-
Notifications
You must be signed in to change notification settings - Fork 10
/
jidctint.c
5240 lines (4334 loc) · 180 KB
/
jidctint.c
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
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
/*
* jidctint.c
*
* Copyright (C) 1991-1998, Thomas G. Lane.
* Modification developed 2002-2018 by Guido Vollbeding.
* This file is part of the Independent JPEG Group's software.
* For conditions of distribution and use, see the accompanying README file.
*
* This file contains a slow-but-accurate integer implementation of the
* inverse DCT (Discrete Cosine Transform). In the IJG code, this routine
* must also perform dequantization of the input coefficients.
*
* A 2-D IDCT can be done by 1-D IDCT on each column followed by 1-D IDCT
* on each row (or vice versa, but it's more convenient to emit a row at
* a time). Direct algorithms are also available, but they are much more
* complex and seem not to be any faster when reduced to code.
*
* This implementation is based on an algorithm described in
* C. Loeffler, A. Ligtenberg and G. Moschytz, "Practical Fast 1-D DCT
* Algorithms with 11 Multiplications", Proc. Int'l. Conf. on Acoustics,
* Speech, and Signal Processing 1989 (ICASSP '89), pp. 988-991.
* The primary algorithm described there uses 11 multiplies and 29 adds.
* We use their alternate method with 12 multiplies and 32 adds.
* The advantage of this method is that no data path contains more than one
* multiplication; this allows a very simple and accurate implementation in
* scaled fixed-point arithmetic, with a minimal number of shifts.
*
* We also provide IDCT routines with various output sample block sizes for
* direct resolution reduction or enlargement and for direct resolving the
* common 2x1 and 1x2 subsampling cases without additional resampling: NxN
* (N=1...16), 2NxN, and Nx2N (N=1...8) pixels for one 8x8 input DCT block.
*
* For N<8 we simply take the corresponding low-frequency coefficients of
* the 8x8 input DCT block and apply an NxN point IDCT on the sub-block
* to yield the downscaled outputs.
* This can be seen as direct low-pass downsampling from the DCT domain
* point of view rather than the usual spatial domain point of view,
* yielding significant computational savings and results at least
* as good as common bilinear (averaging) spatial downsampling.
*
* For N>8 we apply a partial NxN IDCT on the 8 input coefficients as
* lower frequencies and higher frequencies assumed to be zero.
* It turns out that the computational effort is similar to the 8x8 IDCT
* regarding the output size.
* Furthermore, the scaling and descaling is the same for all IDCT sizes.
*
* CAUTION: We rely on the FIX() macro except for the N=1,2,4,8 cases
* since there would be too many additional constants to pre-calculate.
*/
#define JPEG_INTERNALS
#include "jinclude.h"
#include "jpeglib.h"
#include "jdct.h" /* Private declarations for DCT subsystem */
#ifdef DCT_ISLOW_SUPPORTED
/*
* This module is specialized to the case DCTSIZE = 8.
*/
#if DCTSIZE != 8
Sorry, this code only copes with 8x8 DCT blocks. /* deliberate syntax err */
#endif
/*
* The poop on this scaling stuff is as follows:
*
* Each 1-D IDCT step produces outputs which are a factor of sqrt(N)
* larger than the true IDCT outputs. The final outputs are therefore
* a factor of N larger than desired; since N=8 this can be cured by
* a simple right shift at the end of the algorithm. The advantage of
* this arrangement is that we save two multiplications per 1-D IDCT,
* because the y0 and y4 inputs need not be divided by sqrt(N).
*
* We have to do addition and subtraction of the integer inputs, which
* is no problem, and multiplication by fractional constants, which is
* a problem to do in integer arithmetic. We multiply all the constants
* by CONST_SCALE and convert them to integer constants (thus retaining
* CONST_BITS bits of precision in the constants). After doing a
* multiplication we have to divide the product by CONST_SCALE, with proper
* rounding, to produce the correct output. This division can be done
* cheaply as a right shift of CONST_BITS bits. We postpone shifting
* as long as possible so that partial sums can be added together with
* full fractional precision.
*
* The outputs of the first pass are scaled up by PASS1_BITS bits so that
* they are represented to better-than-integral precision. These outputs
* require BITS_IN_JSAMPLE + PASS1_BITS + 3 bits; this fits in a 16-bit word
* with the recommended scaling. (To scale up 12-bit sample data further, an
* intermediate INT32 array would be needed.)
*
* To avoid overflow of the 32-bit intermediate results in pass 2, we must
* have BITS_IN_JSAMPLE + CONST_BITS + PASS1_BITS <= 26. Error analysis
* shows that the values given below are the most effective.
*/
#if BITS_IN_JSAMPLE == 8
#define CONST_BITS 13
#define PASS1_BITS 2
#else
#define CONST_BITS 13
#define PASS1_BITS 1 /* lose a little precision to avoid overflow */
#endif
/* Some C compilers fail to reduce "FIX(constant)" at compile time, thus
* causing a lot of useless floating-point operations at run time.
* To get around this we use the following pre-calculated constants.
* If you change CONST_BITS you may want to add appropriate values.
* (With a reasonable C compiler, you can just rely on the FIX() macro...)
