qpel.c 58.6 KB
Newer Older
1
/*
2 3 4 5 6
 * This is optimized for sh, which have post increment addressing (*p++).
 * Some CPU may be index (p[n]) faster than post increment (*p++).
 *
 * copyright (c) 2001-2003 BERO <bero@geocities.co.jp>
 *
7
 * This file is part of Libav.
8
 *
9
 * Libav is free software; you can redistribute it and/or
10 11
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
12
 * version 2.1 of the License, or (at your option) any later version.
13
 *
14
 * Libav is distributed in the hope that it will be useful,
15 16 17 18 19
 * 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
20
 * License along with Libav; if not, write to the Free Software
21 22
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
 */
23

24 25
#include "libavutil/common.h"

26 27 28 29
#define PIXOP2(OPNAME, OP) \
\
static inline void OPNAME ## _pixels4_l2_aligned(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int dst_stride, int src_stride1, int src_stride2, int h) \
{\
30
        do {\
31
                OP(LP(dst  ),rnd_avg32(LPC(src1  ),LPC(src2  )) ); \
32 33 34 35
                src1+=src_stride1; \
                src2+=src_stride2; \
                dst+=dst_stride; \
        } while(--h); \
36 37 38 39
}\
\
static inline void OPNAME ## _pixels4_l2_aligned2(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int dst_stride, int src_stride1, int src_stride2, int h) \
{\
40
        do {\
41
                OP(LP(dst  ),rnd_avg32(AV_RN32(src1  ),LPC(src2  )) ); \
42 43 44 45
                src1+=src_stride1; \
                src2+=src_stride2; \
                dst+=dst_stride; \
        } while(--h); \
46 47 48 49
}\
\
static inline void OPNAME ## _no_rnd_pixels16_l2_aligned2(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int dst_stride, int src_stride1, int src_stride2, int h) \
{\
50
        do {\
51 52 53 54
                OP(LP(dst  ),no_rnd_avg32(AV_RN32(src1  ),LPC(src2  )) ); \
                OP(LP(dst+4),no_rnd_avg32(AV_RN32(src1+4),LPC(src2+4)) ); \
                OP(LP(dst+8),no_rnd_avg32(AV_RN32(src1+8),LPC(src2+8)) ); \
                OP(LP(dst+12),no_rnd_avg32(AV_RN32(src1+12),LPC(src2+12)) ); \
55 56 57 58
                src1+=src_stride1; \
                src2+=src_stride2; \
                dst+=dst_stride; \
        } while(--h); \
59 60 61 62
}\
\
static inline void OPNAME ## _pixels16_l2_aligned2(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int dst_stride, int src_stride1, int src_stride2, int h) \
{\
63
        do {\
64 65 66 67
                OP(LP(dst  ),rnd_avg32(AV_RN32(src1  ),LPC(src2  )) ); \
                OP(LP(dst+4),rnd_avg32(AV_RN32(src1+4),LPC(src2+4)) ); \
                OP(LP(dst+8),rnd_avg32(AV_RN32(src1+8),LPC(src2+8)) ); \
                OP(LP(dst+12),rnd_avg32(AV_RN32(src1+12),LPC(src2+12)) ); \
68 69 70 71
                src1+=src_stride1; \
                src2+=src_stride2; \
                dst+=dst_stride; \
        } while(--h); \
72 73 74 75
}\
\
static inline void OPNAME ## _no_rnd_pixels8_l2_aligned2(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int dst_stride, int src_stride1, int src_stride2, int h) \
{\
76
        do { /* onlye src2 aligned */\
77 78
                OP(LP(dst  ),no_rnd_avg32(AV_RN32(src1  ),LPC(src2  )) ); \
                OP(LP(dst+4),no_rnd_avg32(AV_RN32(src1+4),LPC(src2+4)) ); \
79 80 81 82
                src1+=src_stride1; \
                src2+=src_stride2; \
                dst+=dst_stride; \
        } while(--h); \
83 84 85 86
}\
\
static inline void OPNAME ## _pixels8_l2_aligned2(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int dst_stride, int src_stride1, int src_stride2, int h) \
{\
87
        do {\
88 89
                OP(LP(dst  ),rnd_avg32(AV_RN32(src1  ),LPC(src2  )) ); \
                OP(LP(dst+4),rnd_avg32(AV_RN32(src1+4),LPC(src2+4)) ); \
90 91 92 93
                src1+=src_stride1; \
                src2+=src_stride2; \
                dst+=dst_stride; \
        } while(--h); \
94 95 96 97
}\
\
static inline void OPNAME ## _no_rnd_pixels8_l2_aligned(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int dst_stride, int src_stride1, int src_stride2, int h) \
{\
98
        do {\
99 100
                OP(LP(dst  ),no_rnd_avg32(LPC(src1  ),LPC(src2  )) ); \
                OP(LP(dst+4),no_rnd_avg32(LPC(src1+4),LPC(src2+4)) ); \
101 102 103 104
                src1+=src_stride1; \
                src2+=src_stride2; \
                dst+=dst_stride; \
        } while(--h); \
105 106 107 108
}\
\
static inline void OPNAME ## _pixels8_l2_aligned(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int dst_stride, int src_stride1, int src_stride2, int h) \
{\
109
        do {\
110 111
                OP(LP(dst  ),rnd_avg32(LPC(src1  ),LPC(src2  )) ); \
                OP(LP(dst+4),rnd_avg32(LPC(src1+4),LPC(src2+4)) ); \
112 113 114 115
                src1+=src_stride1; \
                src2+=src_stride2; \
                dst+=dst_stride; \
        } while(--h); \
116 117 118 119
}\
\
static inline void OPNAME ## _no_rnd_pixels16_l2_aligned(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int dst_stride, int src_stride1, int src_stride2, int h) \
{\
120
        do {\
121 122 123 124
                OP(LP(dst  ),no_rnd_avg32(LPC(src1  ),LPC(src2  )) ); \
                OP(LP(dst+4),no_rnd_avg32(LPC(src1+4),LPC(src2+4)) ); \
                OP(LP(dst+8),no_rnd_avg32(LPC(src1+8),LPC(src2+8)) ); \
                OP(LP(dst+12),no_rnd_avg32(LPC(src1+12),LPC(src2+12)) ); \
125 126 127 128
                src1+=src_stride1; \
                src2+=src_stride2; \
                dst+=dst_stride; \
        } while(--h); \
129 130 131 132
}\
\
static inline void OPNAME ## _pixels16_l2_aligned(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int dst_stride, int src_stride1, int src_stride2, int h) \
{\
133
        do {\
134 135 136 137
                OP(LP(dst  ),rnd_avg32(LPC(src1  ),LPC(src2  )) ); \
                OP(LP(dst+4),rnd_avg32(LPC(src1+4),LPC(src2+4)) ); \
                OP(LP(dst+8),rnd_avg32(LPC(src1+8),LPC(src2+8)) ); \
                OP(LP(dst+12),rnd_avg32(LPC(src1+12),LPC(src2+12)) ); \
138 139 140 141
                src1+=src_stride1; \
                src2+=src_stride2; \
                dst+=dst_stride; \
        } while(--h); \
142 143 144 145 146 147 148 149 150 151 152 153 154 155 156
}\
\
static inline void OPNAME ## _no_rnd_pixels16_l2_aligned1(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int dst_stride, int src_stride1, int src_stride2, int h) \
{ OPNAME ## _no_rnd_pixels16_l2_aligned2(dst,src2,src1,dst_stride,src_stride2,src_stride1,h); } \
\
static inline void OPNAME ## _pixels16_l2_aligned1(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int dst_stride, int src_stride1, int src_stride2, int h) \
{ OPNAME ## _pixels16_l2_aligned2(dst,src2,src1,dst_stride,src_stride2,src_stride1,h); } \
\
static inline void OPNAME ## _no_rnd_pixels8_l2_aligned1(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int dst_stride, int src_stride1, int src_stride2, int h) \
{ OPNAME ## _no_rnd_pixels8_l2_aligned2(dst,src2,src1,dst_stride,src_stride2,src_stride1,h); } \
\
static inline void OPNAME ## _pixels8_l2_aligned1(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int dst_stride, int src_stride1, int src_stride2, int h) \
{ OPNAME ## _pixels8_l2_aligned2(dst,src2,src1,dst_stride,src_stride2,src_stride1,h); } \
\
static inline void OPNAME ## _pixels8_l4_aligned(uint8_t *dst, const uint8_t *src1, uint8_t *src2, uint8_t *src3, uint8_t *src4,int dst_stride, int src_stride1, int src_stride2,int src_stride3,int src_stride4, int h){\
157 158
        do { \
                uint32_t a0,a1,a2,a3; \
159 160
                UNPACK(a0,a1,LPC(src1),LPC(src2)); \
                UNPACK(a2,a3,LPC(src3),LPC(src4)); \
161
                OP(LP(dst),rnd_PACK(a0,a1,a2,a3)); \
162 163
                UNPACK(a0,a1,LPC(src1+4),LPC(src2+4)); \
                UNPACK(a2,a3,LPC(src3+4),LPC(src4+4)); \
164 165 166 167 168 169 170
                OP(LP(dst+4),rnd_PACK(a0,a1,a2,a3)); \
                src1+=src_stride1;\
                src2+=src_stride2;\
                src3+=src_stride3;\
                src4+=src_stride4;\
                dst+=dst_stride;\
        } while(--h); \
171 172 173
} \
\
static inline void OPNAME ## _no_rnd_pixels8_l4_aligned(uint8_t *dst, const uint8_t *src1, uint8_t *src2, uint8_t *src3, uint8_t *src4,int dst_stride, int src_stride1, int src_stride2,int src_stride3,int src_stride4, int h){\
174 175
        do { \
                uint32_t a0,a1,a2,a3; \
176 177
                UNPACK(a0,a1,LPC(src1),LPC(src2)); \
                UNPACK(a2,a3,LPC(src3),LPC(src4)); \
178
                OP(LP(dst),no_rnd_PACK(a0,a1,a2,a3)); \
179 180
                UNPACK(a0,a1,LPC(src1+4),LPC(src2+4)); \
                UNPACK(a2,a3,LPC(src3+4),LPC(src4+4)); \
181 182 183 184 185 186 187
                OP(LP(dst+4),no_rnd_PACK(a0,a1,a2,a3)); \
                src1+=src_stride1;\
                src2+=src_stride2;\
                src3+=src_stride3;\
                src4+=src_stride4;\
                dst+=dst_stride;\
        } while(--h); \
188 189 190
} \
\
static inline void OPNAME ## _pixels8_l4_aligned0(uint8_t *dst, const uint8_t *src1, uint8_t *src2, uint8_t *src3, uint8_t *src4,int dst_stride, int src_stride1, int src_stride2,int src_stride3,int src_stride4, int h){\
191 192
        do { \
                uint32_t a0,a1,a2,a3; /* src1 only not aligned */\
193 194
                UNPACK(a0,a1,AV_RN32(src1),LPC(src2)); \
                UNPACK(a2,a3,LPC(src3),LPC(src4)); \
195
                OP(LP(dst),rnd_PACK(a0,a1,a2,a3)); \
196 197
                UNPACK(a0,a1,AV_RN32(src1+4),LPC(src2+4)); \
                UNPACK(a2,a3,LPC(src3+4),LPC(src4+4)); \
198 199 200 201 202 203 204
                OP(LP(dst+4),rnd_PACK(a0,a1,a2,a3)); \
                src1+=src_stride1;\
                src2+=src_stride2;\
                src3+=src_stride3;\
                src4+=src_stride4;\
                dst+=dst_stride;\
        } while(--h); \
205 206 207
} \
\
static inline void OPNAME ## _no_rnd_pixels8_l4_aligned0(uint8_t *dst, const uint8_t *src1, uint8_t *src2, uint8_t *src3, uint8_t *src4,int dst_stride, int src_stride1, int src_stride2,int src_stride3,int src_stride4, int h){\
208 209
        do { \
                uint32_t a0,a1,a2,a3; \
210 211
                UNPACK(a0,a1,AV_RN32(src1),LPC(src2)); \
                UNPACK(a2,a3,LPC(src3),LPC(src4)); \
212
                OP(LP(dst),no_rnd_PACK(a0,a1,a2,a3)); \
213 214
                UNPACK(a0,a1,AV_RN32(src1+4),LPC(src2+4)); \
                UNPACK(a2,a3,LPC(src3+4),LPC(src4+4)); \
215 216 217 218 219 220 221
                OP(LP(dst+4),no_rnd_PACK(a0,a1,a2,a3)); \
                src1+=src_stride1;\
                src2+=src_stride2;\
                src3+=src_stride3;\
                src4+=src_stride4;\
                dst+=dst_stride;\
        } while(--h); \
222 223 224
} \
\
static inline void OPNAME ## _pixels16_l4_aligned(uint8_t *dst, const uint8_t *src1, uint8_t *src2, uint8_t *src3, uint8_t *src4,int dst_stride, int src_stride1, int src_stride2,int src_stride3,int src_stride4, int h){\
225 226
        do { \
                uint32_t a0,a1,a2,a3; \
227 228
                UNPACK(a0,a1,LPC(src1),LPC(src2)); \
                UNPACK(a2,a3,LPC(src3),LPC(src4)); \
229
                OP(LP(dst),rnd_PACK(a0,a1,a2,a3)); \
230 231
                UNPACK(a0,a1,LPC(src1+4),LPC(src2+4)); \
                UNPACK(a2,a3,LPC(src3+4),LPC(src4+4)); \
232
                OP(LP(dst+8),rnd_PACK(a0,a1,a2,a3)); \
233 234
                UNPACK(a0,a1,LPC(src1+8),LPC(src2+8)); \
                UNPACK(a2,a3,LPC(src3+8),LPC(src4+8)); \
235
                OP(LP(dst+8),rnd_PACK(a0,a1,a2,a3)); \
236 237
                UNPACK(a0,a1,LPC(src1+12),LPC(src2+12)); \
                UNPACK(a2,a3,LPC(src3+12),LPC(src4+12)); \
238 239 240 241 242 243 244
                OP(LP(dst+12),rnd_PACK(a0,a1,a2,a3)); \
                src1+=src_stride1;\
                src2+=src_stride2;\
                src3+=src_stride3;\
                src4+=src_stride4;\
                dst+=dst_stride;\
        } while(--h); \
245 246 247
} \
\
static inline void OPNAME ## _no_rnd_pixels16_l4_aligned(uint8_t *dst, const uint8_t *src1, uint8_t *src2, uint8_t *src3, uint8_t *src4,int dst_stride, int src_stride1, int src_stride2,int src_stride3,int src_stride4, int h){\
248 249
        do { \
                uint32_t a0,a1,a2,a3; \
250 251
                UNPACK(a0,a1,LPC(src1),LPC(src2)); \
                UNPACK(a2,a3,LPC(src3),LPC(src4)); \
252
                OP(LP(dst),no_rnd_PACK(a0,a1,a2,a3)); \
253 254
                UNPACK(a0,a1,LPC(src1+4),LPC(src2+4)); \
                UNPACK(a2,a3,LPC(src3+4),LPC(src4+4)); \
255
                OP(LP(dst+4),no_rnd_PACK(a0,a1,a2,a3)); \
256 257
                UNPACK(a0,a1,LPC(src1+8),LPC(src2+8)); \
                UNPACK(a2,a3,LPC(src3+8),LPC(src4+8)); \
258
                OP(LP(dst+8),no_rnd_PACK(a0,a1,a2,a3)); \
259 260
                UNPACK(a0,a1,LPC(src1+12),LPC(src2+12)); \
                UNPACK(a2,a3,LPC(src3+12),LPC(src4+12)); \
261 262 263 264 265 266 267
                OP(LP(dst+12),no_rnd_PACK(a0,a1,a2,a3)); \
                src1+=src_stride1;\
                src2+=src_stride2;\
                src3+=src_stride3;\
                src4+=src_stride4;\
                dst+=dst_stride;\
        } while(--h); \
268 269 270
} \
\
static inline void OPNAME ## _pixels16_l4_aligned0(uint8_t *dst, const uint8_t *src1, uint8_t *src2, uint8_t *src3, uint8_t *src4,int dst_stride, int src_stride1, int src_stride2,int src_stride3,int src_stride4, int h){\
271 272
        do { /* src1 is unaligned */\
                uint32_t a0,a1,a2,a3; \
273 274
                UNPACK(a0,a1,AV_RN32(src1),LPC(src2)); \
                UNPACK(a2,a3,LPC(src3),LPC(src4)); \
275
                OP(LP(dst),rnd_PACK(a0,a1,a2,a3)); \
276 277
                UNPACK(a0,a1,AV_RN32(src1+4),LPC(src2+4)); \
                UNPACK(a2,a3,LPC(src3+4),LPC(src4+4)); \
278
                OP(LP(dst+8),rnd_PACK(a0,a1,a2,a3)); \
279 280
                UNPACK(a0,a1,AV_RN32(src1+8),LPC(src2+8)); \
                UNPACK(a2,a3,LPC(src3+8),LPC(src4+8)); \
281
                OP(LP(dst+8),rnd_PACK(a0,a1,a2,a3)); \
282 283
                UNPACK(a0,a1,AV_RN32(src1+12),LPC(src2+12)); \
                UNPACK(a2,a3,LPC(src3+12),LPC(src4+12)); \
284 285 286 287 288 289 290
                OP(LP(dst+12),rnd_PACK(a0,a1,a2,a3)); \
                src1+=src_stride1;\
                src2+=src_stride2;\
                src3+=src_stride3;\
                src4+=src_stride4;\
                dst+=dst_stride;\
        } while(--h); \
291 292 293
} \
\
static inline void OPNAME ## _no_rnd_pixels16_l4_aligned0(uint8_t *dst, const uint8_t *src1, uint8_t *src2, uint8_t *src3, uint8_t *src4,int dst_stride, int src_stride1, int src_stride2,int src_stride3,int src_stride4, int h){\
294 295
        do { \
                uint32_t a0,a1,a2,a3; \
296 297
                UNPACK(a0,a1,AV_RN32(src1),LPC(src2)); \
                UNPACK(a2,a3,LPC(src3),LPC(src4)); \
298
                OP(LP(dst),no_rnd_PACK(a0,a1,a2,a3)); \
299 300
                UNPACK(a0,a1,AV_RN32(src1+4),LPC(src2+4)); \
                UNPACK(a2,a3,LPC(src3+4),LPC(src4+4)); \
301
                OP(LP(dst+4),no_rnd_PACK(a0,a1,a2,a3)); \
302 303
                UNPACK(a0,a1,AV_RN32(src1+8),LPC(src2+8)); \
                UNPACK(a2,a3,LPC(src3+8),LPC(src4+8)); \
304
                OP(LP(dst+8),no_rnd_PACK(a0,a1,a2,a3)); \
305 306
                UNPACK(a0,a1,AV_RN32(src1+12),LPC(src2+12)); \
                UNPACK(a2,a3,LPC(src3+12),LPC(src4+12)); \
307 308 309 310 311 312 313
                OP(LP(dst+12),no_rnd_PACK(a0,a1,a2,a3)); \
                src1+=src_stride1;\
                src2+=src_stride2;\
                src3+=src_stride3;\
                src4+=src_stride4;\
                dst+=dst_stride;\
        } while(--h); \
314 315 316
} \
\

