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
|
--- avidemux/ADM_colorspace/yv12rgb_altivec.cpp.bak 2006-01-07 23:37:35.000000000 -0500
+++ avidemux/ADM_colorspace/yv12rgb_altivec.cpp 2006-01-07 23:38:46.000000000 -0500
@@ -53,9 +53,9 @@
#define vecbyte vector unsigned char
#define vecshort vector unsigned short
-uint8_t altivecYV12RGB(uint8_t * ptr_y,
+void altivecYV12RGB(uint8_t * ptr_y,
uint8_t * ptr_u,
- uint8_t * ptr_v, uint8_t * ptr2, uint32_t w);
+ uint8_t * ptr_v, uint8_t * ptr2, int w);
// A pack of constant vectors we will use later
// we more or less rely on gcc to have good register allocation
@@ -78,9 +78,9 @@
// We handle 8 pixels at a time
// My first altivec code :)
//____________________________________
-uint8_t altivecYV12RGB(uint8_t * ptr_y,
+void altivecYV12RGB(uint8_t * ptr_y,
uint8_t * ptr_u,
- uint8_t * ptr_v, uint8_t * ptr2, uint32_t w)
+ uint8_t * ptr_v, uint8_t * ptr2, int w)
{
vector unsigned char y,y2,u,v;
@@ -280,7 +280,7 @@
}
- return 1;
+ return;
}
//
--- avidemux/ADM_colorspace/yv2bmp.cpp.bak 2006-01-07 23:40:11.000000000 -0500
+++ avidemux/ADM_colorspace/yv2bmp.cpp 2006-01-07 23:39:24.000000000 -0500
@@ -29,9 +29,9 @@
#include "colorspace.h"
#include "ADM_toolkit/ADM_cpuCap.h"
#ifdef HAVE_ALTIVEC
-extern uint8_t altivecYV12RGB(uint8_t * ptr_y,
+extern void altivecYV12RGB(uint8_t * ptr_y,
uint8_t * ptr_u,
- uint8_t * ptr_v, uint8_t * ptr2, uint32_t w);
+ uint8_t * ptr_v, uint8_t * ptr2, int w);
#endif
static void swapRnB (uint8_t * ptr, uint32_t w);
--- adm_lavcodec/libpostproc/postprocess_altivec_template.c.bak 2006-01-08 01:52:55.000000000 -0500
+++ adm_lavcodec/libpostproc/postprocess_altivec_template.c 2006-01-08 01:52:14.000000000 -0500
@@ -1228,12 +1228,12 @@
- const VECS16 vec_m=(VECS16)(0,1,2,3,3,2,1,0);
- const VECS16 vec_m2=(VECS16)(0,-1,0,2,0,-1,0,0);
- const VECS16 vec_sign=(VECS16)(0,1,1,1,-1,-1,-1,0);
- const VECS16 vec_sign2=(VECS16)(0,-1,-1,-1,1,1,1,0);
+ const VECS16 vec_m=(VECS16)AVV(0,1,2,3,3,2,1,0);
+ const VECS16 vec_m2=(VECS16)AVV(0,-1,0,2,0,-1,0,0);
+ const VECS16 vec_sign=(VECS16)AVV(0,1,1,1,-1,-1,-1,0);
+ const VECS16 vec_sign2=(VECS16)AVV(0,-1,-1,-1,1,1,1,0);
- const VEC16 vec_const=(VEC16)(4,0x11,0xFFFF,0,0,0,0,0);
+ const VEC16 vec_const=(VEC16)AVV(4,0x11,0xFFFF,0,0,0,0,0);
const VECS8 neg1=(VECS8)vec_splat(vec_const,2); // -1
const VEC16 vec_shift=vec_splat(vec_const,0); // 3
--- adm_lavcodec/libpostproc/postprocess_template.c.bak 2006-01-08 01:57:42.000000000 -0500
+++ adm_lavcodec/libpostproc/postprocess_template.c 2006-01-08 01:54:45.000000000 -0500
@@ -513,7 +513,7 @@
- const VEC16 vec_const=(VEC16)(4,0x11,0xFFFF,0,0,0,0,0);
+ const VEC16 vec_const=(VEC16)AVV(4,0x11,0xFFFF,0,0,0,0,0);
const VEC8 vec_zero=(VEC8)vec_splat(vec_const,3); //0
const VEC16 neg1=vec_splat(vec_const,2); //FFFFF
const VEC16 vec_shift=vec_splat(vec_const,0); //4
--- adm_lavcodec/ppc/fdct_altivec.c.bak 2006-07-08 16:51:08.000000000 -0400