*/
#if CONST_BITS == 13
#define FIX_0_298631336 ((INT32) 2446) /* FIX(0.298631336) */
#define FIX_0_390180644 ((INT32) 3196) /* FIX(0.390180644) */
#define FIX_0_541196100 ((INT32) 4433) /* FIX(0.541196100) */
#define FIX_0_765366865 ((INT32) 6270) /* FIX(0.765366865) */
#define FIX_0_899976223 ((INT32) 7373) /* FIX(0.899976223) */
#define FIX_1_175875602 ((INT32) 9633) /* FIX(1.175875602) */
#define FIX_1_501321110 ((INT32) 12299) /* FIX(1.501321110) */
#define FIX_1_847759065 ((INT32) 15137) /* FIX(1.847759065) */
#define FIX_1_961570560 ((INT32) 16069) /* FIX(1.961570560) */
#define FIX_2_053119869 ((INT32) 16819) /* FIX(2.053119869) */
#define FIX_2_562915447 ((INT32) 20995) /* FIX(2.562915447) */
#define FIX_3_072711026 ((INT32) 25172) /* FIX(3.072711026) */
#else
#define FIX_0_298631336 FIX(0.298631336)
#define FIX_0_390180644 FIX(0.390180644)
#define FIX_0_541196100 FIX(0.541196100)
#define FIX_0_765366865 FIX(0.765366865)
#define FIX_0_899976223 FIX(0.899976223)
#define FIX_1_175875602 FIX(1.175875602)
#define FIX_1_501321110 FIX(1.501321110)
#define FIX_1_847759065 FIX(1.847759065)
#define FIX_1_961570560 FIX(1.961570560)
#define FIX_2_053119869 FIX(2.053119869)
#define FIX_2_562915447 FIX(2.562915447)
#define FIX_3_072711026 FIX(3.072711026)
#endif
/* Multiply an INT32 variable by an INT32 constant to yield an INT32 result.
* For 8-bit samples with the recommended scaling, all the variable
* and constant values involved are no more than 16 bits wide, so a
* 16x16->32 bit multiply can be used instead of a full 32x32 multiply.
* For 12-bit samples, a full 32-bit multiplication will be needed.
*/
#if BITS_IN_JSAMPLE == 8
#define MULTIPLY(var,const) MULTIPLY16C16(var,const)
#else
#define MULTIPLY(var,const) ((var) * (const))
#endif
/* Dequantize a coefficient by multiplying it by the multiplier-table
* entry; produce an int result. In this module, both inputs and result
* are 16 bits or less, so either int or short multiply will work.
*/
#define DEQUANTIZE(coef,quantval) (((ISLOW_MULT_TYPE) (coef)) * (quantval))
/*
* Perform dequantization and inverse DCT on one block of coefficients.
*
* Optimized algorithm with 12 multiplications in the 1-D kernel.
* cK represents sqrt(2) * cos(K*pi/16).
*/
GLOBAL(void)
jpeg_idct_islow (j_decompress_ptr cinfo, jpeg_component_info * compptr,
JCOEFPTR coef_block,
JSAMPARRAY output_buf, JDIMENSION output_col)
{
INT32 tmp0, tmp1, tmp2, tmp3;
INT32 tmp10, tmp11, tmp12, tmp13;
INT32 z1, z2, z3;
JCOEFPTR inptr;
ISLOW_MULT_TYPE * quantptr;
int * wsptr;
JSAMPROW outptr;
JSAMPLE *range_limit = IDCT_range_limit(cinfo);
int ctr;
int workspace[DCTSIZE2]; /* buffers data between passes */
SHIFT_TEMPS
/* Pass 1: process columns from input, store into work array.
* Note results are scaled up by sqrt(8) compared to a true IDCT;
* furthermore, we scale the results by 2**PASS1_BITS.
*/
inptr = coef_block;
quantptr = (ISLOW_MULT_TYPE *) compptr->dct_table;
wsptr = workspace;
for (ctr = DCTSIZE; ctr > 0; ctr--) {
/* Due to quantization, we will usually find that many of the input
* coefficients are zero, especially the AC terms. We can exploit this
* by short-circuiting the IDCT calculation for any column in which all
* the AC terms are zero. In that case each output is equal to the
* DC coefficient (with scale factor as needed).
* With typical images and quantization tables, half or more of the
* column DCT calculations can be simplified this way.
*/
if (inptr[DCTSIZE*1] == 0 && inptr[DCTSIZE*2] == 0 &&
inptr[DCTSIZE*3] == 0 && inptr[DCTSIZE*4] == 0 &&
inptr[DCTSIZE*5] == 0 && inptr[DCTSIZE*6] == 0 &&
inptr[DCTSIZE*7] == 0) {
/* AC terms all zero */
int dcval = DEQUANTIZE(inptr[DCTSIZE*0], quantptr[DCTSIZE*0]) << PASS1_BITS;
wsptr[DCTSIZE*0] = dcval;
wsptr[DCTSIZE*1] = dcval;
wsptr[DCTSIZE*2] = dcval;
wsptr[DCTSIZE*3] = dcval;
wsptr[DCTSIZE*4] = dcval;
wsptr[DCTSIZE*5] = dcval;
wsptr[DCTSIZE*6] = dcval;
wsptr[DCTSIZE*7] = dcval;
inptr++; /* advance pointers to next column */
quantptr++;
wsptr++;
continue;
}
/* Even part: reverse the even part of the forward DCT.
* The rotator is c(-6).
*/
z2 = DEQUANTIZE(inptr[DCTSIZE*0], quantptr[DCTSIZE*0]);
z3 = DEQUANTIZE(inptr[DCTSIZE*4], quantptr[DCTSIZE*4]);
z2 <<= CONST_BITS;
z3 <<= CONST_BITS;
/* Add fudge factor here for final descale. */
z2 += ONE << (CONST_BITS-PASS1_BITS-1);
tmp0 = z2 + z3;
tmp1 = z2 - z3;
z2 = DEQUANTIZE(inptr[DCTSIZE*2], quantptr[DCTSIZE*2]);
z3 = DEQUANTIZE(inptr[DCTSIZE*6], quantptr[DCTSIZE*6]);
z1 = MULTIPLY(z2 + z3, FIX_0_541196100); /* c6 */
tmp2 = z1 + MULTIPLY(z2, FIX_0_765366865); /* c2-c6 */
tmp3 = z1 - MULTIPLY(z3, FIX_1_847759065); /* c2+c6 */
tmp10 = tmp0 + tmp2;
tmp13 = tmp0 - tmp2;
tmp11 = tmp1 + tmp3;
tmp12 = tmp1 - tmp3;
/* Odd part per figure 8; the matrix is unitary and hence its
* transpose is its inverse. i0..i3 are y7,y5,y3,y1 respectively.