Michael Niedermayer's avatar
Michael Niedermayer committed
317
#define op_avg(a, b) a = rnd_avg32(a,b)
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
#define op_put(a, b) a = b

PIXOP2(avg, op_avg)
PIXOP2(put, op_put)
#undef op_avg
#undef op_put

#define avg2(a,b) ((a+b+1)>>1)
#define avg4(a,b,c,d) ((a+b+c+d+2)>>2)


static void gmc1_c(uint8_t *dst, uint8_t *src, int stride, int h, int x16, int y16, int rounder)
{
    const int A=(16-x16)*(16-y16);
    const int B=(   x16)*(16-y16);
    const int C=(16-x16)*(   y16);
    const int D=(   x16)*(   y16);

    do {
        int t0,t1,t2,t3;
        uint8_t *s0 = src;
        uint8_t *s1 = src+stride;
        t0 = *s0++; t2 = *s1++;
        t1 = *s0++; t3 = *s1++;
        dst[0]= (A*t0 + B*t1 + C*t2 + D*t3 + rounder)>>8;
        t0 = *s0++; t2 = *s1++;
        dst[1]= (A*t1 + B*t0 + C*t3 + D*t2 + rounder)>>8;
        t1 = *s0++; t3 = *s1++;
        dst[2]= (A*t0 + B*t1 + C*t2 + D*t3 + rounder)>>8;
        t0 = *s0++; t2 = *s1++;
        dst[3]= (A*t1 + B*t0 + C*t3 + D*t2 + rounder)>>8;
        t1 = *s0++; t3 = *s1++;
        dst[4]= (A*t0 + B*t1 + C*t2 + D*t3 + rounder)>>8;
        t0 = *s0++; t2 = *s1++;
        dst[5]= (A*t1 + B*t0 + C*t3 + D*t2 + rounder)>>8;
        t1 = *s0++; t3 = *s1++;
        dst[6]= (A*t0 + B*t1 + C*t2 + D*t3 + rounder)>>8;
        t0 = *s0++; t2 = *s1++;
        dst[7]= (A*t1 + B*t0 + C*t3 + D*t2 + rounder)>>8;
        dst+= stride;
        src+= stride;
    }while(--h);
}

