Statistics
| Branch: | Revision:

ffmpeg / libavcodec / armv4l / dsputil_arm.c @ be449fca

History | View | Annotate | Download (11.2 KB)

1
/*
2
 * ARMv4L optimized DSP utils
3
 * Copyright (c) 2001 Lionel Ulmer.
4
 *
5
 * This file is part of FFmpeg.
6
 *
7
 * FFmpeg is free software; you can redistribute it and/or
8
 * modify it under the terms of the GNU Lesser General Public
9
 * License as published by the Free Software Foundation; either
10
 * version 2.1 of the License, or (at your option) any later version.
11
 *
12
 * FFmpeg is distributed in the hope that it will be useful,
13
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
15
 * Lesser General Public License for more details.
16
 *
17
 * You should have received a copy of the GNU Lesser General Public
18
 * License along with FFmpeg; if not, write to the Free Software
19
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
20
 */
21

    
22
#include "libavcodec/dsputil.h"
23
#ifdef HAVE_IPP
24
#include <ipp.h>
25
#endif
26

    
27
extern void dsputil_init_iwmmxt(DSPContext* c, AVCodecContext *avctx);
28
extern void ff_float_init_arm_vfp(DSPContext* c, AVCodecContext *avctx);
29

    
30
extern void j_rev_dct_ARM(DCTELEM *data);
31
extern void simple_idct_ARM(DCTELEM *data);
32

    
33
extern void simple_idct_armv5te(DCTELEM *data);
34
extern void simple_idct_put_armv5te(uint8_t *dest, int line_size,
35
                                    DCTELEM *data);
36
extern void simple_idct_add_armv5te(uint8_t *dest, int line_size,
37
                                    DCTELEM *data);
38

    
39
extern void ff_simple_idct_armv6(DCTELEM *data);
40
extern void ff_simple_idct_put_armv6(uint8_t *dest, int line_size,
41
                                     DCTELEM *data);
42
extern void ff_simple_idct_add_armv6(uint8_t *dest, int line_size,
43
                                     DCTELEM *data);
44

    
45
/* XXX: local hack */
46
static void (*ff_put_pixels_clamped)(const DCTELEM *block, uint8_t *pixels, int line_size);
47
static void (*ff_add_pixels_clamped)(const DCTELEM *block, uint8_t *pixels, int line_size);
48

    
49
void put_pixels8_arm(uint8_t *block, const uint8_t *pixels, int line_size, int h);
50
void put_pixels8_x2_arm(uint8_t *block, const uint8_t *pixels, int line_size, int h);
51
void put_pixels8_y2_arm(uint8_t *block, const uint8_t *pixels, int line_size, int h);
52
void put_pixels8_xy2_arm(uint8_t *block, const uint8_t *pixels, int line_size, int h);
53

    
54
void put_no_rnd_pixels8_x2_arm(uint8_t *block, const uint8_t *pixels, int line_size, int h);
55
void put_no_rnd_pixels8_y2_arm(uint8_t *block, const uint8_t *pixels, int line_size, int h);
56
void put_no_rnd_pixels8_xy2_arm(uint8_t *block, const uint8_t *pixels, int line_size, int h);
57

    
58
void put_pixels16_arm(uint8_t *block, const uint8_t *pixels, int line_size, int h);
59

    
60
CALL_2X_PIXELS(put_pixels16_x2_arm , put_pixels8_x2_arm , 8)
61
CALL_2X_PIXELS(put_pixels16_y2_arm , put_pixels8_y2_arm , 8)
62
CALL_2X_PIXELS(put_pixels16_xy2_arm, put_pixels8_xy2_arm, 8)
63
CALL_2X_PIXELS(put_no_rnd_pixels16_x2_arm , put_no_rnd_pixels8_x2_arm , 8)
64
CALL_2X_PIXELS(put_no_rnd_pixels16_y2_arm , put_no_rnd_pixels8_y2_arm , 8)
65
CALL_2X_PIXELS(put_no_rnd_pixels16_xy2_arm, put_no_rnd_pixels8_xy2_arm, 8)
66

