summaryrefslogtreecommitdiff
path: root/libavcodec/sh4/idct_sh4.c
blob: 8d1a31670f8a431c467d7a62b4aabdaa8b0126ee (plain)
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
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
/*
 * idct for sh4
 *
 * Copyright (c) 2001-2003 BERO <bero@geocities.co.jp>
 *
 * 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 "dsputil_sh4.h"
#include "sh4.h"

#define c1      1.38703984532214752434  /* sqrt(2)*cos(1*pi/16) */
#define c2      1.30656296487637657577  /* sqrt(2)*cos(2*pi/16) */
#define c3      1.17587560241935884520  /* sqrt(2)*cos(3*pi/16) */
#define c4      1.00000000000000000000  /* sqrt(2)*cos(4*pi/16) */
#define c5      0.78569495838710234903  /* sqrt(2)*cos(5*pi/16) */
#define c6      0.54119610014619712324  /* sqrt(2)*cos(6*pi/16) */
#define c7      0.27589937928294311353  /* sqrt(2)*cos(7*pi/16) */

static const float even_table[] __attribute__ ((aligned(8))) = {
        c4, c4, c4, c4,
        c2, c6,-c6,-c2,
        c4,-c4,-c4, c4,
        c6,-c2, c2,-c6
};

static const float odd_table[] __attribute__ ((aligned(8))) = {
        c1, c3, c5, c7,
        c3,-c7,-c1,-c5,
        c5,-c1, c7, c3,
        c7,-c5, c3,-c1
};

#undef  c1
#undef  c2
#undef  c3
#undef  c4
#undef  c5
#undef  c6
#undef  c7

#if 1

#define         load_matrix(table) \
    do { \
        const float *t = table; \
        __asm__ volatile( \
        "       fschg\n" \
        "       fmov   @%0+,xd0\n" \
        "       fmov   @%0+,xd2\n" \
        "       fmov   @%0+,xd4\n" \
        "       fmov   @%0+,xd6\n" \
        "       fmov   @%0+,xd8\n" \
        "       fmov   @%0+,xd10\n" \
        "       fmov   @%0+,xd12\n" \
        "       fmov   @%0+,xd14\n" \
        "       fschg\n" \
        : "+r"(t) \
        ); \
    } while (0)

#define         ftrv() \
                __asm__ volatile("ftrv xmtrx,fv0" \
                : "+f"(fr0),"+f"(fr1),"+f"(fr2),"+f"(fr3));

#define         DEFREG        \
        register float fr0 __asm__("fr0"); \
        register float fr1 __asm__("fr1"); \
        register float fr2 __asm__("fr2"); \
        register float fr3 __asm__("fr3")

#else

/* generic C code for check */

static void ftrv_(const float xf[],float fv[])
{
        float f0,f1,f2,f3;
        f0 = fv[0];
        f1 = fv[1];
        f2 = fv[2];
        f3 = fv[3];
        fv[0] = xf[0]*f0 + xf[4]*f1 + xf[ 8]*f2 + xf[12]*f3;
        fv[1] = xf[1]*f0 + xf[5]*f1 + xf[ 9]*f2 + xf[13]*f3;
        fv[2] = xf[2]*f0 + xf[6]*f1 + xf[10]*f2 + xf[14]*f3;
        fv[3] = xf[3]*f0 + xf[7]*f1 + xf[11]*f2 + xf[15]*f3;
}

static void load_matrix_(float xf[],const float table[])
{
        int i;
        for(i=0;i<16;i++) xf[i]=table[i];
}

#define         ftrv()                  ftrv_(xf,fv)
#define         load_matrix(table)      load_matrix_(xf,table)

#define         DEFREG \
        float fv[4],xf[16]

#define         fr0     fv[0]
#define         fr1     fv[1]
#define         fr2     fv[2]
#define         fr3     fv[3]

#endif

#if 1
#define         DESCALE(x,n)    (x)*(1.0f/(1<<(n)))
#else
#define         DESCALE(x,n)    (((int)(x)+(1<<(n-1)))>>(n))
#endif

/* this code work worse on gcc cvs. 3.2.3 work fine */


#if 1
//optimized

void idct_sh4(DCTELEM *block)
{
        DEFREG;

        int i;
        float        tblock[8*8],*fblock;
        int ofs1,ofs2,ofs3;
        int fpscr;

        fp_single_enter(fpscr);

        /* row */

        /* even part */
        load_matrix(even_table);

        fblock = tblock+4;
        i = 8;
        do {
                fr0 = block[0];
                fr1 = block[2];
                fr2 = block[4];
                fr3 = block[6];
                block+=8;
                ftrv();
                *--fblock = fr3;
                *--fblock = fr2;
                *--fblock = fr1;
                *--fblock = fr0;
                fblock+=8+4;
        } while(--i);
        block-=8*8;
        fblock-=8*8+4;

        load_matrix(odd_table);

        i = 8;

        do {
                float t0,t1,t2,t3;
                fr0 = block[1];
                fr1 = block[3];
                fr2 = block[5];
                fr3 = block[7];
                block+=8;
                ftrv();
                t0 = *fblock++;
                t1 = *fblock++;
                t2 = *fblock++;
                t3 = *fblock++;
                fblock+=4;
                *--fblock = t0 - fr0;
                *--fblock = t1 - fr1;
                *--fblock = t2 - fr2;
                *--fblock = t3 - fr3;
                *--fblock = t3 + fr3;
                *--fblock = t2 + fr2;
                *--fblock = t1 + fr1;
                *--fblock = t0 + fr0;
                fblock+=8;
        } while(--i);
        block-=8*8;
        fblock-=8*8;