*/
tmp0 = DEQUANTIZE(inptr[DCTSIZE*7], quantptr[DCTSIZE*7]);
tmp1 = DEQUANTIZE(inptr[DCTSIZE*5], quantptr[DCTSIZE*5]);
tmp2 = DEQUANTIZE(inptr[DCTSIZE*3], quantptr[DCTSIZE*3]);
tmp3 = DEQUANTIZE(inptr[DCTSIZE*1], quantptr[DCTSIZE*1]);
z2 = tmp0 + tmp2;
z3 = tmp1 + tmp3;
z1 = MULTIPLY(z2 + z3, FIX_1_175875602); /* c3 */
z2 = MULTIPLY(z2, - FIX_1_961570560); /* -c3-c5 */
z3 = MULTIPLY(z3, - FIX_0_390180644); /* -c3+c5 */
z2 += z1;
z3 += z1;
z1 = MULTIPLY(tmp0 + tmp3, - FIX_0_899976223); /* -c3+c7 */
tmp0 = MULTIPLY(tmp0, FIX_0_298631336); /* -c1+c3+c5-c7 */
tmp3 = MULTIPLY(tmp3, FIX_1_501321110); /* c1+c3-c5-c7 */
tmp0 += z1 + z2;
tmp3 += z1 + z3;
z1 = MULTIPLY(tmp1 + tmp2, - FIX_2_562915447); /* -c1-c3 */
tmp1 = MULTIPLY(tmp1, FIX_2_053119869); /* c1+c3-c5+c7 */
tmp2 = MULTIPLY(tmp2, FIX_3_072711026); /* c1+c3+c5-c7 */
tmp1 += z1 + z3;
tmp2 += z1 + z2;
/* Final output stage: inputs are tmp10..tmp13, tmp0..tmp3 */
wsptr[DCTSIZE*0] = (int) RIGHT_SHIFT(tmp10 + tmp3, CONST_BITS-PASS1_BITS);
wsptr[DCTSIZE*7] = (int) RIGHT_SHIFT(tmp10 - tmp3, CONST_BITS-PASS1_BITS);
wsptr[DCTSIZE*1] = (int) RIGHT_SHIFT(tmp11 + tmp2, CONST_BITS-PASS1_BITS);
wsptr[DCTSIZE*6] = (int) RIGHT_SHIFT(tmp11 - tmp2, CONST_BITS-PASS1_BITS);
wsptr[DCTSIZE*2] = (int) RIGHT_SHIFT(tmp12 + tmp1, CONST_BITS-PASS1_BITS);
wsptr[DCTSIZE*5] = (int) RIGHT_SHIFT(tmp12 - tmp1, CONST_BITS-PASS1_BITS);
wsptr[DCTSIZE*3] = (int) RIGHT_SHIFT(tmp13 + tmp0, CONST_BITS-PASS1_BITS);
wsptr[DCTSIZE*4] = (int) RIGHT_SHIFT(tmp13 - tmp0, CONST_BITS-PASS1_BITS);
inptr++; /* advance pointers to next column */
quantptr++;
wsptr++;
}
/* Pass 2: process rows from work array, store into output array.
* Note that we must descale the results by a factor of 8 == 2**3,
* and also undo the PASS1_BITS scaling.
*/
wsptr = workspace;
for (ctr = 0; ctr < DCTSIZE; ctr++) {
outptr = output_buf[ctr] + output_col;
/* Add range center and fudge factor for final descale and range-limit. */
z2 = (INT32) wsptr[0] +
((((INT32) RANGE_CENTER) << (PASS1_BITS+3)) +
(ONE << (PASS1_BITS+2)));
/* Rows of zeroes can be exploited in the same way as we did with columns.
* However, the column calculation has created many nonzero AC terms, so
* the simplification applies less often (typically 5% to 10% of the time).
* On machines with very fast multiplication, it's possible that the
* test takes more time than it's worth. In that case this section
* may be commented out.
*/
#ifndef NO_ZERO_ROW_TEST
if (wsptr[1] == 0 && wsptr[2] == 0 && wsptr[3] == 0 && wsptr[4] == 0 &&
wsptr[5] == 0 && wsptr[6] == 0 && wsptr[7] == 0) {
/* AC terms all zero */
JSAMPLE dcval = range_limit[(int) RIGHT_SHIFT(z2, PASS1_BITS+3)
& RANGE_MASK];
outptr[0] = dcval;
outptr[1] = dcval;
outptr[2] = dcval;
outptr[3] = dcval;
outptr[4] = dcval;
outptr[5] = dcval;
outptr[6] = dcval;
outptr[7] = dcval;
wsptr += DCTSIZE; /* advance pointer to next row */
continue;
}
#endif
/* Even part: reverse the even part of the forward DCT.
* The rotator is c(-6).
*/
z3 = (INT32) wsptr[4];
tmp0 = (z2 + z3) << CONST_BITS;
tmp1 = (z2 - z3) << CONST_BITS;
z2 = (INT32) wsptr[2];
z3 = (INT32) wsptr[6];
z1 = MULTIPLY(z2 + z3, FIX_0_541196100); /* c6 */
tmp2 = z1 + MULTIPLY(z2, FIX_0_765366865); /* c2-c6 */
tmp3 = z1 - MULTIPLY(z3, FIX_1_847759065); /* c2+c6 */
tmp10 = tmp0 + tmp2;
tmp13 = tmp0 - tmp2;
tmp11 = tmp1 + tmp3;
tmp12 = tmp1 - tmp3;
/* Odd part per figure 8; the matrix is unitary and hence its
* transpose is its inverse. i0..i3 are y7,y5,y3,y1 respectively.
