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
|
/***************************************************************************
* __________ __ ___.
* Open \______ \ ____ ____ | | _\_ |__ _______ ___
* Source | _// _ \_/ ___\| |/ /| __ \ / _ \ \/ /
* Jukebox | | ( <_> ) \___| < | \_\ ( <_> > < <
* Firmware |____|_ /\____/ \___ >__|_ \|___ /\____/__/\_ \
* \/ \/ \/ \/ \/
* $Id$
*
* Copyright (C) 2005 by Thom Johansen
*
* All files in this archive are subject to the GNU General Public License.
* See the file COPYING in the source tree root for full license agreement.
*
* This software is distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY
* KIND, either express or implied.
*
****************************************************************************/
/* this will also be the home to III_imdct_l in the future */
.global III_imdct_s
III_imdct_s:
/* we need to save 9 registers and 36 samples of temp buffer */
lea.l (-45*4, %sp), %sp
movem.l %d2-%d7/%a2-%a4, (36*4, %sp)
move.l (45*4 + 4, %sp), %a2 /* a2 = X */
move.l %sp, %a3
/* IMDCT */
/* if additional precision is needed in this block, it is possible to
* get more low bits out of the accext01 register _before_ doing the
* movclrs.
*/
sub.l %a0, %a0 /* clear loop variable */
.imdctloop: /* outer loop label */
lea.l imdct_s, %a1 /* load pointer to imdct coefs in a1 */
movem.l (%a2), %d0-%d5 /* load some input data in d0-d5 */
lea.l (6*4, %a2), %a2
clr.l %d7 /* clear loop variable */
move.l (%a1)+, %a4 /* load imdct coef in a4 */
.macloop: /* inner loop label */
mac.l %d0, %a4, (%a1)+, %a4, %acc0 /* mac sequence */
mac.l %d1, %a4, (%a1)+, %a4, %acc0
mac.l %d2, %a4, (%a1)+, %a4, %acc0
mac.l %d3, %a4, (%a1)+, %a4, %acc0
mac.l %d4, %a4, (%a1)+, %a4, %acc0
mac.l %d5, %a4, (%a1)+, %a4, %acc0
movclr.l %acc0, %d6 /* get result, left shifted once */
asl.l #3, %d6 /* one shift free, shift three more */
move.l %d6, (%a3, %d7.l*4) /* yptr[i] = result */
neg.l %d6
neg.l %d7
move.l %d6, (5*4, %a3, %d7.l*4) /* yptr[5 - i] = -result */
mac.l %d0, %a4, (%a1)+, %a4, %acc0 /* mac sequence */
mac.l %d1, %a4, (%a1)+, %a4, %acc0
mac.l %d2, %a4, (%a1)+, %a4, %acc0
mac.l %d3, %a4, (%a1)+, %a4, %acc0
mac.l %d4, %a4, (%a1)+, %a4, %acc0
mac.l %d5, %a4, (%a1)+, %a4, %acc0
movclr.l %acc0, %d6 /* get result */
asl.l #3, %d6
move.l %d6, (11*4, %a3, %d7.l*4) /* yptr[11 - i] = result */
neg.l %d7
move.l %d6, (6*4, %a3, %d7.l*4) /* yptr[i + 6] = result */
addq.l #1, %d7 /* increment inner loop variable */
moveq.l #3, %d6
cmp.l %d6, %d7 /* we do three inner loop iterations */
jne .macloop
lea.l (12*4, %a3), %a3 /* add pointer increment */
addq.l #1, %a0 /* increment outer loop variable */
moveq.l #3, %d0
cmp.l %d0, %a0 /* we do three outer loop iterations */
jne .imdctloop
/* windowing, overlapping and concatenation */
move.l (45*4 + 8, %sp), %a2 /* a2 = z */
move.l %sp, %a3 /* a3 = tmp buffer ptr */
lea.l window_s, %a4 /* a4 = window coef pointer */
moveq.l #6, %d7 /* six iterations */
.overlaploop:
clr.l (%a2) /* z[i + 0] = 0 */
move.l (%a4), %d0
move.l (%a3), %d2
mac.l %d0, %d2, (6*4, %a4), %d1, %acc0
move.l (6*4, %a3), %d2
movclr.l %acc0, %d6
asl.l #3, %d6
move.l %d6, (6*4, %a2) /* z[i + 6] = result */
mac.l %d1, %d2, (12*4, %a3), %d2, %acc0
mac.l %d0, %d2, (18*4, %a3), %d2, %acc0
movclr.l %acc0, %d6
asl.l #3, %d6
move.l %d6, (12*4, %a2) /* z[i + 12] = result */
mac.l %d1, %d2, (24*4, %a3), %d2, %acc0
mac.l %d0, %d2, (30*4, %a3), %d2, %acc0
movclr.l %acc0, %d6
asl.l #3, %d6
move.l %d6, (18*4, %a2) /* z[i + 18] = result */
mac.l %d1, %d2, %acc0
movclr.l %acc0, %d6
asl.l #3, %d6
move.l %d6, (24*4, %a2) /* z[i + 24] = result */
clr.l (30*4, %a2) /* z[i + 30] = 0 */
addq.l #4, %a2 /* increment all pointers */
addq.l #4, %a3
addq.l #4, %a4
subq.l #1, %d7 /* decrement loop counter */
jne .overlaploop
/* fall through to exit if we're done */
/* clean up */
movem.l (36*4, %sp), %d2-%d7/%a2-%a4
lea.l (45*4, %sp), %sp
rts
|