    
67
static void add_pixels_clamped_ARM(short *block, unsigned char *dest, int line_size)
68
{
69
    __asm__ volatile (
70
                  "mov r10, #8 \n\t"
71

    
72
                  "1: \n\t"
73

    
74
                  /* load dest */
75
                  "ldr r4, [%1] \n\t"
76
                  /* block[0] and block[1]*/
77
                  "ldrsh r5, [%0] \n\t"
78
                  "ldrsh r7, [%0, #2] \n\t"
79
                  "and r6, r4, #0xFF \n\t"
80
                  "and r8, r4, #0xFF00 \n\t"
81
                  "add r6, r5, r6 \n\t"
82
                  "add r8, r7, r8, lsr #8 \n\t"
83
                  "mvn r5, r5 \n\t"
84
                  "mvn r7, r7 \n\t"
85
                  "tst r6, #0x100 \n\t"
86
                  "movne r6, r5, lsr #24 \n\t"
87
                  "tst r8, #0x100 \n\t"
88
                  "movne r8, r7, lsr #24 \n\t"
89
                  "mov r9, r6 \n\t"
90
                  "ldrsh r5, [%0, #4] \n\t" /* moved form [A] */
91
                  "orr r9, r9, r8, lsl #8 \n\t"
92
                  /* block[2] and block[3] */
93
                  /* [A] */
94
                  "ldrsh r7, [%0, #6] \n\t"
95
                  "and r6, r4, #0xFF0000 \n\t"
96
                  "and r8, r4, #0xFF000000 \n\t"
97
                  "add r6, r5, r6, lsr #16 \n\t"
98
                  "add r8, r7, r8, lsr #24 \n\t"
99
                  "mvn r5, r5 \n\t"
100
                  "mvn r7, r7 \n\t"
101
                  "tst r6, #0x100 \n\t"
102
                  "movne r6, r5, lsr #24 \n\t"
103
                  "tst r8, #0x100 \n\t"
104
                  "movne r8, r7, lsr #24 \n\t"
105
                  "orr r9, r9, r6, lsl #16 \n\t"
106
                  "ldr r4, [%1, #4] \n\t"       /* moved form [B] */
107
                  "orr r9, r9, r8, lsl #24 \n\t"
108
                  /* store dest */
109
                  "ldrsh r5, [%0, #8] \n\t" /* moved form [C] */
110
                  "str r9, [%1] \n\t"
111

    
112
                  /* load dest */
113
                  /* [B] */
114
                  /* block[4] and block[5] */
115
                  /* [C] */
116
                  "ldrsh r7, [%0, #10] \n\t"
117
                  "and r6, r4, #0xFF \n\t"
118
                  "and r8, r4, #0xFF00 \n\t"
119
                  "add r6, r5, r6 \n\t"
120
                  "add r8, r7, r8, lsr #8 \n\t"
121
                  "mvn r5, r5 \n\t"
122
                  "mvn r7, r7 \n\t"
123
                  "tst r6, #0x100 \n\t"
124
                  "movne r6, r5, lsr #24 \n\t"
125
                  "tst r8, #0x100 \n\t"
126
                  "movne r8, r7, lsr #24 \n\t"
127
                  "mov r9, r6 \n\t"
128
                  "ldrsh r5, [%0, #12] \n\t" /* moved from [D] */
129
                  "orr r9, r9, r8, lsl #8 \n\t"
130
                  /* block[6] and block[7] */
131
                  /* [D] */
132
                  "ldrsh r7, [%0, #14] \n\t"
133
                  "and r6, r4, #0xFF0000 \n\t"
134
                  "and r8, r4, #0xFF000000 \n\t"
135
                  "add r6, r5, r6, lsr #16 \n\t"
136
                  "add r8, r7, r8, lsr #24 \n\t"
137
                  "mvn r5, r5 \n\t"
138
                  "mvn r7, r7 \n\t"
139
                  "tst r6, #0x100 \n\t"
140
                  "movne r6, r5, lsr #24 \n\t"
141
                  "tst r8, #0x100 \n\t"
142
                  "movne r8, r7, lsr #24 \n\t"
143
                  "orr r9, r9, r6, lsl #16 \n\t"
144
                  "add %0, %0, #16 \n\t" /* moved from [E] */
145
                  "orr r9, r9, r8, lsl #24 \n\t"
146
                  "subs r10, r10, #1 \n\t" /* moved from [F] */
147
                  /* store dest */
148
                  "str r9, [%1, #4] \n\t"
149