#define H264_CHROMA_MC(OPNAME, OP)\
363
static void OPNAME ## h264_chroma_mc2_sh4(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int stride, int h, int x, int y){\
364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384
    const int A=(8-x)*(8-y);\
    const int B=(  x)*(8-y);\
    const int C=(8-x)*(  y);\
    const int D=(  x)*(  y);\
    \
    assert(x<8 && y<8 && x>=0 && y>=0);\
\
    do {\
        int t0,t1,t2,t3; \
        uint8_t *s0 = src; \
        uint8_t *s1 = src+stride; \
        t0 = *s0++; t2 = *s1++; \
        t1 = *s0++; t3 = *s1++; \
        OP(dst[0], (A*t0 + B*t1 + C*t2 + D*t3));\
        t0 = *s0++; t2 = *s1++; \
        OP(dst[1], (A*t1 + B*t0 + C*t3 + D*t2));\
        dst+= stride;\
        src+= stride;\
    }while(--h);\
}\
\
385
static void OPNAME ## h264_chroma_mc4_sh4(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int stride, int h, int x, int y){\
386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410
    const int A=(8-x)*(8-y);\
    const int B=(  x)*(8-y);\
    const int C=(8-x)*(  y);\
    const int D=(  x)*(  y);\
    \
    assert(x<8 && y<8 && x>=0 && y>=0);\
\
    do {\
        int t0,t1,t2,t3; \
        uint8_t *s0 = src; \
        uint8_t *s1 = src+stride; \
        t0 = *s0++; t2 = *s1++; \
        t1 = *s0++; t3 = *s1++; \
        OP(dst[0], (A*t0 + B*t1 + C*t2 + D*t3));\
        t0 = *s0++; t2 = *s1++; \
        OP(dst[1], (A*t1 + B*t0 + C*t3 + D*t2));\
        t1 = *s0++; t3 = *s1++; \
        OP(dst[2], (A*t0 + B*t1 + C*t2 + D*t3));\
        t0 = *s0++; t2 = *s1++; \
        OP(dst[3], (A*t1 + B*t0 + C*t3 + D*t2));\
        dst+= stride;\
        src+= stride;\
    }while(--h);\
}\
\
411
static void OPNAME ## h264_chroma_mc8_sh4(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int stride, int h, int x, int y){\
412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454
    const int A=(8-x)*(8-y);\
    const int B=(  x)*(8-y);\
    const int C=(8-x)*(  y);\
    const int D=(  x)*(  y);\
    \
    assert(x<8 && y<8 && x>=0 && y>=0);\
\
    do {\
        int t0,t1,t2,t3; \
        uint8_t *s0 = src; \
        uint8_t *s1 = src+stride; \
        t0 = *s0++; t2 = *s1++; \
        t1 = *s0++; t3 = *s1++; \
        OP(dst[0], (A*t0 + B*t1 + C*t2 + D*t3));\
        t0 = *s0++; t2 = *s1++; \
        OP(dst[1], (A*t1 + B*t0 + C*t3 + D*t2));\
        t1 = *s0++; t3 = *s1++; \
        OP(dst[2], (A*t0 + B*t1 + C*t2 + D*t3));\
        t0 = *s0++; t2 = *s1++; \
        OP(dst[3], (A*t1 + B*t0 + C*t3 + D*t2));\
        t1 = *s0++; t3 = *s1++; \
        OP(dst[4], (A*t0 + B*t1 + C*t2 + D*t3));\
        t0 = *s0++; t2 = *s1++; \
        OP(dst[5], (A*t1 + B*t0 + C*t3 + D*t2));\
        t1 = *s0++; t3 = *s1++; \
        OP(dst[6], (A*t0 + B*t1 + C*t2 + D*t3));\
        t0 = *s0++; t2 = *s1++; \
        OP(dst[7], (A*t1 + B*t0 + C*t3 + D*t2));\
        dst+= stride;\
        src+= stride;\
    }while(--h);\
}

#define op_avg(a, b) a = (((a)+(((b) + 32)>>6)+1)>>1)
#define op_put(a, b) a = (((b) + 32)>>6)

H264_CHROMA_MC(put_       , op_put)
H264_CHROMA_MC(avg_       , op_avg)
#undef op_avg
#undef op_put