*/
tmp0 = (INT32) wsptr[7];
tmp1 = (INT32) wsptr[5];
tmp2 = (INT32) wsptr[3];
tmp3 = (INT32) wsptr[1];
z2 = tmp0 + tmp2;
z3 = tmp1 + tmp3;
z1 = MULTIPLY(z2 + z3, FIX_1_175875602); /* c3 */
z2 = MULTIPLY(z2, - FIX_1_961570560); /* -c3-c5 */
z3 = MULTIPLY(z3, - FIX_0_390180644); /* -c3+c5 */
z2 += z1;
z3 += z1;
z1 = MULTIPLY(tmp0 + tmp3, - FIX_0_899976223); /* -c3+c7 */
tmp0 = MULTIPLY(tmp0, FIX_0_298631336); /* -c1+c3+c5-c7 */
tmp3 = MULTIPLY(tmp3, FIX_1_501321110); /* c1+c3-c5-c7 */
tmp0 += z1 + z2;
tmp3 += z1 + z3;
z1 = MULTIPLY(tmp1 + tmp2, - FIX_2_562915447); /* -c1-c3 */
tmp1 = MULTIPLY(tmp1, FIX_2_053119869); /* c1+c3-c5+c7 */
tmp2 = MULTIPLY(tmp2, FIX_3_072711026); /* c1+c3+c5-c7 */
tmp1 += z1 + z3;
tmp2 += z1 + z2;
/* Final output stage: inputs are tmp10..tmp13, tmp0..tmp3 */
outptr[0] = range_limit[(int) RIGHT_SHIFT(tmp10 + tmp3,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[7] = range_limit[(int) RIGHT_SHIFT(tmp10 - tmp3,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[1] = range_limit[(int) RIGHT_SHIFT(tmp11 + tmp2,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[6] = range_limit[(int) RIGHT_SHIFT(tmp11 - tmp2,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[2] = range_limit[(int) RIGHT_SHIFT(tmp12 + tmp1,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[5] = range_limit[(int) RIGHT_SHIFT(tmp12 - tmp1,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[3] = range_limit[(int) RIGHT_SHIFT(tmp13 + tmp0,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[4] = range_limit[(int) RIGHT_SHIFT(tmp13 - tmp0,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
wsptr += DCTSIZE; /* advance pointer to next row */
}
}
#ifdef IDCT_SCALING_SUPPORTED
/*
* Perform dequantization and inverse DCT on one block of coefficients,
* producing a reduced-size 7x7 output block.
*
* Optimized algorithm with 12 multiplications in the 1-D kernel.
* cK represents sqrt(2) * cos(K*pi/14).
*/
GLOBAL(void)
jpeg_idct_7x7 (j_decompress_ptr cinfo, jpeg_component_info * compptr,
JCOEFPTR coef_block,
JSAMPARRAY output_buf, JDIMENSION output_col)
{
INT32 tmp0, tmp1, tmp2, tmp10, tmp11, tmp12, tmp13;
INT32 z1, z2, z3;
JCOEFPTR inptr;
ISLOW_MULT_TYPE * quantptr;
int * wsptr;
JSAMPROW outptr;
JSAMPLE *range_limit = IDCT_range_limit(cinfo);
int ctr;
int workspace[7*7]; /* buffers data between passes */
SHIFT_TEMPS
/* Pass 1: process columns from input, store into work array. */
inptr = coef_block;
quantptr = (ISLOW_MULT_TYPE *) compptr->dct_table;
wsptr = workspace;
for (ctr = 0; ctr < 7; ctr++, inptr++, quantptr++, wsptr++) {
/* Even part */
tmp13 = DEQUANTIZE(inptr[DCTSIZE*0], quantptr[DCTSIZE*0]);
tmp13 <<= CONST_BITS;
/* Add fudge factor here for final descale. */
tmp13 += ONE << (CONST_BITS-PASS1_BITS-1);
z1 = DEQUANTIZE(inptr[DCTSIZE*2], quantptr[DCTSIZE*2]);
z2 = DEQUANTIZE(inptr[DCTSIZE*4], quantptr[DCTSIZE*4]);
z3 = DEQUANTIZE(inptr[DCTSIZE*6], quantptr[DCTSIZE*6]);
tmp10 = MULTIPLY(z2 - z3, FIX(0.881747734)); /* c4 */
tmp12 = MULTIPLY(z1 - z2, FIX(0.314692123)); /* c6 */
tmp11 = tmp10 + tmp12 + tmp13 - MULTIPLY(z2, FIX(1.841218003)); /* c2+c4-c6 */
tmp0 = z1 + z3;
z2 -= tmp0;
tmp0 = MULTIPLY(tmp0, FIX(1.274162392)) + tmp13; /* c2 */
tmp10 += tmp0 - MULTIPLY(z3, FIX(0.077722536)); /* c2-c4-c6 */
tmp12 += tmp0 - MULTIPLY(z1, FIX(2.470602249)); /* c2+c4+c6 */
tmp13 += MULTIPLY(z2, FIX(1.414213562)); /* c0 */