+++ adm_lavcodec/ppc/fdct_altivec.c 2006-07-08 18:04:23.000000000 -0400
@@ -18,7 +18,7 @@
*/
-#include "../common.h"
+#include "common.h"
#include "../dsputil.h"
#include "dsputil_altivec.h"
#include "gcc_fixes.h"
@@ -214,8 +214,8 @@
/* setup constants {{{ */
/* mzero = -0.0 */
- vu32(mzero) = vec_splat_u32(-1);
- vu32(mzero) = vec_sl(vu32(mzero), vu32(mzero));
+ mzero = ((vector float)vec_splat_u32(-1));
+ mzero = ((vector float)vec_sl(vu32(mzero), vu32(mzero)));
cp = fdctconsts;
cnsts0 = vec_ld(0, cp); cp++;
cnsts1 = vec_ld(0, cp); cp++;
@@ -227,43 +227,43 @@
#define MERGE_S16(hl,a,b) vec_merge##hl(vs16(a), vs16(b))
bp = (vector signed short*)block;
- vs16(b00) = vec_ld(0, bp);
- vs16(b40) = vec_ld(16*4, bp);
- vs16(b01) = MERGE_S16(h, b00, b40);
- vs16(b11) = MERGE_S16(l, b00, b40);
+ b00 = ((vector float)vec_ld(0, bp));
+ b40 = ((vector float)vec_ld(16*4, bp));
+ b01 = ((vector float)MERGE_S16(h, b00, b40));
+ b11 = ((vector float)MERGE_S16(l, b00, b40));
bp++;
- vs16(b10) = vec_ld(0, bp);
- vs16(b50) = vec_ld(16*4, bp);
- vs16(b21) = MERGE_S16(h, b10, b50);
- vs16(b31) = MERGE_S16(l, b10, b50);
+ b10 = ((vector float)vec_ld(0, bp));
+ b50 = ((vector float)vec_ld(16*4, bp));
+ b21 = ((vector float)MERGE_S16(h, b10, b50));
+ b31 = ((vector float)MERGE_S16(l, b10, b50));
bp++;
- vs16(b20) = vec_ld(0, bp);
- vs16(b60) = vec_ld(16*4, bp);
- vs16(b41) = MERGE_S16(h, b20, b60);
- vs16(b51) = MERGE_S16(l, b20, b60);
+ b20 = ((vector float)vec_ld(0, bp));
+ b60 = ((vector float)vec_ld(16*4, bp));
+ b41 = ((vector float)MERGE_S16(h, b20, b60));
+ b51 = ((vector float)MERGE_S16(l, b20, b60));
bp++;
- vs16(b30) = vec_ld(0, bp);
- vs16(b70) = vec_ld(16*4, bp);
- vs16(b61) = MERGE_S16(h, b30, b70);
- vs16(b71) = MERGE_S16(l, b30, b70);
-
- vs16(x0) = MERGE_S16(h, b01, b41);
- vs16(x1) = MERGE_S16(l, b01, b41);
- vs16(x2) = MERGE_S16(h, b11, b51);
- vs16(x3) = MERGE_S16(l, b11, b51);
- vs16(x4) = MERGE_S16(h, b21, b61);
- vs16(x5) = MERGE_S16(l, b21, b61);
- vs16(x6) = MERGE_S16(h, b31, b71);
- vs16(x7) = MERGE_S16(l, b31, b71);
-
- vs16(b00) = MERGE_S16(h, x0, x4);
- vs16(b10) = MERGE_S16(l, x0, x4);
- vs16(b20) = MERGE_S16(h, x1, x5);
- vs16(b30) = MERGE_S16(l, x1, x5);
- vs16(b40) = MERGE_S16(h, x2, x6);
- vs16(b50) = MERGE_S16(l, x2, x6);
- vs16(b60) = MERGE_S16(h, x3, x7);
- vs16(b70) = MERGE_S16(l, x3, x7);
+ b30 = ((vector float)vec_ld(0, bp));
+ b70 = ((vector float)vec_ld(16*4, bp));
+ b61 = ((vector float)MERGE_S16(h, b30, b70));
+ b71 = ((vector float)MERGE_S16(l, b30, b70));
+
+ x0 = ((vector float)MERGE_S16(h, b01, b41));
+ x1 = ((vector float)MERGE_S16(l, b01, b41));
+ x2 = ((vector float)MERGE_S16(h, b11, b51));
+ x3 = ((vector float)MERGE_S16(l, b11, b51));
+ x4 = ((vector float)MERGE_S16(h, b21, b61));
+ x5 = ((vector float)MERGE_S16(l, b21, b61));
+ x6 = ((vector float)MERGE_S16(h, b31, b71));
+ x7 = ((vector float)MERGE_S16(l, b31, b71));
+
+ b00 = ((vector float)MERGE_S16(h, x0, x4));
