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
|
/*
* Copyright (c) 2006 Luca Barbato <lu_zero@gentoo.org>
*
* This file is part of FFmpeg.
*
* FFmpeg is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
* version 2.1 of the License, or (at your option) any later version.
*
* FFmpeg is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with FFmpeg; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
#include "libavcodec/dsputil.h"
#include "gcc_fixes.h"
#include "dsputil_altivec.h"
static void vector_fmul_altivec(float *dst, const float *src, int len)
{
int i;
vector float d0, d1, s, zero = (vector float)vec_splat_u32(0);
for(i=0; i<len-7; i+=8) {
d0 = vec_ld(0, dst+i);
s = vec_ld(0, src+i);
d1 = vec_ld(16, dst+i);
d0 = vec_madd(d0, s, zero);
d1 = vec_madd(d1, vec_ld(16,src+i), zero);
vec_st(d0, 0, dst+i);
vec_st(d1, 16, dst+i);
}
}
static void vector_fmul_reverse_altivec(float *dst, const float *src0,
const float *src1, int len)
{
int i;
vector float d, s0, s1, h0, l0,
s2, s3, zero = (vector float)vec_splat_u32(0);
src1 += len-4;
for(i=0; i<len-7; i+=8) {
s1 = vec_ld(0, src1-i); // [a,b,c,d]
s0 = vec_ld(0, src0+i);
l0 = vec_mergel(s1, s1); // [c,c,d,d]
s3 = vec_ld(-16, src1-i);
h0 = vec_mergeh(s1, s1); // [a,a,b,b]
s2 = vec_ld(16, src0+i);
s1 = vec_mergeh(vec_mergel(l0,h0), // [d,b,d,b]
vec_mergeh(l0,h0)); // [c,a,c,a]
// [d,c,b,a]
l0 = vec_mergel(s3, s3);
d = vec_madd(s0, s1, zero);
h0 = vec_mergeh(s3, s3);
vec_st(d, 0, dst+i);
s3 = vec_mergeh(vec_mergel(l0,h0),
vec_mergeh(l0,h0));
d = vec_madd(s2, s3, zero);
vec_st(d, 16, dst+i);
}
}
static void vector_fmul_add_add_altivec(float *dst, const float *src0,
const float *src1, const float *src2,
int src3, int len, int step)
{
int i;
vector float d, s0, s1, s2, t0, t1, edges;
vector unsigned char align = vec_lvsr(0,dst),
mask = vec_lvsl(0, dst);
#if 0 //FIXME: there is still something wrong
if (step == 2) {
int y;
vector float d0, d1, s3, t2;
vector unsigned int sel =
vec_mergeh(vec_splat_u32(-1), vec_splat_u32(0));
t1 = vec_ld(16, dst);
for (i=0,y=0; i<len-3; i+=4,y+=8) {
s0 = vec_ld(0,src0+i);
s1 = vec_ld(0,src1+i);
s2 = vec_ld(0,src2+i);
// t0 = vec_ld(0, dst+y); //[x x x|a]
// t1 = vec_ld(16, dst+y); //[b c d|e]
t2 = vec_ld(31, dst+y); //[f g h|x]
d = vec_madd(s0,s1,s2); // [A B C D]
// [A A B B]
// [C C D D]
d0 = vec_perm(t0, t1, mask); // [a b c d]
d0 = vec_sel(vec_mergeh(d, d), d0, sel); // [A b B d]
edges = vec_perm(t1, t0, mask);
t0 = vec_perm(edges, d0, align); // [x x x|A]
t1 = vec_perm(d0, edges, align); // [b B d|e]
vec_stl(t0, 0, dst+y);
d1 = vec_perm(t1, t2, mask); // [e f g h]
d1 = vec_sel(vec_mergel(d, d), d1, sel); // [C f D h]
edges = vec_perm(t2, t1, mask);
t1 = vec_perm(edges, d1, align); // [b B d|C]
t2 = vec_perm(d1, edges, align); // [f D h|x]
vec_stl(t1, 16, dst+y);
t0 = t1;
