| 1 | /*
|
|---|
| 2 | * MPEG video MMX templates
|
|---|
| 3 | *
|
|---|
| 4 | * Copyright (c) 2002 Michael Niedermayer <[email protected]>
|
|---|
| 5 | *
|
|---|
| 6 | * This library is free software; you can redistribute it and/or
|
|---|
| 7 | * modify it under the terms of the GNU Lesser General Public
|
|---|
| 8 | * License as published by the Free Software Foundation; either
|
|---|
| 9 | * version 2 of the License, or (at your option) any later version.
|
|---|
| 10 | *
|
|---|
| 11 | * This library is distributed in the hope that it will be useful,
|
|---|
| 12 | * but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|---|
| 13 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
|
|---|
| 14 | * Lesser General Public License for more details.
|
|---|
| 15 | *
|
|---|
| 16 | * You should have received a copy of the GNU Lesser General Public
|
|---|
| 17 | * License along with this library; if not, write to the Free Software
|
|---|
| 18 | * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
|
|---|
| 19 | */
|
|---|
| 20 | #undef SPREADW
|
|---|
| 21 | #undef PMAXW
|
|---|
| 22 | #undef PMAX
|
|---|
| 23 | #ifdef HAVE_MMX2
|
|---|
| 24 | #define SPREADW(a) "pshufw $0, " #a ", " #a " \n\t"
|
|---|
| 25 | #define PMAXW(a,b) "pmaxsw " #a ", " #b " \n\t"
|
|---|
| 26 | #define PMAX(a,b) \
|
|---|
| 27 | "pshufw $0x0E," #a ", " #b " \n\t"\
|
|---|
| 28 | PMAXW(b, a)\
|
|---|
| 29 | "pshufw $0x01," #a ", " #b " \n\t"\
|
|---|
| 30 | PMAXW(b, a)
|
|---|
| 31 | #else
|
|---|
| 32 | #define SPREADW(a) \
|
|---|
| 33 | "punpcklwd " #a ", " #a " \n\t"\
|
|---|
| 34 | "punpcklwd " #a ", " #a " \n\t"
|
|---|
| 35 | #define PMAXW(a,b) \
|
|---|
| 36 | "psubusw " #a ", " #b " \n\t"\
|
|---|
| 37 | "paddw " #a ", " #b " \n\t"
|
|---|
| 38 | #define PMAX(a,b) \
|
|---|
| 39 | "movq " #a ", " #b " \n\t"\
|
|---|
| 40 | "psrlq $32, " #a " \n\t"\
|
|---|
| 41 | PMAXW(b, a)\
|
|---|
| 42 | "movq " #a ", " #b " \n\t"\
|
|---|
| 43 | "psrlq $16, " #a " \n\t"\
|
|---|
| 44 | PMAXW(b, a)
|
|---|
| 45 |
|
|---|
| 46 | #endif
|
|---|
| 47 |
|
|---|
| 48 | static int RENAME(dct_quantize)(MpegEncContext *s,
|
|---|
| 49 | DCTELEM *block, int n,
|
|---|
| 50 | int qscale, int *overflow)
|
|---|
| 51 | {
|
|---|
| 52 | long last_non_zero_p1;
|
|---|
| 53 | int level=0, q; //=0 is cuz gcc says uninitalized ...
|
|---|
| 54 | const uint16_t *qmat, *bias;
|
|---|
| 55 | DECLARE_ALIGNED_8(int16_t, temp_block[64]);
|
|---|
| 56 |
|
|---|
| 57 | assert((7&(int)(&temp_block[0])) == 0); //did gcc align it correctly?
|
|---|
| 58 |
|
|---|
| 59 | //s->fdct (block);
|
|---|
| 60 | RENAMEl(ff_fdct) (block); //cant be anything else ...
|
|---|
| 61 |
|
|---|
| 62 | if(s->dct_error_sum)
|
|---|
| 63 | s->denoise_dct(s, block);
|
|---|
| 64 |
|
|---|
| 65 | if (s->mb_intra) {
|
|---|
| 66 | int dummy;
|
|---|
| 67 | if (n < 4)
|
|---|
| 68 | q = s->y_dc_scale;
|
|---|
| 69 | else
|
|---|
| 70 | q = s->c_dc_scale;
|
|---|
| 71 | /* note: block[0] is assumed to be positive */
|
|---|
| 72 | if (!s->h263_aic) {
|
|---|
| 73 | #if 1
|
|---|
| 74 | asm volatile (
|
|---|
| 75 | "mul %%ecx \n\t"
|
|---|
| 76 | : "=d" (level), "=a"(dummy)
|
|---|
| 77 | : "a" ((block[0]>>2) + q), "c" (inverse[q<<1])
|
|---|
| 78 | );
|
|---|
| 79 | #else
|
|---|
| 80 | asm volatile (
|
|---|
| 81 | "xorl %%edx, %%edx \n\t"
|
|---|
| 82 | "divw %%cx \n\t"
|
|---|
| 83 | "movzwl %%ax, %%eax \n\t"
|
|---|
| 84 | : "=a" (level)
|
|---|
| 85 | : "a" ((block[0]>>2) + q), "c" (q<<1)
|
|---|
| 86 | : "%edx"
|
|---|
| 87 | );
|
