endif
}
#if defined(_M_IX86) && defined(_DEBUG)
void itrans4x4_sse2(const h264_short_macroblock_t tblock, const h264_imgpel_macroblock_t mb_pred, h264_imgpel_macroblock_t mb_rec, int pos_x, int pos_y)
{
__declspec(align(32)) static const short const32[4] = {32, 32, 32, 32};
__asm
{
mov edx, pos_y
shl edx, 4 // imgpel stuff is going to be 16 byte stride
add edx, pos_x
// eax: tblock
lea eax, [edx*2]
add eax, tblock
// ecx: mpr
mov ecx, mb_pred
add ecx, edx
// edx: results
add edx, mb_rec
// load 4x4 matrix
movq mm0, MMWORD PTR 0[eax]
movq mm1, MMWORD PTR 32[eax]
movq mm2, MMWORD PTR 64[eax]
movq mm3, MMWORD PTR 96[eax]
// rotate 4x4 matrix
movq mm4, mm0 // p0 = mm4 (copy)
punpcklwd mm0, mm2 // r0 = mm0
punpckhwd mm4, mm2 // r2 = mm4
movq mm5, mm1 // p1 = mm5 (copy)
punpcklwd mm1, mm3 // r1 = mm1
punpckhwd mm5, mm3 // r3 = mm5
movq mm6, mm0 // r0 = mm6 (copy)
punpcklwd mm0, mm1 // t0 = mm0
punpckhwd mm6, mm1 // t1 = mm6
movq mm1, mm4 // r2 = mm1 (copy)
punpcklwd mm1, mm5 // t2 = mm1
punpckhwd mm4, mm5 // t3 = mm4
/* register state:
mm0: t0
mm1: t2
mm2:
mm3:
mm4: t3
mm5:
mm6: t1
mm7:
*/
/*
p0 = t0 + t2;
p1 = t0 - t2;
p2 = (t1 >> 1) - t3;
p3 = t1 + (t3 >> 1);
*/
movq mm2, mm0 // mm2 = t0 (copy)
paddw mm0, mm1 // mm0 = p0
psubw mm2, mm1 // mm2 = p1, mm1 available
movq mm5, mm6 // mm5 = t1 (copy)
psraw mm5, 1 // mm5 = (t1 >> 1)
psubw mm5, mm4 // mm5 = p2
psraw mm4, 1 // mm4 = (t3 >> 1)
paddw mm6, mm4 // mm6 = p3
/* register state:
mm0: p0
mm1:
mm2: p1
mm3:
mm4:
mm5: p2
mm6: p3
mm7:
*/
/*
*(pTmp++) = p0 + p3;
*(pTmp++) = p1 + p2;
*(pTmp++) = p1 - p2;
*(pTmp++) = p0 - p3;
*/
movq mm3, mm0 // mm3 = p0 (copy)
paddw mm0, mm6 // mm0 = r0
movq mm1, mm2 // mm1 = p1 (copy)
paddw mm1, mm5 // mm1 = r1
psubw mm2, mm5 // mm2 = r2, mm5 available
psubw mm3, mm6 // mm3 = r3
/* register state:
mm0: r0
mm1: r1
mm2: r2
mm3: r3
mm4:
mm5:
mm6:
mm7:
*/
// rotate 4x4 matrix to set up for vertical
movq mm4, mm0 // r0 = mm4 (copy)
punpcklwd mm0, mm2 // p0 = mm0
punpckhwd mm4, mm2 // p2 = mm4
movq mm5, mm1 // r1 = mm5 (copy)
punpcklwd mm1, mm3 // p1 = mm1
punpckhwd mm5, mm3 // p3 = mm5
movq mm6, mm0 // p0 = mm6 (copy)
punpcklwd mm0, mm1 // t0 = mm0
punpckhwd mm6, mm1 // t1 = mm6
movq mm1, mm4 // p2 = mm1 (copy)
punpcklwd mm1, mm5 // t2 = mm1
punpckhwd mm4, mm5 // t3 = mm4
/* register state:
mm0: t0
mm1: t2
mm2:
mm3:
mm4: t3
mm5:
mm6: t1
mm7:
*/
/*
p0 = t0 + t2;
p1 = t0 - t2;
p2 = (t1 >> 1) - t3;
p3 = t1 + (t3 >> 1);
*/