/* Odd part */
z1 = DEQUANTIZE(inptr[DCTSIZE*1], quantptr[DCTSIZE*1]);
z2 = DEQUANTIZE(inptr[DCTSIZE*3], quantptr[DCTSIZE*3]);
z3 = DEQUANTIZE(inptr[DCTSIZE*5], quantptr[DCTSIZE*5]);
tmp1 = MULTIPLY(z1 + z2, FIX(0.935414347)); /* (c3+c1-c5)/2 */
tmp2 = MULTIPLY(z1 - z2, FIX(0.170262339)); /* (c3+c5-c1)/2 */
tmp0 = tmp1 - tmp2;
tmp1 += tmp2;
tmp2 = MULTIPLY(z2 + z3, - FIX(1.378756276)); /* -c1 */
tmp1 += tmp2;
z2 = MULTIPLY(z1 + z3, FIX(0.613604268)); /* c5 */
tmp0 += z2;
tmp2 += z2 + MULTIPLY(z3, FIX(1.870828693)); /* c3+c1-c5 */
/* Final output stage */
wsptr[7*0] = (int) RIGHT_SHIFT(tmp10 + tmp0, CONST_BITS-PASS1_BITS);
wsptr[7*6] = (int) RIGHT_SHIFT(tmp10 - tmp0, CONST_BITS-PASS1_BITS);
wsptr[7*1] = (int) RIGHT_SHIFT(tmp11 + tmp1, CONST_BITS-PASS1_BITS);
wsptr[7*5] = (int) RIGHT_SHIFT(tmp11 - tmp1, CONST_BITS-PASS1_BITS);
wsptr[7*2] = (int) RIGHT_SHIFT(tmp12 + tmp2, CONST_BITS-PASS1_BITS);
wsptr[7*4] = (int) RIGHT_SHIFT(tmp12 - tmp2, CONST_BITS-PASS1_BITS);
wsptr[7*3] = (int) RIGHT_SHIFT(tmp13, CONST_BITS-PASS1_BITS);
}
/* Pass 2: process 7 rows from work array, store into output array. */
wsptr = workspace;
for (ctr = 0; ctr < 7; ctr++) {
outptr = output_buf[ctr] + output_col;
/* Even part */
/* Add range center and fudge factor for final descale and range-limit. */
tmp13 = (INT32) wsptr[0] +
((((INT32) RANGE_CENTER) << (PASS1_BITS+3)) +
(ONE << (PASS1_BITS+2)));
tmp13 <<= CONST_BITS;
z1 = (INT32) wsptr[2];
z2 = (INT32) wsptr[4];
z3 = (INT32) wsptr[6];
tmp10 = MULTIPLY(z2 - z3, FIX(0.881747734)); /* c4 */
tmp12 = MULTIPLY(z1 - z2, FIX(0.314692123)); /* c6 */
tmp11 = tmp10 + tmp12 + tmp13 - MULTIPLY(z2, FIX(1.841218003)); /* c2+c4-c6 */
tmp0 = z1 + z3;
z2 -= tmp0;
tmp0 = MULTIPLY(tmp0, FIX(1.274162392)) + tmp13; /* c2 */
tmp10 += tmp0 - MULTIPLY(z3, FIX(0.077722536)); /* c2-c4-c6 */
tmp12 += tmp0 - MULTIPLY(z1, FIX(2.470602249)); /* c2+c4+c6 */
tmp13 += MULTIPLY(z2, FIX(1.414213562)); /* c0 */
/* Odd part */
z1 = (INT32) wsptr[1];
z2 = (INT32) wsptr[3];
z3 = (INT32) wsptr[5];
tmp1 = MULTIPLY(z1 + z2, FIX(0.935414347)); /* (c3+c1-c5)/2 */
tmp2 = MULTIPLY(z1 - z2, FIX(0.170262339)); /* (c3+c5-c1)/2 */
tmp0 = tmp1 - tmp2;
tmp1 += tmp2;
tmp2 = MULTIPLY(z2 + z3, - FIX(1.378756276)); /* -c1 */
tmp1 += tmp2;
z2 = MULTIPLY(z1 + z3, FIX(0.613604268)); /* c5 */
tmp0 += z2;
tmp2 += z2 + MULTIPLY(z3, FIX(1.870828693)); /* c3+c1-c5 */
/* Final output stage */
outptr[0] = range_limit[(int) RIGHT_SHIFT(tmp10 + tmp0,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[6] = range_limit[(int) RIGHT_SHIFT(tmp10 - tmp0,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[1] = range_limit[(int) RIGHT_SHIFT(tmp11 + tmp1,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[5] = range_limit[(int) RIGHT_SHIFT(tmp11 - tmp1,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[2] = range_limit[(int) RIGHT_SHIFT(tmp12 + tmp2,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[4] = range_limit[(int) RIGHT_SHIFT(tmp12 - tmp2,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[3] = range_limit[(int) RIGHT_SHIFT(tmp13,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
wsptr += 7; /* advance pointer to next row */
}
}
/*
* Perform dequantization and inverse DCT on one block of coefficients,
* producing a reduced-size 6x6 output block.
*
* Optimized algorithm with 3 multiplications in the 1-D kernel.
* cK represents sqrt(2) * cos(K*pi/12).