+ b10 = ((vector float)MERGE_S16(l, x0, x4));
+ b20 = ((vector float)MERGE_S16(h, x1, x5));
+ b30 = ((vector float)MERGE_S16(l, x1, x5));
+ b40 = ((vector float)MERGE_S16(h, x2, x6));
+ b50 = ((vector float)MERGE_S16(l, x2, x6));
+ b60 = ((vector float)MERGE_S16(h, x3, x7));
+ b70 = ((vector float)MERGE_S16(l, x3, x7));
#undef MERGE_S16
/* }}} */
@@ -275,32 +275,32 @@
*/
#if 1
/* fdct rows {{{ */
- vs16(x0) = vec_add(vs16(b00), vs16(b70));
- vs16(x7) = vec_sub(vs16(b00), vs16(b70));
- vs16(x1) = vec_add(vs16(b10), vs16(b60));
- vs16(x6) = vec_sub(vs16(b10), vs16(b60));
- vs16(x2) = vec_add(vs16(b20), vs16(b50));
- vs16(x5) = vec_sub(vs16(b20), vs16(b50));
- vs16(x3) = vec_add(vs16(b30), vs16(b40));
- vs16(x4) = vec_sub(vs16(b30), vs16(b40));
+ x0 = ((vector float)vec_add(vs16(b00), vs16(b70)));
+ x7 = ((vector float)vec_sub(vs16(b00), vs16(b70)));
+ x1 = ((vector float)vec_add(vs16(b10), vs16(b60)));
+ x6 = ((vector float)vec_sub(vs16(b10), vs16(b60)));
+ x2 = ((vector float)vec_add(vs16(b20), vs16(b50)));
+ x5 = ((vector float)vec_sub(vs16(b20), vs16(b50)));
+ x3 = ((vector float)vec_add(vs16(b30), vs16(b40)));
+ x4 = ((vector float)vec_sub(vs16(b30), vs16(b40)));
- vs16(b70) = vec_add(vs16(x0), vs16(x3));
- vs16(b10) = vec_add(vs16(x1), vs16(x2));
+ b70 = ((vector float)vec_add(vs16(x0), vs16(x3)));
+ b10 = ((vector float)vec_add(vs16(x1), vs16(x2)));
- vs16(b00) = vec_add(vs16(b70), vs16(b10));
- vs16(b40) = vec_sub(vs16(b70), vs16(b10));
+ b00 = ((vector float)vec_add(vs16(b70), vs16(b10)));
+ b40 = ((vector float)vec_sub(vs16(b70), vs16(b10)));
#define CTF0(n) \
- vs32(b##n##1) = vec_unpackl(vs16(b##n##0)); \
- vs32(b##n##0) = vec_unpackh(vs16(b##n##0)); \
+ b##n##1 = ((vector float)vec_unpackl(vs16(b##n##0))); \
+ b##n##0 = ((vector float)vec_unpackh(vs16(b##n##0))); \
b##n##1 = vec_ctf(vs32(b##n##1), 0); \
b##n##0 = vec_ctf(vs32(b##n##0), 0);
CTF0(0);
CTF0(4);
- vs16(b20) = vec_sub(vs16(x0), vs16(x3));
- vs16(b60) = vec_sub(vs16(x1), vs16(x2));
+ b20 = ((vector float)vec_sub(vs16(x0), vs16(x3)));
+ b60 = ((vector float)vec_sub(vs16(x1), vs16(x2)));
CTF0(2);
CTF0(6);
@@ -321,8 +321,8 @@
b61 = vec_madd(cnst, b61, x1);
#define CTFX(x,b) \
- vs32(b##0) = vec_unpackh(vs16(x)); \
- vs32(b##1) = vec_unpackl(vs16(x)); \
+ b##0 = ((vector float)vec_unpackh(vs16(x))); \
+ b##1 = ((vector float)vec_unpackl(vs16(x))); \
b##0 = vec_ctf(vs32(b##0), 0); \
b##1 = vec_ctf(vs32(b##1), 0); \
@@ -473,9 +473,9 @@
#define CTS(n) \
b##n##0 = vec_round(b##n##0); \
b##n##1 = vec_round(b##n##1); \
- vs32(b##n##0) = vec_cts(b##n##0, 0); \
- vs32(b##n##1) = vec_cts(b##n##1, 0); \
- vs16(b##n##0) = vec_pack(vs32(b##n##0), vs32(b##n##1)); \
+ b##n##0 = ((vector float)vec_cts(b##n##0, 0)); \
+ b##n##1 = ((vector float)vec_cts(b##n##1, 0)); \
+ b##n##0 = ((vector float)vec_pack(vs32(b##n##0), vs32(b##n##1))); \
vec_st(vs16(b##n##0), 0, bp);
bp = (vector signed short*)block;
|
GitWeb