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
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
|
/*
* rgb_mmx.c
* Copyright (C) 2000-2003 Silicon Integrated System Corp.
* All Rights Reserved.
*
* Author: Olie Lho <ollie@sis.com.tw>
*
* This file is part of mpeg2dec, a free MPEG-2 video stream decoder.
* See http://libmpeg2.sourceforge.net/ for updates.
*
* mpeg2dec is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* mpeg2dec 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 General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
#include "config.h"
#ifdef ARCH_X86
#include "k9common.h"
#include <stdio.h>
#include <stdlib.h>
#include "mpeg2.h"
#include "mpeg2convert.h"
#include "convert_internal.h"
#include "attributes.h"
#include "mmx.h"
#define CPU_MMXEXT 0
#define CPU_MMX 1
/* CPU_MMXEXT/CPU_MMX adaptation layer */
#define movntq(src,dest) \
do { \
if (cpu == CPU_MMXEXT) \
movntq_r2m (src, dest); \
else \
movq_r2m (src, dest); \
} while (0)
static inline void mmx_yuv2rgb (uint8_t * py, uint8_t * pu, uint8_t * pv)
{
static mmx_t mmx_80w = {0x0080008000800080LL};
static mmx_t mmx_U_green = {0xf37df37df37df37dLL};
static mmx_t mmx_U_blue = {0x4093409340934093LL};
static mmx_t mmx_V_red = {0x3312331233123312LL};
static mmx_t mmx_V_green = {0xe5fce5fce5fce5fcLL};
static mmx_t mmx_10w = {0x1010101010101010LL};
static mmx_t mmx_00ffw = {0x00ff00ff00ff00ffLL};
static mmx_t mmx_Y_coeff = {0x253f253f253f253fLL};
movd_m2r (*pu, mm0); /* mm0 = 00 00 00 00 u3 u2 u1 u0 */
movd_m2r (*pv, mm1); /* mm1 = 00 00 00 00 v3 v2 v1 v0 */
movq_m2r (*py, mm6); /* mm6 = Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0 */
pxor_r2r (mm4, mm4); /* mm4 = 0 */
/* XXX might do cache preload for image here */
/*
* Do the multiply part of the conversion for even and odd pixels
* register usage:
* mm0 -> Cblue, mm1 -> Cred, mm2 -> Cgreen even pixels
* mm3 -> Cblue, mm4 -> Cred, mm5 -> Cgreen odd pixels
* mm6 -> Y even, mm7 -> Y odd
*/
punpcklbw_r2r (mm4, mm0); /* mm0 = u3 u2 u1 u0 */
punpcklbw_r2r (mm4, mm1); /* mm1 = v3 v2 v1 v0 */
psubsw_m2r (mmx_80w, mm0); /* u -= 128 */
psubsw_m2r (mmx_80w, mm1); /* v -= 128 */
psllw_i2r (3, mm0); /* promote precision */
psllw_i2r (3, mm1); /* promote precision */
movq_r2r (mm0, mm2); /* mm2 = u3 u2 u1 u0 */
movq_r2r (mm1, mm3); /* mm3 = v3 v2 v1 v0 */
pmulhw_m2r (mmx_U_green, mm2); /* mm2 = u * u_green */
pmulhw_m2r (mmx_V_green, mm3); /* mm3 = v * v_green */
pmulhw_m2r (mmx_U_blue, mm0); /* mm0 = chroma_b */
pmulhw_m2r (mmx_V_red, mm1); /* mm1 = chroma_r */
paddsw_r2r (mm3, mm2); /* mm2 = chroma_g */
psubusb_m2r (mmx_10w, mm6); /* Y -= 16 */
movq_r2r (mm6, mm7); /* mm7 = Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0 */