*/
GLOBAL(void)
jpeg_idct_6x6 (j_decompress_ptr cinfo, jpeg_component_info * compptr,
JCOEFPTR coef_block,
JSAMPARRAY output_buf, JDIMENSION output_col)
{
INT32 tmp0, tmp1, tmp2, tmp10, tmp11, tmp12;
INT32 z1, z2, z3;
JCOEFPTR inptr;
ISLOW_MULT_TYPE * quantptr;
int * wsptr;
JSAMPROW outptr;
JSAMPLE *range_limit = IDCT_range_limit(cinfo);
int ctr;
int workspace[6*6]; /* buffers data between passes */
SHIFT_TEMPS
/* Pass 1: process columns from input, store into work array. */
inptr = coef_block;
quantptr = (ISLOW_MULT_TYPE *) compptr->dct_table;
wsptr = workspace;
for (ctr = 0; ctr < 6; ctr++, inptr++, quantptr++, wsptr++) {
/* Even part */
tmp0 = DEQUANTIZE(inptr[DCTSIZE*0], quantptr[DCTSIZE*0]);
tmp0 <<= CONST_BITS;
/* Add fudge factor here for final descale. */
tmp0 += ONE << (CONST_BITS-PASS1_BITS-1);
tmp2 = DEQUANTIZE(inptr[DCTSIZE*4], quantptr[DCTSIZE*4]);
tmp10 = MULTIPLY(tmp2, FIX(0.707106781)); /* c4 */
tmp1 = tmp0 + tmp10;
tmp11 = RIGHT_SHIFT(tmp0 - tmp10 - tmp10, CONST_BITS-PASS1_BITS);
tmp10 = DEQUANTIZE(inptr[DCTSIZE*2], quantptr[DCTSIZE*2]);
tmp0 = MULTIPLY(tmp10, FIX(1.224744871)); /* c2 */
tmp10 = tmp1 + tmp0;
tmp12 = tmp1 - tmp0;
/* Odd part */
z1 = DEQUANTIZE(inptr[DCTSIZE*1], quantptr[DCTSIZE*1]);
z2 = DEQUANTIZE(inptr[DCTSIZE*3], quantptr[DCTSIZE*3]);
z3 = DEQUANTIZE(inptr[DCTSIZE*5], quantptr[DCTSIZE*5]);
tmp1 = MULTIPLY(z1 + z3, FIX(0.366025404)); /* c5 */
tmp0 = tmp1 + ((z1 + z2) << CONST_BITS);
tmp2 = tmp1 + ((z3 - z2) << CONST_BITS);
tmp1 = (z1 - z2 - z3) << PASS1_BITS;
/* Final output stage */
wsptr[6*0] = (int) RIGHT_SHIFT(tmp10 + tmp0, CONST_BITS-PASS1_BITS);
wsptr[6*5] = (int) RIGHT_SHIFT(tmp10 - tmp0, CONST_BITS-PASS1_BITS);
wsptr[6*1] = (int) (tmp11 + tmp1);
wsptr[6*4] = (int) (tmp11 - tmp1);
wsptr[6*2] = (int) RIGHT_SHIFT(tmp12 + tmp2, CONST_BITS-PASS1_BITS);
wsptr[6*3] = (int) RIGHT_SHIFT(tmp12 - tmp2, CONST_BITS-PASS1_BITS);
}
/* Pass 2: process 6 rows from work array, store into output array. */
wsptr = workspace;
for (ctr = 0; ctr < 6; ctr++) {
outptr = output_buf[ctr] + output_col;
/* Even part */
/* Add range center and fudge factor for final descale and range-limit. */
tmp0 = (INT32) wsptr[0] +
((((INT32) RANGE_CENTER) << (PASS1_BITS+3)) +
(ONE << (PASS1_BITS+2)));
tmp0 <<= CONST_BITS;
tmp2 = (INT32) wsptr[4];
tmp10 = MULTIPLY(tmp2, FIX(0.707106781)); /* c4 */
tmp1 = tmp0 + tmp10;
tmp11 = tmp0 - tmp10 - tmp10;
tmp10 = (INT32) wsptr[2];
tmp0 = MULTIPLY(tmp10, FIX(1.224744871)); /* c2 */
tmp10 = tmp1 + tmp0;
tmp12 = tmp1 - tmp0;
/* Odd part */
z1 = (INT32) wsptr[1];
z2 = (INT32) wsptr[3];
z3 = (INT32) wsptr[5];
tmp1 = MULTIPLY(z1 + z3, FIX(0.366025404)); /* c5 */
tmp0 = tmp1 + ((z1 + z2) << CONST_BITS);
tmp2 = tmp1 + ((z3 - z2) << CONST_BITS);
tmp1 = (z1 - z2 - z3) << CONST_BITS;
/* Final output stage */
outptr[0] = range_limit[(int) RIGHT_SHIFT(tmp10 + tmp0,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[5] = range_limit[(int) RIGHT_SHIFT(tmp10 - tmp0,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[1] = range_limit[(int) RIGHT_SHIFT(tmp11 + tmp1,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[4] = range_limit[(int) RIGHT_SHIFT(tmp11 - tmp1,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[2] = range_limit[(int) RIGHT_SHIFT(tmp12 + tmp2,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[3] = range_limit[(int) RIGHT_SHIFT(tmp12 - tmp2,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
wsptr += 6; /* advance pointer to next row */
}
}
/*
* Perform dequantization and inverse DCT on one block of coefficients,
* producing a reduced-size 5x5 output block.
*
* Optimized algorithm with 5 multiplications in the 1-D kernel.
* cK represents sqrt(2) * cos(K*pi/10).