#define QPEL_MC(r, OPNAME, RND, OP) \
static void OPNAME ## mpeg4_qpel8_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h){\
455
    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481
    do {\
        uint8_t *s = src; \
        int src0,src1,src2,src3,src4,src5,src6,src7,src8;\
        src0= *s++;\
        src1= *s++;\
        src2= *s++;\
        src3= *s++;\
        src4= *s++;\
        OP(dst[0], (src0+src1)*20 - (src0+src2)*6 + (src1+src3)*3 - (src2+src4));\
        src5= *s++;\
        OP(dst[1], (src1+src2)*20 - (src0+src3)*6 + (src0+src4)*3 - (src1+src5));\
        src6= *s++;\
        OP(dst[2], (src2+src3)*20 - (src1+src4)*6 + (src0+src5)*3 - (src0+src6));\
        src7= *s++;\
        OP(dst[3], (src3+src4)*20 - (src2+src5)*6 + (src1+src6)*3 - (src0+src7));\
        src8= *s++;\
        OP(dst[4], (src4+src5)*20 - (src3+src6)*6 + (src2+src7)*3 - (src1+src8));\
        OP(dst[5], (src5+src6)*20 - (src4+src7)*6 + (src3+src8)*3 - (src2+src8));\
        OP(dst[6], (src6+src7)*20 - (src5+src8)*6 + (src4+src8)*3 - (src3+src7));\
        OP(dst[7], (src7+src8)*20 - (src6+src8)*6 + (src5+src7)*3 - (src4+src6));\
        dst+=dstStride;\
        src+=srcStride;\
    }while(--h);\
}\
\
static void OPNAME ## mpeg4_qpel8_v_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
482
    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509
    int w=8;\
    do{\
        uint8_t *s = src, *d=dst;\
        int src0,src1,src2,src3,src4,src5,src6,src7,src8;\
        src0 = *s; s+=srcStride; \
        src1 = *s; s+=srcStride; \
        src2 = *s; s+=srcStride; \
        src3 = *s; s+=srcStride; \
        src4 = *s; s+=srcStride; \
        OP(*d, (src0+src1)*20 - (src0+src2)*6 + (src1+src3)*3 - (src2+src4));d+=dstStride;\
        src5 = *s; s+=srcStride; \
        OP(*d, (src1+src2)*20 - (src0+src3)*6 + (src0+src4)*3 - (src1+src5));d+=dstStride;\
        src6 = *s; s+=srcStride; \
        OP(*d, (src2+src3)*20 - (src1+src4)*6 + (src0+src5)*3 - (src0+src6));d+=dstStride;\
        src7 = *s; s+=srcStride; \
        OP(*d, (src3+src4)*20 - (src2+src5)*6 + (src1+src6)*3 - (src0+src7));d+=dstStride;\
        src8 = *s; \
        OP(*d, (src4+src5)*20 - (src3+src6)*6 + (src2+src7)*3 - (src1+src8));d+=dstStride;\
        OP(*d, (src5+src6)*20 - (src4+src7)*6 + (src3+src8)*3 - (src2+src8));d+=dstStride;\
        OP(*d, (src6+src7)*20 - (src5+src8)*6 + (src4+src8)*3 - (src3+src7));d+=dstStride;\
        OP(*d, (src7+src8)*20 - (src6+src8)*6 + (src5+src7)*3 - (src4+src6));\
        dst++;\
        src++;\
    }while(--w);\
}\
\
static void OPNAME ## mpeg4_qpel16_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h){\
510
    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553
    do {\
        uint8_t *s = src;\
        int src0,src1,src2,src3,src4,src5,src6,src7,src8;\
        int src9,src10,src11,src12,src13,src14,src15,src16;\
        src0= *s++;\
        src1= *s++;\
        src2= *s++;\
        src3= *s++;\
        src4= *s++;\
        OP(dst[ 0], (src0 +src1 )*20 - (src0 +src2 )*6 + (src1 +src3 )*3 - (src2 +src4 ));\
        src5= *s++;\
        OP(dst[ 1], (src1 +src2 )*20 - (src0 +src3 )*6 + (src0 +src4 )*3 - (src1 +src5 ));\
        src6= *s++;\
        OP(dst[ 2], (src2 +src3 )*20 - (src1 +src4 )*6 + (src0 +src5 )*3 - (src0 +src6 ));\
        src7= *s++;\
        OP(dst[ 3], (src3 +src4 )*20 - (src2 +src5 )*6 + (src1 +src6 )*3 - (src0 +src7 ));\
        src8= *s++;\
        OP(dst[ 4], (src4 +src5 )*20 - (src3 +src6 )*6 + (src2 +src7 )*3 - (src1 +src8 ));\
        src9= *s++;\
        OP(dst[ 5], (src5 +src6 )*20 - (src4 +src7 )*6 + (src3 +src8 )*3 - (src2 +src9 ));\
        src10= *s++;\
        OP(dst[ 6], (src6 +src7 )*20 - (src5 +src8 )*6 + (src4 +src9 )*3 - (src3 +src10));\
        src11= *s++;\
        OP(dst[ 7], (src7 +src8 )*20 - (src6 +src9 )*6 + (src5 +src10)*3 - (src4 +src11));\
        src12= *s++;\
        OP(dst[ 8], (src8 +src9 )*20 - (src7 +src10)*6 + (src6 +src11)*3 - (src5 +src12));\
        src13= *s++;\
        OP(dst[ 9], (src9 +src10)*20 - (src8 +src11)*6 + (src7 +src12)*3 - (src6 +src13));\
        src14= *s++;\
        OP(dst[10], (src10+src11)*20 - (src9 +src12)*6 + (src8 +src13)*3 - (src7 +src14));\
        src15= *s++;\
        OP(dst[11], (src11+src12)*20 - (src10+src13)*6 + (src9 +src14)*3 - (src8 +src15));\
        src16= *s++;\
        OP(dst[12], (src12+src13)*20 - (src11+src14)*6 + (src10+src15)*3 - (src9 +src16));\
        OP(dst[13], (src13+src14)*20 - (src12+src15)*6 + (src11+src16)*3 - (src10+src16));\
        OP(dst[14], (src14+src15)*20 - (src13+src16)*6 + (src12+src16)*3 - (src11+src15));\
        OP(dst[15], (src15+src16)*20 - (src14+src16)*6 + (src13+src15)*3 - (src12+src14));\
        dst+=dstStride;\
        src+=srcStride;\
    }while(--h);\
}\
\
static void OPNAME ## mpeg4_qpel16_v_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
554
    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597
    int w=16;\
    do {\
        uint8_t *s = src, *d=dst;\
        int src0,src1,src2,src3,src4,src5,src6,src7,src8;\
        int src9,src10,src11,src12,src13,src14,src15,src16;\
        src0 = *s; s+=srcStride; \
        src1 = *s; s+=srcStride; \
        src2 = *s; s+=srcStride; \
        src3 = *s; s+=srcStride; \
        src4 = *s; s+=srcStride; \
        OP(*d, (src0 +src1 )*20 - (src0 +src2 )*6 + (src1 +src3 )*3 - (src2 +src4 ));d+=dstStride;\
        src5 = *s; s+=srcStride; \
        OP(*d, (src1 +src2 )*20 - (src0 +src3 )*6 + (src0 +src4 )*3 - (src1 +src5 ));d+=dstStride;\
        src6 = *s; s+=srcStride; \
        OP(*d, (src2 +src3 )*20 - (src1 +src4 )*6 + (src0 +src5 )*3 - (src0 +src6 ));d+=dstStride;\
        src7 = *s; s+=srcStride; \
        OP(*d, (src3 +src4 )*20 - (src2 +src5 )*6 + (src1 +src6 )*3 - (src0 +src7 ));d+=dstStride;\
        src8 = *s; s+=srcStride; \
        OP(*d, (src4 +src5 )*20 - (src3 +src6 )*6 + (src2 +src7 )*3 - (src1 +src8 ));d+=dstStride;\
        src9 = *s; s+=srcStride; \
        OP(*d, (src5 +src6 )*20 - (src4 +src7 )*6 + (src3 +src8 )*3 - (src2 +src9 ));d+=dstStride;\
        src10 = *s; s+=srcStride; \
        OP(*d, (src6 +src7 )*20 - (src5 +src8 )*6 + (src4 +src9 )*3 - (src3 +src10));d+=dstStride;\
        src11 = *s; s+=srcStride; \
        OP(*d, (src7 +src8 )*20 - (src6 +src9 )*6 + (src5 +src10)*3 - (src4 +src11));d+=dstStride;\
        src12 = *s; s+=srcStride; \
        OP(*d, (src8 +src9 )*20 - (src7 +src10)*6 + (src6 +src11)*3 - (src5 +src12));d+=dstStride;\
        src13 = *s; s+=srcStride; \
        OP(*d, (src9 +src10)*20 - (src8 +src11)*6 + (src7 +src12)*3 - (src6 +src13));d+=dstStride;\
        src14 = *s; s+=srcStride; \
        OP(*d, (src10+src11)*20 - (src9 +src12)*6 + (src8 +src13)*3 - (src7 +src14));d+=dstStride;\
        src15 = *s; s+=srcStride; \
        OP(*d, (src11+src12)*20 - (src10+src13)*6 + (src9 +src14)*3 - (src8 +src15));d+=dstStride;\
        src16 = *s; \
        OP(*d, (src12+src13)*20 - (src11+src14)*6 + (src10+src15)*3 - (src9 +src16));d+=dstStride;\
        OP(*d, (src13+src14)*20 - (src12+src15)*6 + (src11+src16)*3 - (src10+src16));d+=dstStride;\
        OP(*d, (src14+src15)*20 - (src13+src16)*6 + (src12+src16)*3 - (src11+src15));d+=dstStride;\
        OP(*d, (src15+src16)*20 - (src14+src16)*6 + (src13+src15)*3 - (src12+src14));\
        dst++;\
        src++;\
    }while(--w);\
}\
\
598
static void OPNAME ## qpel8_mc00_sh4 (uint8_t *dst, uint8_t *src, int stride){\
599 600 601
    OPNAME ## pixels8_c(dst, src, stride, 8);\
}\
\
602
static void OPNAME ## qpel8_mc10_sh4(uint8_t *dst, uint8_t *src, int stride){\
603 604 605 606 607
    uint8_t half[64];\
    put ## RND ## mpeg4_qpel8_h_lowpass(half, src, 8, stride, 8);\
    OPNAME ## pixels8_l2_aligned2(dst, src, half, stride, stride, 8, 8);\
}\
\
608
static void OPNAME ## qpel8_mc20_sh4(uint8_t *dst, uint8_t *src, int stride){\
609 610 611
    OPNAME ## mpeg4_qpel8_h_lowpass(dst, src, stride, stride, 8);\
}\
\
612
static void OPNAME ## qpel8_mc30_sh4(uint8_t *dst, uint8_t *src, int stride){\
613 614 615 616 617
    uint8_t half[64];\
    put ## RND ## mpeg4_qpel8_h_lowpass(half, src, 8, stride, 8);\
    OPNAME ## pixels8_l2_aligned2(dst, src+1, half, stride, stride, 8, 8);\
}\
\
618
static void OPNAME ## qpel8_mc01_sh4(uint8_t *dst, uint8_t *src, int stride){\
619 620 621 622 623 624 625
    uint8_t full[16*9];\
    uint8_t half[64];\
    copy_block9(full, src, 16, stride, 9);\
    put ## RND ## mpeg4_qpel8_v_lowpass(half, full, 8, 16);\
    OPNAME ## pixels8_l2_aligned(dst, full, half, stride, 16, 8, 8);\
}\
\
626
static void OPNAME ## qpel8_mc02_sh4(uint8_t *dst, uint8_t *src, int stride){\
627 628 629 630 631
    uint8_t full[16*9];\
    copy_block9(full, src, 16, stride, 9);\
    OPNAME ## mpeg4_qpel8_v_lowpass(dst, full, stride, 16);\
}\
\
632
static void OPNAME ## qpel8_mc03_sh4(uint8_t *dst, uint8_t *src, int stride){\
633 634 635 636 637 638
    uint8_t full[16*9];\
    uint8_t half[64];\
    copy_block9(full, src, 16, stride, 9);\
    put ## RND ## mpeg4_qpel8_v_lowpass(half, full, 8, 16);\
    OPNAME ## pixels8_l2_aligned(dst, full+16, half, stride, 16, 8, 8);\
}\
639
static void OPNAME ## qpel8_mc11_sh4(uint8_t *dst, uint8_t *src, int stride){\
640 641 642 643 644 645 646 647 648
    uint8_t full[16*9];\
    uint8_t halfH[72];\
    uint8_t halfHV[64];\
    copy_block9(full, src, 16, stride, 9);\
    put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9);\