    
150
                  /* [E] */
151
                  /* [F] */
152
                  "add %1, %1, %2 \n\t"
153
                  "bne 1b \n\t"
154
                  : "+r"(block),
155
                    "+r"(dest)
156
                  : "r"(line_size)
157
                  : "r4", "r5", "r6", "r7", "r8", "r9", "r10", "cc", "memory" );
158
}
159

    
160
/* XXX: those functions should be suppressed ASAP when all IDCTs are
161
   converted */
162
static void j_rev_dct_ARM_put(uint8_t *dest, int line_size, DCTELEM *block)
163
{
164
    j_rev_dct_ARM (block);
165
    ff_put_pixels_clamped(block, dest, line_size);
166
}
167
static void j_rev_dct_ARM_add(uint8_t *dest, int line_size, DCTELEM *block)
168
{
169
    j_rev_dct_ARM (block);
170
    ff_add_pixels_clamped(block, dest, line_size);
171
}
172
static void simple_idct_ARM_put(uint8_t *dest, int line_size, DCTELEM *block)
173
{
174
    simple_idct_ARM (block);
175
    ff_put_pixels_clamped(block, dest, line_size);
176
}
177
static void simple_idct_ARM_add(uint8_t *dest, int line_size, DCTELEM *block)
178
{
179
    simple_idct_ARM (block);
180
    ff_add_pixels_clamped(block, dest, line_size);
181
}
182

    
183
#ifdef HAVE_IPP
184
static void simple_idct_ipp(DCTELEM *block)
185
{
186
    ippiDCT8x8Inv_Video_16s_C1I(block);
187
}
188
static void simple_idct_ipp_put(uint8_t *dest, int line_size, DCTELEM *block)
189
{
190
    ippiDCT8x8Inv_Video_16s8u_C1R(block, dest, line_size);
191
}
192

    
193
void add_pixels_clamped_iwmmxt(const DCTELEM *block, uint8_t *pixels, int line_size);
194

    
195
static void simple_idct_ipp_add(uint8_t *dest, int line_size, DCTELEM *block)
196
{
197
    ippiDCT8x8Inv_Video_16s_C1I(block);
198
#ifdef HAVE_IWMMXT
199
    add_pixels_clamped_iwmmxt(block, dest, line_size);
200
#else
201
    add_pixels_clamped_ARM(block, dest, line_size);
202
#endif
203
}
204
#endif
205

    
206
#ifdef HAVE_ARMV5TE
207
static void prefetch_arm(void *mem, int stride, int h)
208
{
209
    __asm__ volatile(
210
        "1:              \n\t"
211
        "subs %0, %0, #1 \n\t"
212
        "pld  [%1]       \n\t"
213
        "add  %1, %1, %2 \n\t"
214
        "bgt  1b         \n\t"
215
        : "+r"(h), "+r"(mem) : "r"(stride));
216
}
217
#endif
218

    
219
int mm_support(void)
220
{
221
    return ENABLE_IWMMXT * MM_IWMMXT;
222
}
223

    
224
void dsputil_init_armv4l(DSPContext* c, AVCodecContext *avctx)
225
{
226
    int idct_algo= avctx->idct_algo;
227

    
228
    ff_put_pixels_clamped = c->put_pixels_clamped;
229
    ff_add_pixels_clamped = c->add_pixels_clamped;
230

    
231
    if (avctx->lowres == 0) {
232
        if(idct_algo == FF_IDCT_AUTO){
233
#if defined(HAVE_IPP)
234
            idct_algo = FF_IDCT_IPP;
235
#elif defined(HAVE_ARMV6)
236
            idct_algo = FF_IDCT_SIMPLEARMV6;
237
#elif defined(HAVE_ARMV5TE)
238
            idct_algo = FF_IDCT_SIMPLEARMV5TE;
239
#else
240
            idct_algo = FF_IDCT_ARM;
241
#endif
242
        }
243