        /* col */

        /* even part */
        load_matrix(even_table);

        ofs1 = sizeof(float)*2*8;
        ofs2 = sizeof(float)*4*8;
        ofs3 = sizeof(float)*6*8;

        i = 8;

#define        OA(fblock,ofs)   *(float*)((char*)fblock + ofs)

        do {
                fr0 = OA(fblock,   0);
                fr1 = OA(fblock,ofs1);
                fr2 = OA(fblock,ofs2);
                fr3 = OA(fblock,ofs3);
                ftrv();
                OA(fblock,0   ) = fr0;
                OA(fblock,ofs1) = fr1;
                OA(fblock,ofs2) = fr2;
                OA(fblock,ofs3) = fr3;
                fblock++;
        } while(--i);
        fblock-=8;

        load_matrix(odd_table);

        i=8;
        do {
                float t0,t1,t2,t3;
                t0 = OA(fblock,   0); /* [8*0] */
                t1 = OA(fblock,ofs1); /* [8*2] */
                t2 = OA(fblock,ofs2); /* [8*4] */
                t3 = OA(fblock,ofs3); /* [8*6] */
                fblock+=8;
                fr0 = OA(fblock,   0); /* [8*1] */
                fr1 = OA(fblock,ofs1); /* [8*3] */
                fr2 = OA(fblock,ofs2); /* [8*5] */
                fr3 = OA(fblock,ofs3); /* [8*7] */
                fblock+=-8+1;
                ftrv();
                block[8*0] = DESCALE(t0 + fr0,3);
                block[8*7] = DESCALE(t0 - fr0,3);
                block[8*1] = DESCALE(t1 + fr1,3);
                block[8*6] = DESCALE(t1 - fr1,3);
                block[8*2] = DESCALE(t2 + fr2,3);
                block[8*5] = DESCALE(t2 - fr2,3);
                block[8*3] = DESCALE(t3 + fr3,3);
                block[8*4] = DESCALE(t3 - fr3,3);
                block++;
        } while(--i);

        fp_single_leave(fpscr);
}
#else
void idct_sh4(DCTELEM *block)
{
        DEFREG;

        int i;
        float   tblock[8*8],*fblock;

        /* row */

        /* even part */
        load_matrix(even_table);

        fblock = tblock;
        i = 8;
        do {
                fr0 = block[0];
                fr1 = block[2];
                fr2 = block[4];
                fr3 = block[6];
                block+=8;
                ftrv();
                fblock[0] = fr0;
                fblock[2] = fr1;
                fblock[4] = fr2;
                fblock[6] = fr3;
                fblock+=8;
        } while(--i);
        block-=8*8;
        fblock-=8*8;

        load_matrix(odd_table);

        i = 8;

        do {
                float t0,t1,t2,t3;
                fr0 = block[1];
                fr1 = block[3];
                fr2 = block[5];
                fr3 = block[7];
                block+=8;
                ftrv();
                t0 = fblock[0];
                t1 = fblock[2];
                t2 = fblock[4];
                t3 = fblock[6];
                fblock[0] = t0 + fr0;
                fblock[7] = t0 - fr0;
                fblock[1] = t1 + fr1;
                fblock[6] = t1 - fr1;
                fblock[2] = t2 + fr2;
                fblock[5] = t2 - fr2;
                fblock[3] = t3 + fr3;
                fblock[4] = t3 - fr3;
                fblock+=8;
        } while(--i);
        block-=8*8;
        fblock-=8*8;

        /* col */

        /* even part */
        load_matrix(even_table);

        i = 8;

        do {
                fr0 = fblock[8*0];
                fr1 = fblock[8*2];
                fr2 = fblock[8*4];
                fr3 = fblock[8*6];
                ftrv();
                fblock[8*0] = fr0;
                fblock[8*2] = fr1;
                fblock[8*4] = fr2;
                fblock[8*6] = fr3;
                fblock++;
        } while(--i);
        fblock-=8;

        load_matrix(odd_table);

        i=8;
        do {
                float t0,t1,t2,t3;
                fr0 = fblock[8*1];
                fr1 = fblock[8*3];
                fr2 = fblock[8*5];
                fr3 = fblock[8*7];
                ftrv();
                t0 = fblock[8*0];
                t1 = fblock[8*2];
                t2 = fblock[8*4];
                t3 = fblock[8*6];
                fblock++;
                block[8*0] = DESCALE(t0 + fr0,3);
                block[8*7] = DESCALE(t0 - fr0,3);
                block[8*1] = DESCALE(t1 + fr1,3);
                block[8*6] = DESCALE(t1 - fr1,3);
                block[8*2] = DESCALE(t2 + fr2,3);
                block[8*5] = DESCALE(t2 - fr2,3);
                block[8*3] = DESCALE(t3 + fr3,3);
                block[8*4] = DESCALE(t3 - fr3,3);
                block++;
        } while(--i);
}
#endif