    put ## RND ## pixels8_l2_aligned(halfH, halfH, full, 8, 8, 16, 9);\
    put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8);\
    OPNAME ## pixels8_l2_aligned(dst, halfH, halfHV, stride, 8, 8, 8);\
}\
649
static void OPNAME ## qpel8_mc31_sh4(uint8_t *dst, uint8_t *src, int stride){\
650 651 652 653 654 655 656 657 658
    uint8_t full[16*9];\
    uint8_t halfH[72];\
    uint8_t halfHV[64];\
    copy_block9(full, src, 16, stride, 9);\
    put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9);\
    put ## RND ## pixels8_l2_aligned1(halfH, halfH, full+1, 8, 8, 16, 9);\
    put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8);\
    OPNAME ## pixels8_l2_aligned(dst, halfH, halfHV, stride, 8, 8, 8);\
}\
659
static void OPNAME ## qpel8_mc13_sh4(uint8_t *dst, uint8_t *src, int stride){\
660 661 662 663 664 665 666 667 668
    uint8_t full[16*9];\
    uint8_t halfH[72];\
    uint8_t halfHV[64];\
    copy_block9(full, src, 16, stride, 9);\
    put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9);\
    put ## RND ## pixels8_l2_aligned(halfH, halfH, full, 8, 8, 16, 9);\
    put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8);\
    OPNAME ## pixels8_l2_aligned(dst, halfH+8, halfHV, stride, 8, 8, 8);\
}\
669
static void OPNAME ## qpel8_mc33_sh4(uint8_t *dst, uint8_t *src, int stride){\
670 671 672 673 674 675 676 677 678
    uint8_t full[16*9];\
    uint8_t halfH[72];\
    uint8_t halfHV[64];\
    copy_block9(full, src, 16, stride, 9);\
    put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9);\
    put ## RND ## pixels8_l2_aligned1(halfH, halfH, full+1, 8, 8, 16, 9);\
    put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8);\
    OPNAME ## pixels8_l2_aligned(dst, halfH+8, halfHV, stride, 8, 8, 8);\
}\
679
static void OPNAME ## qpel8_mc21_sh4(uint8_t *dst, uint8_t *src, int stride){\
680 681 682 683 684 685
    uint8_t halfH[72];\
    uint8_t halfHV[64];\
    put ## RND ## mpeg4_qpel8_h_lowpass(halfH, src, 8, stride, 9);\
    put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8);\
    OPNAME ## pixels8_l2_aligned(dst, halfH, halfHV, stride, 8, 8, 8);\
}\
686
static void OPNAME ## qpel8_mc23_sh4(uint8_t *dst, uint8_t *src, int stride){\
687 688 689 690 691 692
    uint8_t halfH[72];\
    uint8_t halfHV[64];\
    put ## RND ## mpeg4_qpel8_h_lowpass(halfH, src, 8, stride, 9);\
    put ## RND ## mpeg4_qpel8_v_lowpass(halfHV, halfH, 8, 8);\
    OPNAME ## pixels8_l2_aligned(dst, halfH+8, halfHV, stride, 8, 8, 8);\
}\
693
static void OPNAME ## qpel8_mc12_sh4(uint8_t *dst, uint8_t *src, int stride){\
694 695 696 697 698 699 700
    uint8_t full[16*9];\
    uint8_t halfH[72];\
    copy_block9(full, src, 16, stride, 9);\
    put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9);\
    put ## RND ## pixels8_l2_aligned(halfH, halfH, full, 8, 8, 16, 9);\
    OPNAME ## mpeg4_qpel8_v_lowpass(dst, halfH, stride, 8);\
}\
701
static void OPNAME ## qpel8_mc32_sh4(uint8_t *dst, uint8_t *src, int stride){\
702 703 704 705 706 707 708
    uint8_t full[16*9];\
    uint8_t halfH[72];\
    copy_block9(full, src, 16, stride, 9);\
    put ## RND ## mpeg4_qpel8_h_lowpass(halfH, full, 8, 16, 9);\
    put ## RND ## pixels8_l2_aligned1(halfH, halfH, full+1, 8, 8, 16, 9);\
    OPNAME ## mpeg4_qpel8_v_lowpass(dst, halfH, stride, 8);\
}\
709
static void OPNAME ## qpel8_mc22_sh4(uint8_t *dst, uint8_t *src, int stride){\
710 711 712 713
    uint8_t halfH[72];\
    put ## RND ## mpeg4_qpel8_h_lowpass(halfH, src, 8, stride, 9);\
    OPNAME ## mpeg4_qpel8_v_lowpass(dst, halfH, stride, 8);\
}\
714
static void OPNAME ## qpel16_mc00_sh4 (uint8_t *dst, uint8_t *src, int stride){\
715 716 717
    OPNAME ## pixels16_c(dst, src, stride, 16);\
}\
\
718
static void OPNAME ## qpel16_mc10_sh4(uint8_t *dst, uint8_t *src, int stride){\
719 720 721 722 723
    uint8_t half[256];\
    put ## RND ## mpeg4_qpel16_h_lowpass(half, src, 16, stride, 16);\
    OPNAME ## pixels16_l2_aligned2(dst, src, half, stride, stride, 16, 16);\
}\
\
724
static void OPNAME ## qpel16_mc20_sh4(uint8_t *dst, uint8_t *src, int stride){\
725 726 727
    OPNAME ## mpeg4_qpel16_h_lowpass(dst, src, stride, stride, 16);\
}\
\
728
static void OPNAME ## qpel16_mc30_sh4(uint8_t *dst, uint8_t *src, int stride){\
729 730 731 732 733
    uint8_t half[256];\
    put ## RND ## mpeg4_qpel16_h_lowpass(half, src, 16, stride, 16);\
    OPNAME ## pixels16_l2_aligned2(dst, src+1, half, stride, stride, 16, 16);\
}\
\
734
static void OPNAME ## qpel16_mc01_sh4(uint8_t *dst, uint8_t *src, int stride){\
735 736 737 738 739 740 741
    uint8_t full[24*17];\
    uint8_t half[256];\
    copy_block17(full, src, 24, stride, 17);\
    put ## RND ## mpeg4_qpel16_v_lowpass(half, full, 16, 24);\
    OPNAME ## pixels16_l2_aligned(dst, full, half, stride, 24, 16, 16);\
}\
\
742
static void OPNAME ## qpel16_mc02_sh4(uint8_t *dst, uint8_t *src, int stride){\
743 744 745 746 747
    uint8_t full[24*17];\
    copy_block17(full, src, 24, stride, 17);\
    OPNAME ## mpeg4_qpel16_v_lowpass(dst, full, stride, 24);\
}\
\
748
static void OPNAME ## qpel16_mc03_sh4(uint8_t *dst, uint8_t *src, int stride){\
749 750 751 752 753 754
    uint8_t full[24*17];\
    uint8_t half[256];\
    copy_block17(full, src, 24, stride, 17);\
    put ## RND ## mpeg4_qpel16_v_lowpass(half, full, 16, 24);\
    OPNAME ## pixels16_l2_aligned(dst, full+24, half, stride, 24, 16, 16);\
}\
755
static void OPNAME ## qpel16_mc11_sh4(uint8_t *dst, uint8_t *src, int stride){\
756 757 758 759 760 761 762 763 764
    uint8_t full[24*17];\
    uint8_t halfH[272];\
    uint8_t halfHV[256];\
    copy_block17(full, src, 24, stride, 17);\
    put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17);\
    put ## RND ## pixels16_l2_aligned(halfH, halfH, full, 16, 16, 24, 17);\
    put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16);\
    OPNAME ## pixels16_l2_aligned(dst, halfH, halfHV, stride, 16, 16, 16);\
}\
765
static void OPNAME ## qpel16_mc31_sh4(uint8_t *dst, uint8_t *src, int stride){\
766 767 768 769 770 771 772 773 774
    uint8_t full[24*17];\
    uint8_t halfH[272];\
    uint8_t halfHV[256];\
    copy_block17(full, src, 24, stride, 17);\
    put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17);\
    put ## RND ## pixels16_l2_aligned1(halfH, halfH, full+1, 16, 16, 24, 17);\
    put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16);\
    OPNAME ## pixels16_l2_aligned(dst, halfH, halfHV, stride, 16, 16, 16);\
}\
775
static void OPNAME ## qpel16_mc13_sh4(uint8_t *dst, uint8_t *src, int stride){\
776 777 778 779 780 781 782 783 784
    uint8_t full[24*17];\
    uint8_t halfH[272];\
    uint8_t halfHV[256];\
    copy_block17(full, src, 24, stride, 17);\
    put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17);\
    put ## RND ## pixels16_l2_aligned(halfH, halfH, full, 16, 16, 24, 17);\
    put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16);\
    OPNAME ## pixels16_l2_aligned(dst, halfH+16, halfHV, stride, 16, 16, 16);\
}\
785
static void OPNAME ## qpel16_mc33_sh4(uint8_t *dst, uint8_t *src, int stride){\
786 787 788 789 790 791 792 793 794
    uint8_t full[24*17];\
    uint8_t halfH[272];\
    uint8_t halfHV[256];\
    copy_block17(full, src, 24, stride, 17);\
    put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17);\
    put ## RND ## pixels16_l2_aligned1(halfH, halfH, full+1, 16, 16, 24, 17);\
    put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16);\
    OPNAME ## pixels16_l2_aligned(dst, halfH+16, halfHV, stride, 16, 16, 16);\
}\
795
static void OPNAME ## qpel16_mc21_sh4(uint8_t *dst, uint8_t *src, int stride){\
796 797 798 799 800 801
    uint8_t halfH[272];\
    uint8_t halfHV[256];\
    put ## RND ## mpeg4_qpel16_h_lowpass(halfH, src, 16, stride, 17);\
    put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16);\
    OPNAME ## pixels16_l2_aligned(dst, halfH, halfHV, stride, 16, 16, 16);\
}\
802
static void OPNAME ## qpel16_mc23_sh4(uint8_t *dst, uint8_t *src, int stride){\
803 804 805 806 807 808
    uint8_t halfH[272];\
    uint8_t halfHV[256];\
    put ## RND ## mpeg4_qpel16_h_lowpass(halfH, src, 16, stride, 17);\
    put ## RND ## mpeg4_qpel16_v_lowpass(halfHV, halfH, 16, 16);\
    OPNAME ## pixels16_l2_aligned(dst, halfH+16, halfHV, stride, 16, 16, 16);\
}\
809
static void OPNAME ## qpel16_mc12_sh4(uint8_t *dst, uint8_t *src, int stride){\
810 811 812 813 814 815 816
    uint8_t full[24*17];\
    uint8_t halfH[272];\
    copy_block17(full, src, 24, stride, 17);\
    put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17);\
    put ## RND ## pixels16_l2_aligned(halfH, halfH, full, 16, 16, 24, 17);\
    OPNAME ## mpeg4_qpel16_v_lowpass(dst, halfH, stride, 16);\
}\
817
static void OPNAME ## qpel16_mc32_sh4(uint8_t *dst, uint8_t *src, int stride){\
818 819 820 821 822 823 824
    uint8_t full[24*17];\
    uint8_t halfH[272];\
    copy_block17(full, src, 24, stride, 17);\
    put ## RND ## mpeg4_qpel16_h_lowpass(halfH, full, 16, 24, 17);\
    put ## RND ## pixels16_l2_aligned1(halfH, halfH, full+1, 16, 16, 24, 17);\
    OPNAME ## mpeg4_qpel16_v_lowpass(dst, halfH, stride, 16);\
}\
825
static void OPNAME ## qpel16_mc22_sh4(uint8_t *dst, uint8_t *src, int stride){\
826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846
    uint8_t halfH[272];\
    put ## RND ## mpeg4_qpel16_h_lowpass(halfH, src, 16, stride, 17);\
    OPNAME ## mpeg4_qpel16_v_lowpass(dst, halfH, stride, 16);\
}