vec_stl(t2, 31, dst+y);
t1 = t2;
}
} else
#endif
if (step == 1 && src3 == 0)
for (i=0; i<len-3; i+=4) {
t0 = vec_ld(0, dst+i);
t1 = vec_ld(15, dst+i);
s0 = vec_ld(0, src0+i);
s1 = vec_ld(0, src1+i);
s2 = vec_ld(0, src2+i);
edges = vec_perm(t1 ,t0, mask);
d = vec_madd(s0,s1,s2);
t1 = vec_perm(d, edges, align);
t0 = vec_perm(edges, d, align);
vec_st(t1, 15, dst+i);
vec_st(t0, 0, dst+i);
}
else
ff_vector_fmul_add_add_c(dst, src0, src1, src2, src3, len, step);
}
static vector signed short
float_to_int16_one_altivec(const float *src)
{
vector float s0 = vec_ld(0, src);
vector float s1 = vec_ld(16, src);
vector signed int t0 = vec_cts(s0, 0);
vector signed int t1 = vec_cts(s1, 0);
return vec_packs(t0,t1);
}
static void float_to_int16_altivec(int16_t *dst, const float *src, int len)
{
int i;
vector signed short d0, d1, d;
vector unsigned char align;
if(((long)dst)&15) //FIXME
for(i=0; i<len-7; i+=8) {
d0 = vec_ld(0, dst+i);
d = float_to_int16_one_altivec(src+i);
d1 = vec_ld(15, dst+i);
d1 = vec_perm(d1, d0, vec_lvsl(0,dst+i));
align = vec_lvsr(0, dst+i);
d0 = vec_perm(d1, d, align);
d1 = vec_perm(d, d1, align);
vec_st(d0, 0, dst+i);
vec_st(d1,15, dst+i);
}
else
for(i=0; i<len-7; i+=8) {
d = float_to_int16_one_altivec(src+i);
vec_st(d, 0, dst+i);
}
}
static void
float_to_int16_interleave_altivec(int16_t *dst, const float **src,
long len, int channels)
{
int i;
vector signed short d0, d1, d2, c0, c1, t0, t1;
vector unsigned char align;
if(channels == 1)
float_to_int16_altivec(dst, src[0], len);
else
if (channels == 2) {
if(((long)dst)&15)
for(i=0; i<len-7; i+=8) {
d0 = vec_ld(0, dst + i);
t0 = float_to_int16_one_altivec(src[0] + i);
d1 = vec_ld(31, dst + i);
t1 = float_to_int16_one_altivec(src[1] + i);
c0 = vec_mergeh(t0, t1);
c1 = vec_mergel(t0, t1);
d2 = vec_perm(d1, d0, vec_lvsl(0, dst + i));
align = vec_lvsr(0, dst + i);
d0 = vec_perm(d2, c0, align);
d1 = vec_perm(c0, c1, align);
vec_st(d0, 0, dst + i);
d0 = vec_perm(c1, d2, align);
vec_st(d1, 15, dst + i);
vec_st(d0, 31, dst + i);
dst+=8;
}
else
for(i=0; i<len-7; i+=8) {
t0 = float_to_int16_one_altivec(src[0] + i);
t1 = float_to_int16_one_altivec(src[1] + i);
d0 = vec_mergeh(t0, t1);
d1 = vec_mergel(t0, t1);
vec_st(d0, 0, dst + i);
vec_st(d1, 16, dst + i);
dst+=8;
}
} else {
DECLARE_ALIGNED(16, int16_t, tmp[len]);
int c, j;
for (c = 0; c < channels; c++) {
float_to_int16_altivec(tmp, src[c], len);
for (i = 0, j = c; i < len; i++, j+=channels) {
dst[j] = tmp[i];
}
}
}
}
void float_init_altivec(DSPContext* c, AVCodecContext *avctx)
{
c->vector_fmul = vector_fmul_altivec;
c->vector_fmul_reverse = vector_fmul_reverse_altivec;
c->vector_fmul_add_add = vector_fmul_add_add_altivec;
if(!(avctx->flags & CODEC_FLAG_BITEXACT)) {
c->float_to_int16 = float_to_int16_altivec;
c->float_to_int16_interleave = float_to_int16_interleave_altivec;
}
}
|
