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
|
/*
wrapper for dcts
Copyright (C) 2001 Martin Vogt
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU Library General Public License as published by
the Free Software Foundation.
For more information look at the file COPYRIGHT in this package
*/
#include "dct.h"
#include "common.h"
ATTR_ALIGN(64) static REAL hcos_64[16];
ATTR_ALIGN(64) static REAL hcos_32[8];
ATTR_ALIGN(64) static REAL hcos_16[4];
ATTR_ALIGN(64) static REAL hcos_8[2];
ATTR_ALIGN(64) static REAL hcos_4;
/**
This was some time ago a standalone dct class,
but to get more speed I made it an inline dct
int the filter classes
*/
static int dct64Init=false;
void initialize_dct64() {
if (dct64Init==true) {
return;
}
dct64Init=true;
int i;
for(i=0;i<16;i++) {
hcos_64[i]=1.0/(2.0*cos(MY_PI*double(i*2+1)/64.0));
}
for(i=0;i< 8;i++) {
hcos_32[i]=1.0/(2.0*cos(MY_PI*double(i*2+1)/32.0));
}
for(i=0;i< 4;i++) {
hcos_16[i]=1.0/(2.0*cos(MY_PI*double(i*2+1)/16.0));
}
for(i=0;i< 2;i++) {
hcos_8[i]=1.0/(2.0*cos(MY_PI*double(i*2+1)/ 8.0));
}
hcos_4=1.0/(2.0*cos(MY_PI*1.0/4.0));
}
//
// splay dct64 , faster than mpeg123 dct. (from decode.c)
//
inline void dct64(REAL* out1,REAL* out2,REAL *fraction) {
ATTR_ALIGN(64) REAL p0,p1,p2,p3,p4,p5,p6,p7,p8,p9,pa,pb,pc,pd,pe,pf;
ATTR_ALIGN(64) REAL q0,q1,q2,q3,q4,q5,q6,q7,q8,q9,qa,qb,qc,qd,qe,qf;
#define OUT1(v,t) out1[(32-(v))*16] =(-(out1[(v)*16]=t))
#define OUT2(v) out2[(96-(v)-32)*16]=out2[((v)-32)*16]
// compute new values via a fast cosine transform:
{
// put to buffer 0..15
register REAL* x=fraction;
p0=x[ 0]+x[31];p1=x[ 1]+x[30];p2=x[ 2]+x[29];p3=x[ 3]+x[28];
p4=x[ 4]+x[27];p5=x[ 5]+x[26];p6=x[ 6]+x[25];p7=x[ 7]+x[24];
p8=x[ 8]+x[23];p9=x[ 9]+x[22];pa=x[10]+x[21];pb=x[11]+x[20];
pc=x[12]+x[19];pd=x[13]+x[18];pe=x[14]+x[17];pf=x[15]+x[16];
}
// put to buffer 32..39
q0=p0+pf;q1=p1+pe;q2=p2+pd;q3=p3+pc;
q4=p4+pb;q5=p5+pa;q6=p6+p9;q7=p7+p8;
// put to buffer 40..47
q8=hcos_32[0]*(p0-pf);q9=hcos_32[1]*(p1-pe);
qa=hcos_32[2]*(p2-pd);qb=hcos_32[3]*(p3-pc);
qc=hcos_32[4]*(p4-pb);qd=hcos_32[5]*(p5-pa);
qe=hcos_32[6]*(p6-p9);qf=hcos_32[7]*(p7-p8);
p0=q0+q7;p1=q1+q6;p2=q2+q5;p3=q3+q4;
p4=hcos_16[0]*(q0-q7);p5=hcos_16[1]*(q1-q6);
p6=hcos_16[2]*(q2-q5);p7=hcos_16[3]*(q3-q4);
p8=q8+qf;p9=q9+qe;pa=qa+qd;pb=qb+qc;
pc=hcos_16[0]*(q8-qf);pd=hcos_16[1]*(q9-qe);
pe=hcos_16[2]*(qa-qd);pf=hcos_16[3]*(qb-qc);
q0=p0+p3;q1=p1+p2;q2=hcos_8[0]*(p0-p3);q3=hcos_8[1]*(p1-p2);
q4=p4+p7;q5=p5+p6;q6=hcos_8[0]*(p4-p7);q7=hcos_8[1]*(p5-p6);
q8=p8+pb;q9=p9+pa;qa=hcos_8[0]*(p8-pb);qb=hcos_8[1]*(p9-pa);