#define op_avg(a, b) a = (((a)+cm[((b) + 16)>>5]+1)>>1)
#define op_avg_no_rnd(a, b) a = (((a)+cm[((b) + 15)>>5])>>1)
#define op_put(a, b) a = cm[((b) + 16)>>5]
#define op_put_no_rnd(a, b) a = cm[((b) + 15)>>5]

QPEL_MC(0, put_       , _       , op_put)
QPEL_MC(1, put_no_rnd_, _no_rnd_, op_put_no_rnd)
QPEL_MC(0, avg_       , _       , op_avg)
//QPEL_MC(1, avg_no_rnd , _       , op_avg)
#undef op_avg
#undef op_avg_no_rnd
#undef op_put
#undef op_put_no_rnd

#define H264_LOWPASS(OPNAME, OP, OP2) \
static inline void OPNAME ## h264_qpel_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride,int w,int h){\
847
    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899
    do {\
        int srcB,srcA,src0,src1,src2,src3,src4,src5,src6;\
        uint8_t *s = src-2;\
        srcB = *s++;\
        srcA = *s++;\
        src0 = *s++;\
        src1 = *s++;\
        src2 = *s++;\
        src3 = *s++;\
        OP(dst[0], (src0+src1)*20 - (srcA+src2)*5 + (srcB+src3));\
        src4 = *s++;\
        OP(dst[1], (src1+src2)*20 - (src0+src3)*5 + (srcA+src4));\
        src5 = *s++;\
        OP(dst[2], (src2+src3)*20 - (src1+src4)*5 + (src0+src5));\
        src6 = *s++;\
        OP(dst[3], (src3+src4)*20 - (src2+src5)*5 + (src1+src6));\
      if (w>4) { /* it optimized */ \
        int src7,src8,src9,src10; \
        src7 = *s++;\
        OP(dst[4], (src4+src5)*20 - (src3+src6)*5 + (src2+src7));\
        src8 = *s++;\
        OP(dst[5], (src5+src6)*20 - (src4+src7)*5 + (src3+src8));\
        src9 = *s++;\
        OP(dst[6], (src6+src7)*20 - (src5+src8)*5 + (src4+src9));\
        src10 = *s++;\
        OP(dst[7], (src7+src8)*20 - (src6+src9)*5 + (src5+src10));\
       if (w>8) { \
        int src11,src12,src13,src14,src15,src16,src17,src18; \
        src11 = *s++;\
        OP(dst[8] , (src8 +src9 )*20 - (src7 +src10)*5 + (src6 +src11));\
        src12 = *s++;\
        OP(dst[9] , (src9 +src10)*20 - (src8 +src11)*5 + (src7 +src12));\
        src13 = *s++;\
        OP(dst[10], (src10+src11)*20 - (src9 +src12)*5 + (src8 +src13));\
        src14 = *s++;\
        OP(dst[11], (src11+src12)*20 - (src10+src13)*5 + (src9 +src14));\
        src15 = *s++;\
        OP(dst[12], (src12+src13)*20 - (src11+src14)*5 + (src10+src15));\
        src16 = *s++;\
        OP(dst[13], (src13+src14)*20 - (src12+src15)*5 + (src11+src16));\
        src17 = *s++;\
        OP(dst[14], (src14+src15)*20 - (src13+src16)*5 + (src12+src17));\
        src18 = *s++;\
        OP(dst[15], (src15+src16)*20 - (src14+src17)*5 + (src13+src18));\
       } \
      } \
        dst+=dstStride;\
        src+=srcStride;\
    }while(--h);\
}\
\
static inline void OPNAME ## h264_qpel_v_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride,int w,int h){\
900
    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952
    do{\
        int srcB,srcA,src0,src1,src2,src3,src4,src5,src6;\
        uint8_t *s = src-2*srcStride,*d=dst;\
        srcB = *s; s+=srcStride;\
        srcA = *s; s+=srcStride;\
        src0 = *s; s+=srcStride;\
        src1 = *s; s+=srcStride;\
        src2 = *s; s+=srcStride;\
        src3 = *s; s+=srcStride;\
        OP(*d, (src0+src1)*20 - (srcA+src2)*5 + (srcB+src3));d+=dstStride;\
        src4 = *s; s+=srcStride;\
        OP(*d, (src1+src2)*20 - (src0+src3)*5 + (srcA+src4));d+=dstStride;\
        src5 = *s; s+=srcStride;\
        OP(*d, (src2+src3)*20 - (src1+src4)*5 + (src0+src5));d+=dstStride;\
        src6 = *s; s+=srcStride;\
        OP(*d, (src3+src4)*20 - (src2+src5)*5 + (src1+src6));d+=dstStride;\
      if (h>4) { \
        int src7,src8,src9,src10; \
        src7 = *s; s+=srcStride;\
        OP(*d, (src4+src5)*20 - (src3+src6)*5 + (src2+src7));d+=dstStride;\
        src8 = *s; s+=srcStride;\
        OP(*d, (src5+src6)*20 - (src4+src7)*5 + (src3+src8));d+=dstStride;\
        src9 = *s; s+=srcStride;\
        OP(*d, (src6+src7)*20 - (src5+src8)*5 + (src4+src9));d+=dstStride;\
        src10 = *s; s+=srcStride;\
        OP(*d, (src7+src8)*20 - (src6+src9)*5 + (src5+src10));d+=dstStride;\
       if (h>8) { \
        int src11,src12,src13,src14,src15,src16,src17,src18; \
        src11 = *s; s+=srcStride;\
        OP(*d , (src8 +src9 )*20 - (src7 +src10)*5 + (src6 +src11));d+=dstStride;\
        src12 = *s; s+=srcStride;\
        OP(*d , (src9 +src10)*20 - (src8 +src11)*5 + (src7 +src12));d+=dstStride;\
        src13 = *s; s+=srcStride;\
        OP(*d, (src10+src11)*20 - (src9 +src12)*5 + (src8 +src13));d+=dstStride;\
        src14 = *s; s+=srcStride;\
        OP(*d, (src11+src12)*20 - (src10+src13)*5 + (src9 +src14));d+=dstStride;\
        src15 = *s; s+=srcStride;\
        OP(*d, (src12+src13)*20 - (src11+src14)*5 + (src10+src15));d+=dstStride;\
        src16 = *s; s+=srcStride;\
        OP(*d, (src13+src14)*20 - (src12+src15)*5 + (src11+src16));d+=dstStride;\
        src17 = *s; s+=srcStride;\
        OP(*d, (src14+src15)*20 - (src13+src16)*5 + (src12+src17));d+=dstStride;\
        src18 = *s; s+=srcStride;\
        OP(*d, (src15+src16)*20 - (src14+src17)*5 + (src13+src18));d+=dstStride;\
       } \
      } \
        dst++;\
        src++;\
    }while(--w);\
}\
\
static inline void OPNAME ## h264_qpel_hv_lowpass(uint8_t *dst, int16_t *tmp, uint8_t *src, int dstStride, int tmpStride, int srcStride,int w,int h){\
953
    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089
    int i;\
    src -= 2*srcStride;\
    i= h+5; \
    do {\
        int srcB,srcA,src0,src1,src2,src3,src4,src5,src6;\
        uint8_t *s = src-2;\
        srcB = *s++;\
        srcA = *s++;\
        src0 = *s++;\
        src1 = *s++;\
        src2 = *s++;\
        src3 = *s++;\
        tmp[0] = ((src0+src1)*20 - (srcA+src2)*5 + (srcB+src3));\
        src4 = *s++;\
        tmp[1] = ((src1+src2)*20 - (src0+src3)*5 + (srcA+src4));\
        src5 = *s++;\
        tmp[2] = ((src2+src3)*20 - (src1+src4)*5 + (src0+src5));\
        src6 = *s++;\
        tmp[3] = ((src3+src4)*20 - (src2+src5)*5 + (src1+src6));\
      if (w>4) { /* it optimized */ \
        int src7,src8,src9,src10; \
        src7 = *s++;\
        tmp[4] = ((src4+src5)*20 - (src3+src6)*5 + (src2+src7));\
        src8 = *s++;\
        tmp[5] = ((src5+src6)*20 - (src4+src7)*5 + (src3+src8));\
        src9 = *s++;\
        tmp[6] = ((src6+src7)*20 - (src5+src8)*5 + (src4+src9));\
        src10 = *s++;\
        tmp[7] = ((src7+src8)*20 - (src6+src9)*5 + (src5+src10));\
       if (w>8) { \
        int src11,src12,src13,src14,src15,src16,src17,src18; \
        src11 = *s++;\
        tmp[8] = ((src8 +src9 )*20 - (src7 +src10)*5 + (src6 +src11));\
        src12 = *s++;\
        tmp[9] = ((src9 +src10)*20 - (src8 +src11)*5 + (src7 +src12));\
        src13 = *s++;\
        tmp[10] = ((src10+src11)*20 - (src9 +src12)*5 + (src8 +src13));\
        src14 = *s++;\
        tmp[11] = ((src11+src12)*20 - (src10+src13)*5 + (src9 +src14));\
        src15 = *s++;\
        tmp[12] = ((src12+src13)*20 - (src11+src14)*5 + (src10+src15));\
        src16 = *s++;\
        tmp[13] = ((src13+src14)*20 - (src12+src15)*5 + (src11+src16));\
        src17 = *s++;\
        tmp[14] = ((src14+src15)*20 - (src13+src16)*5 + (src12+src17));\
        src18 = *s++;\
        tmp[15] = ((src15+src16)*20 - (src14+src17)*5 + (src13+src18));\
       } \
      } \
        tmp+=tmpStride;\
        src+=srcStride;\
    }while(--i);\
    tmp -= tmpStride*(h+5-2);\
    i = w; \
    do {\
        int tmpB,tmpA,tmp0,tmp1,tmp2,tmp3,tmp4,tmp5,tmp6;\
        int16_t *s = tmp-2*tmpStride; \
        uint8_t *d=dst;\
        tmpB = *s; s+=tmpStride;\
        tmpA = *s; s+=tmpStride;\
        tmp0 = *s; s+=tmpStride;\
        tmp1 = *s; s+=tmpStride;\
        tmp2 = *s; s+=tmpStride;\
        tmp3 = *s; s+=tmpStride;\
        OP2(*d, (tmp0+tmp1)*20 - (tmpA+tmp2)*5 + (tmpB+tmp3));d+=dstStride;\
        tmp4 = *s; s+=tmpStride;\
        OP2(*d, (tmp1+tmp2)*20 - (tmp0+tmp3)*5 + (tmpA+tmp4));d+=dstStride;\
        tmp5 = *s; s+=tmpStride;\
        OP2(*d, (tmp2+tmp3)*20 - (tmp1+tmp4)*5 + (tmp0+tmp5));d+=dstStride;\
        tmp6 = *s; s+=tmpStride;\
        OP2(*d, (tmp3+tmp4)*20 - (tmp2+tmp5)*5 + (tmp1+tmp6));d+=dstStride;\
      if (h>4) { \
        int tmp7,tmp8,tmp9,tmp10; \
        tmp7 = *s; s+=tmpStride;\
        OP2(*d, (tmp4+tmp5)*20 - (tmp3+tmp6)*5 + (tmp2+tmp7));d+=dstStride;\
        tmp8 = *s; s+=tmpStride;\
        OP2(*d, (tmp5+tmp6)*20 - (tmp4+tmp7)*5 + (tmp3+tmp8));d+=dstStride;\
        tmp9 = *s; s+=tmpStride;\
        OP2(*d, (tmp6+tmp7)*20 - (tmp5+tmp8)*5 + (tmp4+tmp9));d+=dstStride;\
        tmp10 = *s; s+=tmpStride;\
        OP2(*d, (tmp7+tmp8)*20 - (tmp6+tmp9)*5 + (tmp5+tmp10));d+=dstStride;\
       if (h>8) { \
        int tmp11,tmp12,tmp13,tmp14,tmp15,tmp16,tmp17,tmp18; \
        tmp11 = *s; s+=tmpStride;\
        OP2(*d , (tmp8 +tmp9 )*20 - (tmp7 +tmp10)*5 + (tmp6 +tmp11));d+=dstStride;\
        tmp12 = *s; s+=tmpStride;\
        OP2(*d , (tmp9 +tmp10)*20 - (tmp8 +tmp11)*5 + (tmp7 +tmp12));d+=dstStride;\
        tmp13 = *s; s+=tmpStride;\
        OP2(*d, (tmp10+tmp11)*20 - (tmp9 +tmp12)*5 + (tmp8 +tmp13));d+=dstStride;\
        tmp14 = *s; s+=tmpStride;\
        OP2(*d, (tmp11+tmp12)*20 - (tmp10+tmp13)*5 + (tmp9 +tmp14));d+=dstStride;\
        tmp15 = *s; s+=tmpStride;\
        OP2(*d, (tmp12+tmp13)*20 - (tmp11+tmp14)*5 + (tmp10+tmp15));d+=dstStride;\
        tmp16 = *s; s+=tmpStride;\
        OP2(*d, (tmp13+tmp14)*20 - (tmp12+tmp15)*5 + (tmp11+tmp16));d+=dstStride;\
        tmp17 = *s; s+=tmpStride;\
        OP2(*d, (tmp14+tmp15)*20 - (tmp13+tmp16)*5 + (tmp12+tmp17));d+=dstStride;\
        tmp18 = *s; s+=tmpStride;\
        OP2(*d, (tmp15+tmp16)*20 - (tmp14+tmp17)*5 + (tmp13+tmp18));d+=dstStride;\
       } \
      } \
        dst++;\
        tmp++;\
    }while(--i);\
}\
\
static void OPNAME ## h264_qpel4_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
    OPNAME ## h264_qpel_h_lowpass(dst,src,dstStride,srcStride,4,4); \
}\
static void OPNAME ## h264_qpel8_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
   OPNAME ## h264_qpel_h_lowpass(dst,src,dstStride,srcStride,8,8); \
}\
static void OPNAME ## h264_qpel16_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
   OPNAME ## h264_qpel_h_lowpass(dst,src,dstStride,srcStride,16,16); \
}\
\
static void OPNAME ## h264_qpel4_v_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
   OPNAME ## h264_qpel_v_lowpass(dst,src,dstStride,srcStride,4,4); \
}\
static void OPNAME ## h264_qpel8_v_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
   OPNAME ## h264_qpel_v_lowpass(dst,src,dstStride,srcStride,8,8); \
}\
static void OPNAME ## h264_qpel16_v_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
   OPNAME ## h264_qpel_v_lowpass(dst,src,dstStride,srcStride,16,16); \
}\
static void OPNAME ## h264_qpel4_hv_lowpass(uint8_t *dst, int16_t *tmp, uint8_t *src, int dstStride, int tmpStride, int srcStride){\
   OPNAME ## h264_qpel_hv_lowpass(dst,tmp,src,dstStride,tmpStride,srcStride,4,4); \
}\
static void OPNAME ## h264_qpel8_hv_lowpass(uint8_t *dst, int16_t *tmp, uint8_t *src, int dstStride, int tmpStride, int srcStride){\
   OPNAME ## h264_qpel_hv_lowpass(dst,tmp,src,dstStride,tmpStride,srcStride,8,8); \
}\
static void OPNAME ## h264_qpel16_hv_lowpass(uint8_t *dst, int16_t *tmp, uint8_t *src, int dstStride, int tmpStride, int srcStride){\
   OPNAME ## h264_qpel_hv_lowpass(dst,tmp,src,dstStride,tmpStride,srcStride,16,16); \
}\