pand_m2r (mmx_00ffw, mm6); /* mm6 = Y6 Y4 Y2 Y0 */
psrlw_i2r (8, mm7); /* mm7 = Y7 Y5 Y3 Y1 */
psllw_i2r (3, mm6); /* promote precision */
psllw_i2r (3, mm7); /* promote precision */
pmulhw_m2r (mmx_Y_coeff, mm6); /* mm6 = luma_rgb even */
pmulhw_m2r (mmx_Y_coeff, mm7); /* mm7 = luma_rgb odd */
/*
* Do the addition part of the conversion for even and odd pixels
* register usage:
* mm0 -> Cblue, mm1 -> Cred, mm2 -> Cgreen even pixels
* mm3 -> Cblue, mm4 -> Cred, mm5 -> Cgreen odd pixels
* mm6 -> Y even, mm7 -> Y odd
*/
movq_r2r (mm0, mm3); /* mm3 = chroma_b */
movq_r2r (mm1, mm4); /* mm4 = chroma_r */
movq_r2r (mm2, mm5); /* mm5 = chroma_g */
paddsw_r2r (mm6, mm0); /* mm0 = B6 B4 B2 B0 */
paddsw_r2r (mm7, mm3); /* mm3 = B7 B5 B3 B1 */
paddsw_r2r (mm6, mm1); /* mm1 = R6 R4 R2 R0 */
paddsw_r2r (mm7, mm4); /* mm4 = R7 R5 R3 R1 */
paddsw_r2r (mm6, mm2); /* mm2 = G6 G4 G2 G0 */
paddsw_r2r (mm7, mm5); /* mm5 = G7 G5 G3 G1 */
packuswb_r2r (mm0, mm0); /* saturate to 0-255 */
packuswb_r2r (mm1, mm1); /* saturate to 0-255 */
packuswb_r2r (mm2, mm2); /* saturate to 0-255 */
packuswb_r2r (mm3, mm3); /* saturate to 0-255 */
packuswb_r2r (mm4, mm4); /* saturate to 0-255 */
packuswb_r2r (mm5, mm5); /* saturate to 0-255 */
punpcklbw_r2r (mm3, mm0); /* mm0 = B7 B6 B5 B4 B3 B2 B1 B0 */
punpcklbw_r2r (mm4, mm1); /* mm1 = R7 R6 R5 R4 R3 R2 R1 R0 */
punpcklbw_r2r (mm5, mm2); /* mm2 = G7 G6 G5 G4 G3 G2 G1 G0 */
}
static inline void mmx_unpack_16rgb (uint8_t * image, const int cpu)
{
static mmx_t mmx_bluemask = {0xf8f8f8f8f8f8f8f8LL};
static mmx_t mmx_greenmask = {0xfcfcfcfcfcfcfcfcLL};
static mmx_t mmx_redmask = {0xf8f8f8f8f8f8f8f8LL};
/*
* convert RGB plane to RGB 16 bits
* mm0 -> B, mm1 -> R, mm2 -> G
* mm4 -> GB, mm5 -> AR pixel 4-7
* mm6 -> GB, mm7 -> AR pixel 0-3
*/
pand_m2r (mmx_bluemask, mm0); /* mm0 = b7b6b5b4b3______ */
pand_m2r (mmx_greenmask, mm2); /* mm2 = g7g6g5g4g3g2____ */
pand_m2r (mmx_redmask, mm1); /* mm1 = r7r6r5r4r3______ */
psrlq_i2r (3, mm0); /* mm0 = ______b7b6b5b4b3 */
pxor_r2r (mm4, mm4); /* mm4 = 0 */
movq_r2r (mm0, mm5); /* mm5 = ______b7b6b5b4b3 */
movq_r2r (mm2, mm7); /* mm7 = g7g6g5g4g3g2____ */
punpcklbw_r2r (mm4, mm2);
punpcklbw_r2r (mm1, mm0);
psllq_i2r (3, mm2);
por_r2r (mm2, mm0);
movntq (mm0, *image);
punpckhbw_r2r (mm4, mm7);
punpckhbw_r2r (mm1, mm5);
psllq_i2r (3, mm7);
por_r2r (mm7, mm5);
movntq (mm5, *(image+8));
}
static inline void mmx_unpack_32rgb (uint8_t * image, const int cpu)
{
/*
* convert RGB plane to RGB packed format,
* mm0 -> B, mm1 -> R, mm2 -> G, mm3 -> 0,
* mm4 -> GB, mm5 -> AR pixel 4-7,
* mm6 -> GB, mm7 -> AR pixel 0-3
*/
pxor_r2r (mm3, mm3);
movq_r2r (mm0, mm6);
movq_r2r (mm1, mm7);
movq_r2r (mm0, mm4);
movq_r2r (mm1, mm5);
punpcklbw_r2r (mm2, mm6);
