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
| __asm {
section .text local
global _FullComplexAuto
align 16
_FullComplexAuto type func
; r0: in -- r6
; r1: sc -- r5
; (sp-28): idx -- r3
[ push r6
push r7
]
[ push d6
push d7
]
[bmclr #$4,sr.l ; saturation off
tfra r0,r6 ; r6 -> in
]
[ move.l (sp-28),r3 ; r3 -> idx
move.l #4,n0
]
adda #4,r6,r7
[ dosetup1 inner1
doen1 #160 ; AUTO_LEN/2
]
loopstart1
inner1
[ clr d12
clr d13
clr d14
clr d15
tfra r1,r5 ; r5 -> sc
move.2f (r6)+n0,d4:d5 ; load in_r(1),in_i(1)
]
[ move.2f (r5)+,d0:d1 ; load sc_r(1),sc_ri(1)
move.2f (r7)+n0,d6:d7 ; load in_r(4),in_i(4)
] ; software pipelining
[ dosetup2 inner2
doen2 #10
]
loopstart2
inner2
[ clr d8
clr d9
clr d10
clr d11
]
[ dosetup3 inner3
doen3 #128
]
loopstart3
inner3
[ mac d4,d0,d8 ; A_r(1) += in_r(1) * sc_r(1)
mac d5,d0,d9 ; A_i(1) += in_i(1) * sc_r(1)
mac d6,d0,d10 ; A_r(4) += in_r(4) * sc_r(1)
mac d7,d0,d11 ; A_i(4) += in_i(4) * sc_r(1)
move.2f (r6)+n0,d2:d3 ; load in_r(4),in_i(4)
]
[ mac d5,d1,d8 ; A_r(1) += in_i(1) * sc_i(1)
mac -d4,d1,d9 ; A_i(1) += -in_r(1) * sc_i(1)
mac d7,d1,d10 ; A_r(4) += in_i(4) * sc_i(1)
mac -d6,d1,d11 ; A_i(4) += -in_r(4) * sc_i(1)
move.2f (r5)+,d0:d1 ; load sc_r(2),sc_ri(2)
move.2f (r7)+n0,d6:d7 ; load in_r(8),in_i(8)
]
[ mac d2,d0,d8 ; A_r(1) += in_r(4) * sc_r(2)
mac d3,d0,d9 ; A_i(1) += in_i(4) * sc_r(2)
mac d6,d0,d10 ; A_r(4) += in_r(8) * sc_r(2)
mac d7,d0,d11 ; A_i(4) += in_i(8) * sc_r(2)
move.2f (r6)+n0,d4:d5 ; load in_r(8),in_i(8)
]
[ mac d3,d1,d8 ; A_r(1) += in_i(4) * sc_i(2)
mac -d2,d1,d9 ; A_i(1) += -in_r(4) * sc_i(2)
mac d7,d1,d10 ; A_r(4) += in_i(8) * sc_i(2)
mac -d6,d1,d11 ; A_i(4) += -in_r(8) * sc_i(2)
move.2f (r5)+,d0:d1 ; load sc_r(3),sc_ri(3)
move.2f (r7)+n0,d6:d7 ; load in_r(12),in_i(12)
]
loopend3
[ asrr #14,d8 ; scaling
asrr #14,d9
asrr #14,d10
asrr #14,d11
]
[ add d8,d12,d12 ; A_r(1)
add d9,d13,d13 ; A_i(1)
add d10,d14,d14 ; A_r(4)
add d11,d15,d15 ; A_i(4)
]
loopend2
[ abs d12
abs d13
abs d14
abs d15
adda #8,r0,r0
]
[ add d12,d13,d12 ; |A_r(1)| + |A_i(1)|
add d14,d15,d13 ; |A_r(4)| + |A_i(4)|tfra r0,r6 ; r6 -> in tfra r1,r5 ; r5 -> sc][move.2l d12:d13,(r3)+adda #4,r6,r7
]
loopend1 bmset #$4,sr.l ; saturation mode on
nop
nop
[ pop d6
pop d7
]
[ pop r6
pop r7
]
rts
endsec
}; |