#define H264_MC(OPNAME, SIZE) \
1090
static void OPNAME ## h264_qpel ## SIZE ## _mc00_sh4 (uint8_t *dst, uint8_t *src, int stride){\
1091 1092 1093
    OPNAME ## pixels ## SIZE ## _c(dst, src, stride, SIZE);\
}\
\
1094
static void OPNAME ## h264_qpel ## SIZE ## _mc10_sh4(uint8_t *dst, uint8_t *src, int stride){\
1095 1096 1097 1098 1099
    uint8_t half[SIZE*SIZE];\
    put_h264_qpel ## SIZE ## _h_lowpass(half, src, SIZE, stride);\
    OPNAME ## pixels ## SIZE ## _l2_aligned2(dst, src, half, stride, stride, SIZE, SIZE);\
}\
\
1100
static void OPNAME ## h264_qpel ## SIZE ## _mc20_sh4(uint8_t *dst, uint8_t *src, int stride){\
1101 1102 1103
    OPNAME ## h264_qpel ## SIZE ## _h_lowpass(dst, src, stride, stride);\
}\
\
1104
static void OPNAME ## h264_qpel ## SIZE ## _mc30_sh4(uint8_t *dst, uint8_t *src, int stride){\
1105 1106 1107 1108 1109
    uint8_t half[SIZE*SIZE];\
    put_h264_qpel ## SIZE ## _h_lowpass(half, src, SIZE, stride);\
    OPNAME ## pixels ## SIZE ## _l2_aligned2(dst, src+1, half, stride, stride, SIZE, SIZE);\
}\
\
1110
static void OPNAME ## h264_qpel ## SIZE ## _mc01_sh4(uint8_t *dst, uint8_t *src, int stride){\
1111 1112 1113 1114 1115 1116 1117 1118
    uint8_t full[SIZE*(SIZE+5)];\
    uint8_t * const full_mid= full + SIZE*2;\
    uint8_t half[SIZE*SIZE];\
    copy_block ## SIZE (full, src - stride*2, SIZE,  stride, SIZE + 5);\
    put_h264_qpel ## SIZE ## _v_lowpass(half, full_mid, SIZE, SIZE);\
    OPNAME ## pixels ## SIZE ## _l2_aligned(dst, full_mid, half, stride, SIZE, SIZE, SIZE);\
}\
\
1119
static void OPNAME ## h264_qpel ## SIZE ## _mc02_sh4(uint8_t *dst, uint8_t *src, int stride){\
1120 1121 1122 1123 1124 1125
    uint8_t full[SIZE*(SIZE+5)];\
    uint8_t * const full_mid= full + SIZE*2;\
    copy_block ## SIZE (full, src - stride*2, SIZE,  stride, SIZE + 5);\
    OPNAME ## h264_qpel ## SIZE ## _v_lowpass(dst, full_mid, stride, SIZE);\
}\
\
1126
static void OPNAME ## h264_qpel ## SIZE ## _mc03_sh4(uint8_t *dst, uint8_t *src, int stride){\
1127 1128 1129 1130 1131 1132 1133 1134
    uint8_t full[SIZE*(SIZE+5)];\
    uint8_t * const full_mid= full + SIZE*2;\
    uint8_t half[SIZE*SIZE];\
    copy_block ## SIZE (full, src - stride*2, SIZE,  stride, SIZE + 5);\
    put_h264_qpel ## SIZE ## _v_lowpass(half, full_mid, SIZE, SIZE);\
    OPNAME ## pixels ## SIZE ## _l2_aligned(dst, full_mid+SIZE, half, stride, SIZE, SIZE, SIZE);\
}\
\
1135
static void OPNAME ## h264_qpel ## SIZE ## _mc11_sh4(uint8_t *dst, uint8_t *src, int stride){\
1136 1137 1138 1139 1140 1141 1142 1143 1144 1145
    uint8_t full[SIZE*(SIZE+5)];\
    uint8_t * const full_mid= full + SIZE*2;\
    uint8_t halfH[SIZE*SIZE];\
    uint8_t halfV[SIZE*SIZE];\
    put_h264_qpel ## SIZE ## _h_lowpass(halfH, src, SIZE, stride);\
    copy_block ## SIZE (full, src - stride*2, SIZE,  stride, SIZE + 5);\
    put_h264_qpel ## SIZE ## _v_lowpass(halfV, full_mid, SIZE, SIZE);\
    OPNAME ## pixels ## SIZE ## _l2_aligned(dst, halfH, halfV, stride, SIZE, SIZE, SIZE);\
}\
\
1146
static void OPNAME ## h264_qpel ## SIZE ## _mc31_sh4(uint8_t *dst, uint8_t *src, int stride){\
1147 1148 1149 1150 1151 1152 1153 1154 1155 1156
    uint8_t full[SIZE*(SIZE+5)];\
    uint8_t * const full_mid= full + SIZE*2;\
    uint8_t halfH[SIZE*SIZE];\
    uint8_t halfV[SIZE*SIZE];\
    put_h264_qpel ## SIZE ## _h_lowpass(halfH, src, SIZE, stride);\
    copy_block ## SIZE (full, src - stride*2 + 1, SIZE,  stride, SIZE + 5);\
    put_h264_qpel ## SIZE ## _v_lowpass(halfV, full_mid, SIZE, SIZE);\
    OPNAME ## pixels ## SIZE ## _l2_aligned(dst, halfH, halfV, stride, SIZE, SIZE, SIZE);\
}\
\
1157
static void OPNAME ## h264_qpel ## SIZE ## _mc13_sh4(uint8_t *dst, uint8_t *src, int stride){\
1158 1159 1160 1161 1162 1163 1164 1165 1166 1167
    uint8_t full[SIZE*(SIZE+5)];\
    uint8_t * const full_mid= full + SIZE*2;\
    uint8_t halfH[SIZE*SIZE];\
    uint8_t halfV[SIZE*SIZE];\
    put_h264_qpel ## SIZE ## _h_lowpass(halfH, src + stride, SIZE, stride);\
    copy_block ## SIZE (full, src - stride*2, SIZE,  stride, SIZE + 5);\
    put_h264_qpel ## SIZE ## _v_lowpass(halfV, full_mid, SIZE, SIZE);\
    OPNAME ## pixels ## SIZE ## _l2_aligned(dst, halfH, halfV, stride, SIZE, SIZE, SIZE);\
}\
\
1168
static void OPNAME ## h264_qpel ## SIZE ## _mc33_sh4(uint8_t *dst, uint8_t *src, int stride){\
1169 1170 1171 1172 1173 1174 1175 1176 1177 1178
    uint8_t full[SIZE*(SIZE+5)];\
    uint8_t * const full_mid= full + SIZE*2;\
    uint8_t halfH[SIZE*SIZE];\
    uint8_t halfV[SIZE*SIZE];\
    put_h264_qpel ## SIZE ## _h_lowpass(halfH, src + stride, SIZE, stride);\
    copy_block ## SIZE (full, src - stride*2 + 1, SIZE,  stride, SIZE + 5);\
    put_h264_qpel ## SIZE ## _v_lowpass(halfV, full_mid, SIZE, SIZE);\
    OPNAME ## pixels ## SIZE ## _l2_aligned(dst, halfH, halfV, stride, SIZE, SIZE, SIZE);\
}\
\
1179
static void OPNAME ## h264_qpel ## SIZE ## _mc22_sh4(uint8_t *dst, uint8_t *src, int stride){\
1180 1181 1182 1183
    int16_t tmp[SIZE*(SIZE+5)];\
    OPNAME ## h264_qpel ## SIZE ## _hv_lowpass(dst, tmp, src, stride, SIZE, stride);\
}\
\
1184
static void OPNAME ## h264_qpel ## SIZE ## _mc21_sh4(uint8_t *dst, uint8_t *src, int stride){\
1185 1186 1187 1188 1189 1190 1191 1192
    int16_t tmp[SIZE*(SIZE+5)];\
    uint8_t halfH[SIZE*SIZE];\
    uint8_t halfHV[SIZE*SIZE];\
    put_h264_qpel ## SIZE ## _h_lowpass(halfH, src, SIZE, stride);\
    put_h264_qpel ## SIZE ## _hv_lowpass(halfHV, tmp, src, SIZE, SIZE, stride);\
    OPNAME ## pixels ## SIZE ## _l2_aligned(dst, halfH, halfHV, stride, SIZE, SIZE, SIZE);\
}\
\
1193
static void OPNAME ## h264_qpel ## SIZE ## _mc23_sh4(uint8_t *dst, uint8_t *src, int stride){\
1194 1195 1196 1197 1198 1199 1200 1201
    int16_t tmp[SIZE*(SIZE+5)];\
    uint8_t halfH[SIZE*SIZE];\
    uint8_t halfHV[SIZE*SIZE];\
    put_h264_qpel ## SIZE ## _h_lowpass(halfH, src + stride, SIZE, stride);\
    put_h264_qpel ## SIZE ## _hv_lowpass(halfHV, tmp, src, SIZE, SIZE, stride);\
    OPNAME ## pixels ## SIZE ## _l2_aligned(dst, halfH, halfHV, stride, SIZE, SIZE, SIZE);\
}\
\
1202
static void OPNAME ## h264_qpel ## SIZE ## _mc12_sh4(uint8_t *dst, uint8_t *src, int stride){\
1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213
    uint8_t full[SIZE*(SIZE+5)];\
    uint8_t * const full_mid= full + SIZE*2;\
    int16_t tmp[SIZE*(SIZE+5)];\
    uint8_t halfV[SIZE*SIZE];\
    uint8_t halfHV[SIZE*SIZE];\
    copy_block ## SIZE (full, src - stride*2, SIZE,  stride, SIZE + 5);\
    put_h264_qpel ## SIZE ## _v_lowpass(halfV, full_mid, SIZE, SIZE);\
    put_h264_qpel ## SIZE ## _hv_lowpass(halfHV, tmp, src, SIZE, SIZE, stride);\
    OPNAME ## pixels ## SIZE ## _l2_aligned(dst, halfV, halfHV, stride, SIZE, SIZE, SIZE);\
}\
\
1214
static void OPNAME ## h264_qpel ## SIZE ## _mc32_sh4(uint8_t *dst, uint8_t *src, int stride){\
1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246
    uint8_t full[SIZE*(SIZE+5)];\
    uint8_t * const full_mid= full + SIZE*2;\
    int16_t tmp[SIZE*(SIZE+5)];\
    uint8_t halfV[SIZE*SIZE];\
    uint8_t halfHV[SIZE*SIZE];\
    copy_block ## SIZE (full, src - stride*2 + 1, SIZE,  stride, SIZE + 5);\
    put_h264_qpel ## SIZE ## _v_lowpass(halfV, full_mid, SIZE, SIZE);\
    put_h264_qpel ## SIZE ## _hv_lowpass(halfHV, tmp, src, SIZE, SIZE, stride);\
    OPNAME ## pixels ## SIZE ## _l2_aligned(dst, halfV, halfHV, stride, SIZE, SIZE, SIZE);\
}\