movq mm2, mm0 // mm2 = t0 (copy)
paddw mm0, mm1 // mm0 = p0
psubw mm2, mm1 // mm2 = p1, mm1 available
movq mm5, mm6 // mm5 = t1 (copy)
psraw mm5, 1 // mm5 = (t1 >> 1)
psubw mm5, mm4 // mm5 = p2
psraw mm4, 1 // mm4 = (t3 >> 1)
paddw mm6, mm4 // mm6 = p3
/* register state:
mm0: p0
mm1:
mm2: p1
mm3:
mm4:
mm5: p2
mm6: p3
mm7:
*/
/*
*(pTmp++) = p0 + p3;
*(pTmp++) = p1 + p2;
*(pTmp++) = p1 - p2;
*(pTmp++) = p0 - p3;
*/
movq mm3, mm0 // mm3 = p0 (copy)
paddw mm0, mm6 // mm0 = r0
movq mm1, mm2 // mm1 = p1 (copy)
paddw mm1, mm5 // mm1 = r1
psubw mm2, mm5 // mm2 = r2, mm5 available
psubw mm3, mm6 // mm3 = r3
/* register state:
mm0: r0
mm1: r1
mm2: r2
mm3: r3
mm4:
mm5:
mm6:
mm7:
*/
/* --- 4x4 iDCT done, now time to combine with mpr --- */
// mm0 : constant value 32
movq mm7, const32
paddw mm0, mm7 // rres + 32
psraw mm0, 6 // (rres + 32) >> 6
paddw mm1, mm7 // rres + 32
psraw mm1, 6 // (rres + 32) >> 6
paddw mm2, mm7 // rres + 32
psraw mm2, 6 // (rres + 32) >> 6
paddw mm3, mm7 // rres + 32
psraw mm3, 6 // (rres + 32) >> 6
pxor mm7, mm7
// convert mpr from unsigned char to short
movd mm4, DWORD PTR 0[ecx]
movd mm5, DWORD PTR 16[ecx]
movd mm6, DWORD PTR 32[ecx]
punpcklbw mm4, mm7
punpcklbw mm5, mm7
punpcklbw mm6, mm7
paddsw mm4, mm0 // pred_row + rres_row
movd mm0, DWORD PTR 48[ecx] // reuse mm0 for mpr[3]
paddsw mm5, mm1 // pred_row + rres_row
punpcklbw mm0, mm7
paddsw mm6, mm2 // pred_row + rres_row
paddsw mm0, mm3 // pred_row + rres_row
// results in mm4, mm5, mm6, mm0
// move back to 8 bit
packuswb mm4, mm7
packuswb mm5, mm7
packuswb mm6, mm7
packuswb mm0, mm7
movd DWORD PTR 0[edx], mm4
movd DWORD PTR 16[edx], mm5
movd DWORD PTR 32[edx], mm6
movd DWORD PTR 48[edx], mm0
}
}
#elif defined(_M_X64)
static void itrans4x4_sse2(const h264_int_macroblock_t tblock, const h264_imgpel_macroblock_t mb_pred, h264_imgpel_macroblock_t mb_rec, int pos_x, int pos_y)
{
__declspec(align(32)) static const int const32[4] = {32, 32, 32, 32};
__m128i p0,p1,p2,p3;
__m128i t0,t1,t2,t3;
__m128i r0,r1,r2,r3;
__m128i c32, zero;
// horizontal
// load registers in vertical mode, we'll rotate them next
p0 = _mm_loadu_si128((__m128i *)&tblock[pos_y][pos_x]); // 00 01 02 03
p1 = _mm_loadu_si128((__m128i *)&tblock[pos_y+1][pos_x]); // 10 11 12 13
p2 = _mm_loadu_si128((__m128i *)&tblock[pos_y+2][pos_x]); // 20 21 22 23
p3 = _mm_loadu_si128((__m128i *)&tblock[pos_y+3][pos_x]); // 30 31 32 33
// rotate 4x4 matrix
r0 = _mm_unpacklo_epi32(p0, p2); // 00 20 01 21