qc=pc+pf;qd=pd+pe;qe=hcos_8[0]*(pc-pf);qf=hcos_8[1]*(pd-pe);
p0=q0+q1;p1=hcos_4*(q0-q1);p2=q2+q3;p3=hcos_4*(q2-q3);
p4=q4+q5;p5=hcos_4*(q4-q5);p6=q6+q7;p7=hcos_4*(q6-q7);
p8=q8+q9;p9=hcos_4*(q8-q9);pa=qa+qb;pb=hcos_4*(qa-qb);
pc=qc+qd;pd=hcos_4*(qc-qd);pe=qe+qf;pf=hcos_4*(qe-qf);
{
register REAL tmp;
tmp=p6+p7;
OUT2(36)=-(p5+tmp);
OUT2(44)=-(p4+tmp);
tmp=pb+pf;
OUT1(10,tmp);
OUT1(6,pd+tmp);
tmp=pe+pf;
OUT2(46)=-(p8+pc+tmp);
OUT2(34)=-(p9+pd+tmp);
tmp+=pa+pb;
OUT2(38)=-(pd+tmp);
OUT2(42)=-(pc+tmp);
OUT1(2,p9+pd+pf);
OUT1(4,p5+p7);
OUT2(48)=-p0;
out2[0]=-(out1[0]=p1);
OUT1( 8,p3);
OUT1(12,p7);
OUT1(14,pf);
OUT2(40)=-(p2+p3);
}
{
// put to buffer 16..31
register REAL *x=fraction;
p0=hcos_64[ 0]*(x[ 0]-x[31]);p1=hcos_64[ 1]*(x[ 1]-x[30]);
p2=hcos_64[ 2]*(x[ 2]-x[29]);p3=hcos_64[ 3]*(x[ 3]-x[28]);
p4=hcos_64[ 4]*(x[ 4]-x[27]);p5=hcos_64[ 5]*(x[ 5]-x[26]);
p6=hcos_64[ 6]*(x[ 6]-x[25]);p7=hcos_64[ 7]*(x[ 7]-x[24]);
p8=hcos_64[ 8]*(x[ 8]-x[23]);p9=hcos_64[ 9]*(x[ 9]-x[22]);
pa=hcos_64[10]*(x[10]-x[21]);pb=hcos_64[11]*(x[11]-x[20]);
pc=hcos_64[12]*(x[12]-x[19]);pd=hcos_64[13]*(x[13]-x[18]);
pe=hcos_64[14]*(x[14]-x[17]);pf=hcos_64[15]*(x[15]-x[16]);
}
// put to 48..63
q0=p0+pf;q1=p1+pe;q2=p2+pd;q3=p3+pc;
q4=p4+pb;q5=p5+pa;q6=p6+p9;q7=p7+p8;
q8=hcos_32[0]*(p0-pf);q9=hcos_32[1]*(p1-pe);
qa=hcos_32[2]*(p2-pd);qb=hcos_32[3]*(p3-pc);
qc=hcos_32[4]*(p4-pb);qd=hcos_32[5]*(p5-pa);
qe=hcos_32[6]*(p6-p9);qf=hcos_32[7]*(p7-p8);
p0=q0+q7;p1=q1+q6;p2=q2+q5;p3=q3+q4;
p4=hcos_16[0]*(q0-q7);p5=hcos_16[1]*(q1-q6);
p6=hcos_16[2]*(q2-q5);p7=hcos_16[3]*(q3-q4);
p8=q8+qf;p9=q9+qe;pa=qa+qd;pb=qb+qc;
pc=hcos_16[0]*(q8-qf);pd=hcos_16[1]*(q9-qe);
pe=hcos_16[2]*(qa-qd);pf=hcos_16[3]*(qb-qc);
q0=p0+p3;q1=p1+p2;q2=hcos_8[0]*(p0-p3);q3=hcos_8[1]*(p1-p2);
q4=p4+p7;q5=p5+p6;q6=hcos_8[0]*(p4-p7);q7=hcos_8[1]*(p5-p6);
q8=p8+pb;q9=p9+pa;qa=hcos_8[0]*(p8-pb);qb=hcos_8[1]*(p9-pa);
qc=pc+pf;qd=pd+pe;qe=hcos_8[0]*(pc-pf);qf=hcos_8[1]*(pd-pe);
p0=q0+q1;p1=hcos_4*(q0-q1);
p2=q2+q3;p3=hcos_4*(q2-q3);
p4=q4+q5;p5=hcos_4*(q4-q5);
p6=q6+q7;p7=hcos_4*(q6-q7);
p8=q8+q9;p9=hcos_4*(q8-q9);
pa=qa+qb;pb=hcos_4*(qa-qb);
pc=qc+qd;pd=hcos_4*(qc-qd);
pe=qe+qf;pf=hcos_4*(qe-qf);
{
REAL tmp;
tmp=pd+pf;
OUT1(5,p5+p7+pb+tmp);
tmp+=p9;
OUT1(1,p1+tmp);
OUT2(33)=-(p1+pe+tmp);
tmp+=p5+p7;
OUT1(3,tmp);
OUT2(35)=-(p6+pe+tmp);
tmp=pa+pb+pc+pd+pe+pf;
OUT2(39)=-(p2+p3+tmp-pc);
OUT2(43)=-(p4+p6+p7+tmp-pd);
OUT2(37)=-(p5+p6+p7+tmp-pc);
OUT2(41)=-(p2+p3+tmp-pd);
tmp=p8+pc+pe+pf;
OUT2(47)=-(p0+tmp);
OUT2(45)=-(p4+p6+p7+tmp);
tmp=pb+pf;
OUT1(11,p7+tmp);
tmp+=p3;
OUT1( 9,tmp);
OUT1( 7,pd+tmp);
OUT1(13,p7+pf);
OUT1(15,pf);
}
}
|