#define op_avg(a, b)  a = (((a)+cm[((b) + 16)>>5]+1)>>1)
//#define op_avg2(a, b) a = (((a)*w1+cm[((b) + 16)>>5]*w2 + o + 64)>>7)
#define op_put(a, b)  a = cm[((b) + 16)>>5]
#define op2_avg(a, b)  a = (((a)+cm[((b) + 512)>>10]+1)>>1)
#define op2_put(a, b)  a = cm[((b) + 512)>>10]

H264_LOWPASS(put_       , op_put, op2_put)
H264_LOWPASS(avg_       , op_avg, op2_avg)
H264_MC(put_, 4)
H264_MC(put_, 8)
H264_MC(put_, 16)
H264_MC(avg_, 4)
H264_MC(avg_, 8)
H264_MC(avg_, 16)

#undef op_avg
#undef op_put
#undef op2_avg
#undef op2_put

static void wmv2_mspel8_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h){
1247
    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271

    do{
        int src_1,src0,src1,src2,src3,src4,src5,src6,src7,src8,src9;
        uint8_t *s = src;
        src_1 = s[-1];
        src0 = *s++;
        src1 = *s++;
        src2 = *s++;
        dst[0]= cm[(9*(src0 + src1) - (src_1 + src2) + 8)>>4];
        src3 = *s++;
        dst[1]= cm[(9*(src1 + src2) - (src0 + src3) + 8)>>4];
        src4 = *s++;
        dst[2]= cm[(9*(src2 + src3) - (src1 + src4) + 8)>>4];
        src5 = *s++;
        dst[3]= cm[(9*(src3 + src4) - (src2 + src5) + 8)>>4];
        src6 = *s++;
        dst[4]= cm[(9*(src4 + src5) - (src3 + src6) + 8)>>4];
        src7 = *s++;
        dst[5]= cm[(9*(src5 + src6) - (src4 + src7) + 8)>>4];
        src8 = *s++;
        dst[6]= cm[(9*(src6 + src7) - (src5 + src8) + 8)>>4];
        src9 = *s++;
        dst[7]= cm[(9*(src7 + src8) - (src6 + src9) + 8)>>4];
        dst+=dstStride;
1272
        src+=srcStride;
1273 1274 1275 1276
    }while(--h);
}

static void wmv2_mspel8_v_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int w){
1277
    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305

    do{
        int src_1,src0,src1,src2,src3,src4,src5,src6,src7,src8,src9;
        uint8_t *s = src,*d = dst;
        src_1 = *(s-srcStride);
        src0 = *s; s+=srcStride;
        src1 = *s; s+=srcStride;
        src2 = *s; s+=srcStride;
        *d= cm[(9*(src0 + src1) - (src_1 + src2) + 8)>>4]; d+=dstStride;
        src3 = *s; s+=srcStride;
        *d= cm[(9*(src1 + src2) - (src0  + src3) + 8)>>4]; d+=dstStride;
        src4 = *s; s+=srcStride;
        *d= cm[(9*(src2 + src3) - (src1  + src4) + 8)>>4]; d+=dstStride;
        src5 = *s; s+=srcStride;
        *d= cm[(9*(src3 + src4) - (src2  + src5) + 8)>>4]; d+=dstStride;
        src6 = *s; s+=srcStride;
        *d= cm[(9*(src4 + src5) - (src3  + src6) + 8)>>4]; d+=dstStride;
        src7 = *s; s+=srcStride;
        *d= cm[(9*(src5 + src6) - (src4  + src7) + 8)>>4]; d+=dstStride;
        src8 = *s; s+=srcStride;
        *d= cm[(9*(src6 + src7) - (src5  + src8) + 8)>>4]; d+=dstStride;
        src9 = *s;
        *d= cm[(9*(src7 + src8) - (src6  + src9) + 8)>>4]; d+=dstStride;
        src++;
        dst++;
    }while(--w);
}

1306
static void put_mspel8_mc00_sh4 (uint8_t *dst, uint8_t *src, int stride){
1307 1308 1309
    put_pixels8_c(dst, src, stride, 8);
}

1310
static void put_mspel8_mc10_sh4(uint8_t *dst, uint8_t *src, int stride){
1311 1312 1313 1314 1315
    uint8_t half[64];
    wmv2_mspel8_h_lowpass(half, src, 8, stride, 8);
    put_pixels8_l2_aligned2(dst, src, half, stride, stride, 8, 8);
}

1316
static void put_mspel8_mc20_sh4(uint8_t *dst, uint8_t *src, int stride){
1317 1318 1319
    wmv2_mspel8_h_lowpass(dst, src, stride, stride, 8);
}

1320
static void put_mspel8_mc30_sh4(uint8_t *dst, uint8_t *src, int stride){
1321 1322 1323 1324 1325
    uint8_t half[64];
    wmv2_mspel8_h_lowpass(half, src, 8, stride, 8);
    put_pixels8_l2_aligned2(dst, src+1, half, stride, stride, 8, 8);
}

1326
static void put_mspel8_mc02_sh4(uint8_t *dst, uint8_t *src, int stride){
1327 1328 1329
    wmv2_mspel8_v_lowpass(dst, src, stride, stride, 8);
}

1330
static void put_mspel8_mc12_sh4(uint8_t *dst, uint8_t *src, int stride){
1331 1332 1333 1334 1335 1336 1337 1338
    uint8_t halfH[88];
    uint8_t halfV[64];
    uint8_t halfHV[64];
    wmv2_mspel8_h_lowpass(halfH, src-stride, 8, stride, 11);
    wmv2_mspel8_v_lowpass(halfV, src, 8, stride, 8);
    wmv2_mspel8_v_lowpass(halfHV, halfH+8, 8, 8, 8);
    put_pixels8_l2_aligned(dst, halfV, halfHV, stride, 8, 8, 8);
}
1339
static void put_mspel8_mc32_sh4(uint8_t *dst, uint8_t *src, int stride){
1340 1341 1342 1343 1344 1345 1346 1347
    uint8_t halfH[88];
    uint8_t halfV[64];
    uint8_t halfHV[64];
    wmv2_mspel8_h_lowpass(halfH, src-stride, 8, stride, 11);
    wmv2_mspel8_v_lowpass(halfV, src+1, 8, stride, 8);
    wmv2_mspel8_v_lowpass(halfHV, halfH+8, 8, 8, 8);
    put_pixels8_l2_aligned(dst, halfV, halfHV, stride, 8, 8, 8);
}
1348
static void put_mspel8_mc22_sh4(uint8_t *dst, uint8_t *src, int stride){
1349 1350 1351 1352
    uint8_t halfH[88];
    wmv2_mspel8_h_lowpass(halfH, src-stride, 8, stride, 11);
    wmv2_mspel8_v_lowpass(dst, halfH+8, stride, 8, 8);
}