r1 = _mm_unpacklo_epi32(p1, p3); // 10 30 11 31
r2 = _mm_unpackhi_epi32(p0, p2); // 02 22 03 23
r3 = _mm_unpackhi_epi32(p1, p3); // 12 32 13 33
t0 = _mm_unpacklo_epi32(r0, r1); // 00 10 20 30
t1 = _mm_unpackhi_epi32(r0, r1); // 01 11 21 31
t2 = _mm_unpacklo_epi32(r2, r3); // 02 12 22 32
t3 = _mm_unpackhi_epi32(r2, r3); // 03 13 23 33
p0 = _mm_add_epi32(t0, t2); //t0 + t2;
p1 = _mm_sub_epi32(t0, t2); // t0 - t2;
p2 = _mm_srai_epi32(t1, 1); // t1 >> 1
p2 = _mm_sub_epi32(p2, t3); // (t1 >> 1) - t3;
p3 = _mm_srai_epi32(t3, 1); // (t3 >> 1)
p3 = _mm_add_epi32(p3, t1); // t1 + (t3 >> 1);
t0 = _mm_add_epi32(p0, p3); //p0 + p3;
t1 = _mm_add_epi32(p1, p2);//p1 + p2;
t2 = _mm_sub_epi32(p1, p2); //p1 - p2;
t3 = _mm_sub_epi32(p0, p3); //p0 - p3;
// rotate 4x4 matrix to set up for vertical
r0 = _mm_unpacklo_epi32(t0, t2);
r1 = _mm_unpacklo_epi32(t1, t3);
r2 = _mm_unpackhi_epi32(t0, t2);
r3 = _mm_unpackhi_epi32(t1, t3);
t0 = _mm_unpacklo_epi32(r0, r1);
t1 = _mm_unpackhi_epi32(r0, r1);
t2 = _mm_unpacklo_epi32(r2, r3);
t3 = _mm_unpackhi_epi32(r2, r3);
// vertical
p0 = _mm_add_epi32(t0, t2); //t0 + t2;
p3 = _mm_srai_epi32(t3, 1); // (t3 >> 1)
p3 = _mm_add_epi32(p3, t1); // t1 + (t3 >> 1);
r0 = _mm_add_epi32(p0, p3); //p0 + p3;
r3 = _mm_sub_epi32(p0, p3); //p0 - p3;
p1 = _mm_sub_epi32(t0, t2); // t0 - t2;
p2 = _mm_srai_epi32(t1, 1); // t1 >> 1
p2 = _mm_sub_epi32(p2, t3); // (t1 >> 1) - t3;
r1 = _mm_add_epi32(p1, p2);//p1 + p2;
r2 = _mm_sub_epi32(p1, p2); //p1 - p2;
c32 = _mm_load_si128((const __m128i *)const32);
zero = _mm_setzero_si128();
// (x + 32) >> 6
r0 = _mm_add_epi32(r0, c32);
r0 = _mm_srai_epi32(r0, 6);
r1 = _mm_add_epi32(r1, c32);
r1 = _mm_srai_epi32(r1, 6);
r2 = _mm_add_epi32(r2, c32);
r2 = _mm_srai_epi32(r2, 6);
r3 = _mm_add_epi32(r3, c32);
r3 = _mm_srai_epi32(r3, 6);
// convert to 16bit values
r0 = _mm_packs_epi32(r0, r1);
r2 = _mm_packs_epi32(r2, r3);
// convert mpr from unsigned char to short
p0 = _mm_cvtsi32_si128(*(int32_t *)&mb_pred[pos_y][pos_x]);
p1 = _mm_cvtsi32_si128(*(int32_t *)&mb_pred[pos_y+1][pos_x]);
p0 = _mm_unpacklo_epi32(p0, p1);
p0 = _mm_unpacklo_epi8(p0, zero); // convert to short
r0 = _mm_add_epi16(r0, p0);
p0 = _mm_cvtsi32_si128(*(int32_t *)&mb_pred[pos_y+2][pos_x]);
p1 = _mm_cvtsi32_si128(*(int32_t *)&mb_pred[pos_y+3][pos_x]);
p0 = _mm_unpacklo_epi32(p0, p1);
p0 = _mm_unpacklo_epi8(p0, zero); // convert to short
r2 = _mm_add_epi16(r2, p0);