    
244
        if(idct_algo==FF_IDCT_ARM){
245
            c->idct_put= j_rev_dct_ARM_put;
246
            c->idct_add= j_rev_dct_ARM_add;
247
            c->idct    = j_rev_dct_ARM;
248
            c->idct_permutation_type= FF_LIBMPEG2_IDCT_PERM;/* FF_NO_IDCT_PERM */
249
        } else if (idct_algo==FF_IDCT_SIMPLEARM){
250
            c->idct_put= simple_idct_ARM_put;
251
            c->idct_add= simple_idct_ARM_add;
252
            c->idct    = simple_idct_ARM;
253
            c->idct_permutation_type= FF_NO_IDCT_PERM;
254
#ifdef HAVE_ARMV6
255
        } else if (idct_algo==FF_IDCT_SIMPLEARMV6){
256
            c->idct_put= ff_simple_idct_put_armv6;
257
            c->idct_add= ff_simple_idct_add_armv6;
258
            c->idct    = ff_simple_idct_armv6;
259
            c->idct_permutation_type= FF_LIBMPEG2_IDCT_PERM;
260
#endif
261
#ifdef HAVE_ARMV5TE
262
        } else if (idct_algo==FF_IDCT_SIMPLEARMV5TE){
263
            c->idct_put= simple_idct_put_armv5te;
264
            c->idct_add= simple_idct_add_armv5te;
265
            c->idct    = simple_idct_armv5te;
266
            c->idct_permutation_type = FF_NO_IDCT_PERM;
267
#endif
268
#ifdef HAVE_IPP
269
        } else if (idct_algo==FF_IDCT_IPP){
270
            c->idct_put= simple_idct_ipp_put;
271
            c->idct_add= simple_idct_ipp_add;
272
            c->idct    = simple_idct_ipp;
273
            c->idct_permutation_type= FF_NO_IDCT_PERM;
274
#endif
275
        }
276
    }
277

    
278
    c->put_pixels_tab[0][0] = put_pixels16_arm;
279
    c->put_pixels_tab[0][1] = put_pixels16_x2_arm; //OK!
280
    c->put_pixels_tab[0][2] = put_pixels16_y2_arm; //OK!
281
    c->put_pixels_tab[0][3] = put_pixels16_xy2_arm;
282
    c->put_no_rnd_pixels_tab[0][0] = put_pixels16_arm;
283
    c->put_no_rnd_pixels_tab[0][1] = put_no_rnd_pixels16_x2_arm; // OK
284
    c->put_no_rnd_pixels_tab[0][2] = put_no_rnd_pixels16_y2_arm; //OK
285
    c->put_no_rnd_pixels_tab[0][3] = put_no_rnd_pixels16_xy2_arm;
286
    c->put_pixels_tab[1][0] = put_pixels8_arm; //OK
287
    c->put_pixels_tab[1][1] = put_pixels8_x2_arm; //OK
288
    c->put_pixels_tab[1][2] = put_pixels8_y2_arm;
289
    c->put_pixels_tab[1][3] = put_pixels8_xy2_arm;
290
    c->put_no_rnd_pixels_tab[1][0] = put_pixels8_arm;//OK
291
    c->put_no_rnd_pixels_tab[1][1] = put_no_rnd_pixels8_x2_arm; //OK
292
    c->put_no_rnd_pixels_tab[1][2] = put_no_rnd_pixels8_y2_arm; //OK
293
    c->put_no_rnd_pixels_tab[1][3] = put_no_rnd_pixels8_xy2_arm;
294

    
295
#ifdef HAVE_ARMV5TE
296
    c->prefetch = prefetch_arm;
297
#endif
298

    
299
#ifdef HAVE_IWMMXT
300
    dsputil_init_iwmmxt(c, avctx);
301
#endif
302
#ifdef HAVE_ARMVFP
303
    ff_float_init_arm_vfp(c, avctx);
304
#endif
305
}