*/
GLOBAL(void)
jpeg_idct_5x5 (j_decompress_ptr cinfo, jpeg_component_info * compptr,
JCOEFPTR coef_block,
JSAMPARRAY output_buf, JDIMENSION output_col)
{
INT32 tmp0, tmp1, tmp10, tmp11, tmp12;
INT32 z1, z2, z3;
JCOEFPTR inptr;
ISLOW_MULT_TYPE * quantptr;
int * wsptr;
JSAMPROW outptr;
JSAMPLE *range_limit = IDCT_range_limit(cinfo);
int ctr;
int workspace[5*5]; /* buffers data between passes */
SHIFT_TEMPS
/* Pass 1: process columns from input, store into work array. */
inptr = coef_block;
quantptr = (ISLOW_MULT_TYPE *) compptr->dct_table;
wsptr = workspace;
for (ctr = 0; ctr < 5; ctr++, inptr++, quantptr++, wsptr++) {
/* Even part */
tmp12 = DEQUANTIZE(inptr[DCTSIZE*0], quantptr[DCTSIZE*0]);
tmp12 <<= CONST_BITS;
/* Add fudge factor here for final descale. */
tmp12 += ONE << (CONST_BITS-PASS1_BITS-1);
tmp0 = DEQUANTIZE(inptr[DCTSIZE*2], quantptr[DCTSIZE*2]);
tmp1 = DEQUANTIZE(inptr[DCTSIZE*4], quantptr[DCTSIZE*4]);
z1 = MULTIPLY(tmp0 + tmp1, FIX(0.790569415)); /* (c2+c4)/2 */
z2 = MULTIPLY(tmp0 - tmp1, FIX(0.353553391)); /* (c2-c4)/2 */
z3 = tmp12 + z2;
tmp10 = z3 + z1;
tmp11 = z3 - z1;
tmp12 -= z2 << 2;
/* Odd part */
z2 = DEQUANTIZE(inptr[DCTSIZE*1], quantptr[DCTSIZE*1]);
z3 = DEQUANTIZE(inptr[DCTSIZE*3], quantptr[DCTSIZE*3]);
z1 = MULTIPLY(z2 + z3, FIX(0.831253876)); /* c3 */
tmp0 = z1 + MULTIPLY(z2, FIX(0.513743148)); /* c1-c3 */
tmp1 = z1 - MULTIPLY(z3, FIX(2.176250899)); /* c1+c3 */
/* Final output stage */
wsptr[5*0] = (int) RIGHT_SHIFT(tmp10 + tmp0, CONST_BITS-PASS1_BITS);
wsptr[5*4] = (int) RIGHT_SHIFT(tmp10 - tmp0, CONST_BITS-PASS1_BITS);
wsptr[5*1] = (int) RIGHT_SHIFT(tmp11 + tmp1, CONST_BITS-PASS1_BITS);
wsptr[5*3] = (int) RIGHT_SHIFT(tmp11 - tmp1, CONST_BITS-PASS1_BITS);
wsptr[5*2] = (int) RIGHT_SHIFT(tmp12, CONST_BITS-PASS1_BITS);
}
/* Pass 2: process 5 rows from work array, store into output array. */
wsptr = workspace;
for (ctr = 0; ctr < 5; ctr++) {
outptr = output_buf[ctr] + output_col;
/* Even part */
/* Add range center and fudge factor for final descale and range-limit. */
tmp12 = (INT32) wsptr[0] +
((((INT32) RANGE_CENTER) << (PASS1_BITS+3)) +
(ONE << (PASS1_BITS+2)));
tmp12 <<= CONST_BITS;
tmp0 = (INT32) wsptr[2];
tmp1 = (INT32) wsptr[4];
z1 = MULTIPLY(tmp0 + tmp1, FIX(0.790569415)); /* (c2+c4)/2 */
z2 = MULTIPLY(tmp0 - tmp1, FIX(0.353553391)); /* (c2-c4)/2 */
z3 = tmp12 + z2;
tmp10 = z3 + z1;
tmp11 = z3 - z1;
tmp12 -= z2 << 2;
/* Odd part */
z2 = (INT32) wsptr[1];
z3 = (INT32) wsptr[3];
z1 = MULTIPLY(z2 + z3, FIX(0.831253876)); /* c3 */
tmp0 = z1 + MULTIPLY(z2, FIX(0.513743148)); /* c1-c3 */
tmp1 = z1 - MULTIPLY(z3, FIX(2.176250899)); /* c1+c3 */
/* Final output stage */
outptr[0] = range_limit[(int) RIGHT_SHIFT(tmp10 + tmp0,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[4] = range_limit[(int) RIGHT_SHIFT(tmp10 - tmp0,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[1] = range_limit[(int) RIGHT_SHIFT(tmp11 + tmp1,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[3] = range_limit[(int) RIGHT_SHIFT(tmp11 - tmp1,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[2] = range_limit[(int) RIGHT_SHIFT(tmp12,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
wsptr += 5; /* advance pointer to next row */
}
}
/*
* Perform dequantization and inverse DCT on one block of coefficients,
* producing a reduced-size 4x4 output block.
*
* Optimized algorithm with 3 multiplications in the 1-D kernel.
* cK represents sqrt(2) * cos(K*pi/16) [refers to 8-point IDCT].