punpcklbw_r2r (mm3, mm7);
punpcklwd_r2r (mm7, mm6);
movntq (mm6, *image);
movq_r2r (mm0, mm6);
punpcklbw_r2r (mm2, mm6);
punpckhwd_r2r (mm7, mm6);
movntq (mm6, *(image+8));
punpckhbw_r2r (mm2, mm4);
punpckhbw_r2r (mm3, mm5);
punpcklwd_r2r (mm5, mm4);
movntq (mm4, *(image+16));
movq_r2r (mm0, mm4);
punpckhbw_r2r (mm2, mm4);
punpckhwd_r2r (mm5, mm4);
movntq (mm4, *(image+24));
}
static inline void rgb16 (void * const _id, uint8_t * const * src,
const unsigned int v_offset, const int cpu)
{
convert_rgb_t * const id = (convert_rgb_t *) _id;
uint8_t * dst;
uint8_t * py, * pu, * pv;
int i, j;
dst = id->rgb_ptr + id->rgb_stride * v_offset;
py = src[0]; pu = src[1]; pv = src[2];
i = 16;
do {
j = id->width;
do {
mmx_yuv2rgb (py, pu, pv);
mmx_unpack_16rgb (dst, cpu);
py += 8;
pu += 4;
pv += 4;
dst += 16;
} while (--j);
dst += id->rgb_increm;
py += id->y_increm;
if (! (--i & id->chroma420)) {
pu += id->uv_increm;
pv += id->uv_increm;
} else {
pu -= id->uv_stride_frame;
pv -= id->uv_stride_frame;
}
} while (i);
}
static inline void argb32 (void * const _id, uint8_t * const * src,
const unsigned int v_offset, const int cpu)
{
convert_rgb_t * const id = (convert_rgb_t *) _id;
uint8_t * dst;
uint8_t * py, * pu, * pv;
int i, j;
dst = id->rgb_ptr + id->rgb_stride * v_offset;
py = src[0]; pu = src[1]; pv = src[2];
i = 16;
do {
j = id->width;
do {
mmx_yuv2rgb (py, pu, pv);
mmx_unpack_32rgb (dst, cpu);
py += 8;
pu += 4;
pv += 4;
dst += 32;
} while (--j);
dst += id->rgb_increm;
py += id->y_increm;
if (! (--i & id->chroma420)) {
pu += id->uv_increm;
pv += id->uv_increm;
} else {
pu -= id->uv_stride_frame;
pv -= id->uv_stride_frame;
}
} while (i);
}
static void mmxext_rgb16 (void * id, uint8_t * const * src,
unsigned int v_offset)
{
rgb16 (id, src, v_offset, CPU_MMXEXT);
}
static void mmxext_argb32 (void * id, uint8_t * const * src,
unsigned int v_offset)
{
argb32 (id, src, v_offset, CPU_MMXEXT);
}
static void mmx_rgb16 (void * id, uint8_t * const * src, unsigned int v_offset)
{
rgb16 (id, src, v_offset, CPU_MMX);
}
static void mmx_argb32 (void * id, uint8_t * const * src,
unsigned int v_offset)
{
argb32 (id, src, v_offset, CPU_MMX);
}
mpeg2convert_copy_t * mpeg2convert_rgb_mmxext (int order, int bpp,
const mpeg2_sequence_t * seq)
{
if (order == MPEG2CONVERT_RGB && seq->chroma_width < seq->width) {
if (bpp == 16)
return mmxext_rgb16;
else if (bpp == 32)
return mmxext_argb32;
}
return NULL; /* Fallback to C */
}
mpeg2convert_copy_t * mpeg2convert_rgb_mmx (int order, int bpp,
const mpeg2_sequence_t * seq)
{
if (order == MPEG2CONVERT_RGB && seq->chroma_width < seq->width) {
if (bpp == 16)
return mmx_rgb16;
else if (bpp == 32)
return mmx_argb32;
}
return NULL; /* Fallback to C */
}
#endif
|