r0 = _mm_packus_epi16(r0, r2); // convert to unsigned char
*(int32_t *)&mb_rec[pos_y][pos_x] = _mm_cvtsi128_si32(r0);
r0 = _mm_srli_si128(r0, 4);
*(int32_t *)&mb_rec[pos_y+1][pos_x] = _mm_cvtsi128_si32(r0);
r0 = _mm_srli_si128(r0, 4);
*(int32_t *)&mb_rec[pos_y+2][pos_x] = _mm_cvtsi128_si32(r0);
r0 = _mm_srli_si128(r0, 4);
*(int32_t *)&mb_rec[pos_y+3][pos_x] = _mm_cvtsi128_si32(r0);
}
#endif
void itrans4x4_c(const h264_short_block_t tblock, const h264_imgpel_macroblock_t mb_pred, h264_imgpel_macroblock_t mb_rec, int pos_x, int pos_y)
{
inverse4x4(tblock, (h264_short_block_row_t *)tblock,pos_y,pos_x);
sample_reconstruct(mb_rec, mb_pred, tblock, pos_y, pos_x, 255);
}
void ihadamard4x4(int block[4][4])
{
int i;
int tmp[16];
int *pTmp = tmp;
int p0,p1,p2,p3;
int t0,t1,t2,t3;
// Horizontal
for (i = 0; i < BLOCK_SIZE; i++)
{
t0 = block[i][0];
t1 = block[i][1];
t2 = block[i][2];
t3 = block[i][3];
p0 = t0 + t2;
p1 = t0 - t2;
p2 = t1 - t3;
p3 = t1 + t3;
*(pTmp++) = p0 + p3;
*(pTmp++) = p1 + p2;
*(pTmp++) = p1 - p2;
*(pTmp++) = p0 - p3;
}
// Vertical
for (i = 0; i < BLOCK_SIZE; i++)
{
pTmp = tmp + i;
t0 = *pTmp;
t1 = *(pTmp += BLOCK_SIZE);
t2 = *(pTmp += BLOCK_SIZE);
t3 = *(pTmp += BLOCK_SIZE);
p0 = t0 + t2;
p1 = t0 - t2;
p2 = t1 - t3;
p3 = t1 + t3;
block[0][i] = p0 + p3;
block[1][i] = p1 + p2;
block[2][i] = p1 - p2;
block[3][i] = p0 - p3;
}
}
void ihadamard4x2(int **tblock, int **block)
{
int i;
int tmp[8];
int *pTmp = tmp;
int p0,p1,p2,p3;
int t0,t1,t2,t3;
// Horizontal
*(pTmp++) = tblock[0][0] + tblock[1][0];
*(pTmp++) = tblock[0][1] + tblock[1][1];
*(pTmp++) = tblock[0][2] + tblock[1][2];
*(pTmp++) = tblock[0][3] + tblock[1][3];
*(pTmp++) = tblock[0][0] - tblock[1][0];
*(pTmp++) = tblock[0][1] - tblock[1][1];
*(pTmp++) = tblock[0][2] - tblock[1][2];
*(pTmp ) = tblock[0][3] - tblock[1][3];
// Vertical
pTmp = tmp;
for (i = 0; i < 2; i++)
{
p0 = *(pTmp++);
p1 = *(pTmp++);
p2 = *(pTmp++);
p3 = *(pTmp++);
t0 = p0 + p2;
t1 = p0 - p2;
t2 = p1 - p3;
t3 = p1 + p3;
// coefficients (transposed)
block[0][i] = t0 + t3;
block[1][i] = t1 + t2;
block[2][i] = t1 - t2;
block[3][i] = t0 - t3;
}
}
//following functions perform 8 additions, 8 assignments. Should be a bit faster
void ihadamard2x2(int tblock[4], int block[4])
{
int t0,t1,t2,t3;
t0 = tblock[0] + tblock[1];
t1 = tblock[0] - tblock[1];
t2 = tblock[2] + tblock[3];
t3 = tblock[2] - tblock[3];
block[0] = (t0 + t2);
block[1] = (t1 + t3);
block[2] = (t0 - t2);
block[3] = (t1 - t3);
}
|