*/
GLOBAL(void)
jpeg_idct_4x4 (j_decompress_ptr cinfo, jpeg_component_info * compptr,
JCOEFPTR coef_block,
JSAMPARRAY output_buf, JDIMENSION output_col)
{
INT32 tmp0, tmp2, tmp10, tmp12;
INT32 z1, z2, z3;
JCOEFPTR inptr;
ISLOW_MULT_TYPE * quantptr;
int * wsptr;
JSAMPROW outptr;
JSAMPLE *range_limit = IDCT_range_limit(cinfo);
int ctr;
int workspace[4*4]; /* buffers data between passes */
SHIFT_TEMPS
/* Pass 1: process columns from input, store into work array. */
inptr = coef_block;
quantptr = (ISLOW_MULT_TYPE *) compptr->dct_table;
wsptr = workspace;
for (ctr = 0; ctr < 4; ctr++, inptr++, quantptr++, wsptr++) {
/* Even part */
tmp0 = DEQUANTIZE(inptr[DCTSIZE*0], quantptr[DCTSIZE*0]);
tmp2 = DEQUANTIZE(inptr[DCTSIZE*2], quantptr[DCTSIZE*2]);
tmp10 = (tmp0 + tmp2) << PASS1_BITS;
tmp12 = (tmp0 - tmp2) << PASS1_BITS;
/* Odd part */
/* Same rotation as in the even part of the 8x8 LL&M IDCT */
z2 = DEQUANTIZE(inptr[DCTSIZE*1], quantptr[DCTSIZE*1]);
z3 = DEQUANTIZE(inptr[DCTSIZE*3], quantptr[DCTSIZE*3]);
z1 = MULTIPLY(z2 + z3, FIX_0_541196100); /* c6 */
/* Add fudge factor here for final descale. */
z1 += ONE << (CONST_BITS-PASS1_BITS-1);
tmp0 = RIGHT_SHIFT(z1 + MULTIPLY(z2, FIX_0_765366865), /* c2-c6 */
CONST_BITS-PASS1_BITS);
tmp2 = RIGHT_SHIFT(z1 - MULTIPLY(z3, FIX_1_847759065), /* c2+c6 */
CONST_BITS-PASS1_BITS);
/* Final output stage */
wsptr[4*0] = (int) (tmp10 + tmp0);
wsptr[4*3] = (int) (tmp10 - tmp0);
wsptr[4*1] = (int) (tmp12 + tmp2);
wsptr[4*2] = (int) (tmp12 - tmp2);
}
/* Pass 2: process 4 rows from work array, store into output array. */
wsptr = workspace;
for (ctr = 0; ctr < 4; ctr++) {
outptr = output_buf[ctr] + output_col;
/* Even part */
/* Add range center and fudge factor for final descale and range-limit. */
tmp0 = (INT32) wsptr[0] +
((((INT32) RANGE_CENTER) << (PASS1_BITS+3)) +
(ONE << (PASS1_BITS+2)));
tmp2 = (INT32) wsptr[2];
tmp10 = (tmp0 + tmp2) << CONST_BITS;
tmp12 = (tmp0 - tmp2) << CONST_BITS;
/* Odd part */
/* Same rotation as in the even part of the 8x8 LL&M IDCT */
z2 = (INT32) wsptr[1];
z3 = (INT32) wsptr[3];
z1 = MULTIPLY(z2 + z3, FIX_0_541196100); /* c6 */
tmp0 = z1 + MULTIPLY(z2, FIX_0_765366865); /* c2-c6 */
tmp2 = z1 - MULTIPLY(z3, FIX_1_847759065); /* c2+c6 */
/* Final output stage */
outptr[0] = range_limit[(int) RIGHT_SHIFT(tmp10 + tmp0,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[3] = range_limit[(int) RIGHT_SHIFT(tmp10 - tmp0,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[1] = range_limit[(int) RIGHT_SHIFT(tmp12 + tmp2,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
outptr[2] = range_limit[(int) RIGHT_SHIFT(tmp12 - tmp2,
CONST_BITS+PASS1_BITS+3)
& RANGE_MASK];
wsptr += 4; /* advance pointer to next row */
}
}
/*
* Perform dequantization and inverse DCT on one block of coefficients,
* producing a reduced-size 3x3 output block.
*
* Optimized algorithm with 2 multiplications in the 1-D kernel.
* cK represents sqrt(2) * cos(K*pi/6).
*/
GLOBAL(void)
jpeg_idct_3x3 (j_decompress_ptr cinfo, jpeg_component_info * compptr,
JCOEFPTR coef_block,
JSAMPARRAY output_buf, JDIMENSION output_col)
{
INT32 tmp0, tmp2, tmp10, tmp12;
JCOEFPTR inptr;
ISLOW_MULT_TYPE * quantptr;
int * wsptr;
JSAMPROW outptr;
JSAMPLE *range_limit = IDCT_range_limit(cinfo);
int ctr;
int workspace[3*3]; /* buffers data between passes */
SHIFT_TEMPS
/* Pass 1: process columns from input, store into work array. */
inptr = coef_block;
quantptr = (ISLOW_MULT_TYPE *) compptr->dct_table;
wsptr = workspace;
for (ctr = 0; ctr < 3; ctr++, inptr++, quantptr++, wsptr++) {
/* Even part */
tmp0 = DEQUANTIZE(inptr[DCTSIZE*0], quantptr[DCTSIZE*0]);
tmp0 <<= CONST_BITS;
/* Add fudge factor here for final descale. */
tmp0 += ONE << (CONST_BITS-PASS1_BITS-1);
tmp2 = DEQUANTIZE(inptr[DCTSIZE*2], quantptr[DCTSIZE*2]);
tmp12 = MULTIPLY(tmp2, FIX(0.707106781)); /* c2 */
tmp10 = tmp0 + tmp12;
tmp2 = tmp0 - tmp12 - tmp12;
/* Odd part */
tmp12 = DEQUANTIZE(inptr[DCTSIZE*1], quantptr[DCTSIZE*1]);
tmp0 = MULTIPLY(tmp12, FIX(1.224744871)); /* c1 */
/* Final output stage */
wsptr[3*0] = (int) RIGHT_SHIFT(tmp10 + tmp0, CONST_BITS-PASS1_BITS);
wsptr[3*2] = (int) RIGHT_SHIFT(tmp10 - tmp0, CONST_BITS-PASS1_BITS);
wsptr[3*1] = (int) RIGHT_SHIFT(tmp2, CONST_BITS-PASS1_BITS);
}
/* Pass 2: process 3 rows from work array, store into output array. */
wsptr = workspace;
for (ctr = 0; ctr < 3; ctr++) {
outptr = output_buf[ctr] + output_col;
/* Even part */
/* Add range center and fudge factor for final descale and range-limit. */
tmp0 = (INT32) wsptr[0] +
((((INT32) RANGE_CENTER) << (PASS1_BITS+3)) +
(ONE << (PASS1_BITS+2)));
tmp0 <<= CONST_BITS;
tmp2 = (INT32) wsptr[2];
tmp12 = MULTIPLY(tmp2, FIX(0.707106781)); /* c2 */
tmp10 = tmp0 + tmp12;
tmp2 = tmp0 - tmp12 - tmp12;
/* Odd part */
tmp12 = (INT32) wsptr[1];