|---|
| 88 | #endif
|
|---|
| 89 | } else
|
|---|
| 90 | /* For AIC we skip quant/dequant of INTRADC */
|
|---|
| 91 | level = (block[0] + 4)>>3;
|
|---|
| 92 |
|
|---|
| 93 | block[0]=0; //avoid fake overflow
|
|---|
| 94 | // temp_block[0] = (block[0] + (q >> 1)) / q;
|
|---|
| 95 | last_non_zero_p1 = 1;
|
|---|
| 96 | bias = s->q_intra_matrix16[qscale][1];
|
|---|
| 97 | qmat = s->q_intra_matrix16[qscale][0];
|
|---|
| 98 | } else {
|
|---|
| 99 | last_non_zero_p1 = 0;
|
|---|
| 100 | bias = s->q_inter_matrix16[qscale][1];
|
|---|
| 101 | qmat = s->q_inter_matrix16[qscale][0];
|
|---|
| 102 | }
|
|---|
| 103 |
|
|---|
| 104 | if((s->out_format == FMT_H263 || s->out_format == FMT_H261) && s->mpeg_quant==0){
|
|---|
| 105 |
|
|---|
| 106 | asm volatile(
|
|---|
| 107 | "movd %%"REG_a", %%mm3 \n\t" // last_non_zero_p1
|
|---|
| 108 | SPREADW(%%mm3)
|
|---|
| 109 | "pxor %%mm7, %%mm7 \n\t" // 0
|
|---|
| 110 | "pxor %%mm4, %%mm4 \n\t" // 0
|
|---|
| 111 | "movq (%2), %%mm5 \n\t" // qmat[0]
|
|---|
| 112 | "pxor %%mm6, %%mm6 \n\t"
|
|---|
| 113 | "psubw (%3), %%mm6 \n\t" // -bias[0]
|
|---|
| 114 | "mov $-128, %%"REG_a" \n\t"
|
|---|
| 115 | #if !defined(VBOX) || !defined(__DARWIN__)
|
|---|
| 116 | ".balign 16 \n\t"
|
|---|
| 117 | #else
|
|---|
| 118 | ".align 4 \n\t"
|
|---|
| 119 | #endif
|
|---|
| 120 | "1: \n\t"
|
|---|
| 121 | "pxor %%mm1, %%mm1 \n\t" // 0
|
|---|
| 122 | "movq (%1, %%"REG_a"), %%mm0 \n\t" // block[i]
|
|---|
| 123 | "pcmpgtw %%mm0, %%mm1 \n\t" // block[i] <= 0 ? 0xFF : 0x00
|
|---|
| 124 | "pxor %%mm1, %%mm0 \n\t"
|
|---|
| 125 | "psubw %%mm1, %%mm0 \n\t" // ABS(block[i])
|
|---|
| 126 | "psubusw %%mm6, %%mm0 \n\t" // ABS(block[i]) + bias[0]
|
|---|
| 127 | "pmulhw %%mm5, %%mm0 \n\t" // (ABS(block[i])*qmat[0] - bias[0]*qmat[0])>>16
|
|---|
| 128 | "por %%mm0, %%mm4 \n\t"
|
|---|
| 129 | "pxor %%mm1, %%mm0 \n\t"
|
|---|
| 130 | "psubw %%mm1, %%mm0 \n\t" // out=((ABS(block[i])*qmat[0] - bias[0]*qmat[0])>>16)*sign(block[i])
|
|---|
| 131 | "movq %%mm0, (%5, %%"REG_a") \n\t"
|
|---|
| 132 | "pcmpeqw %%mm7, %%mm0 \n\t" // out==0 ? 0xFF : 0x00
|
|---|
| 133 | "movq (%4, %%"REG_a"), %%mm1 \n\t"
|
|---|
| 134 | "movq %%mm7, (%1, %%"REG_a") \n\t" // 0
|
|---|
| 135 | "pandn %%mm1, %%mm0 \n\t"
|
|---|
| 136 | PMAXW(%%mm0, %%mm3)
|
|---|
| 137 | "add $8, %%"REG_a" \n\t"
|
|---|
| 138 | " js 1b \n\t"
|
|---|
| 139 | PMAX(%%mm3, %%mm0)
|
|---|
| 140 | "movd %%mm3, %%"REG_a" \n\t"
|
|---|
| 141 | "movzb %%al, %%"REG_a" \n\t" // last_non_zero_p1
|
|---|
| 142 | : "+a" (last_non_zero_p1)
|
|---|
| 143 | : "r" (block+64), "r" (qmat), "r" (bias),
|
|---|
| 144 | "r" (inv_zigzag_direct16+64), "r" (temp_block+64)
|
|---|
| 145 | );
|
|---|
| 146 | // note the asm is split cuz gcc doesnt like that many operands ...
|
|---|
| 147 | asm volatile(
|
|---|
| 148 | "movd %1, %%mm1 \n\t" // max_qcoeff
|
|---|
| 149 | SPREADW(%%mm1)
|
|---|
| 150 | "psubusw %%mm1, %%mm4 \n\t"
|
|---|
| 151 | "packuswb %%mm4, %%mm4 \n\t"
|
|---|
| 152 | "movd %%mm4, %0 \n\t" // *overflow
|
|---|
| 153 | : "=g" (*overflow)
|
|---|
| 154 | : "g" (s->max_qcoeff)
|
|---|
| 155 | );
|
|---|
| 156 | }else{ // FMT_H263
|
|---|
| 157 | asm volatile(
|
|---|
| 158 | "movd %%"REG_a", %%mm3 \n\t" // last_non_zero_p1
|
|---|
| 159 | SPREADW(%%mm3)
|
|---|
| 160 | "pxor %%mm7, %%mm7 \n\t" // 0
|
|---|
| 161 | "pxor %%mm4, %%mm4 \n\t" // 0
|
|---|
| 162 | "mov $-128, %%"REG_a" \n\t"
|
|---|
| 163 | #if !defined(VBOX) || !defined(__DARWIN__)
|
|---|
| 164 | ".balign 16 \n\t"
|
|---|
| 165 | #else
|
|---|
| 166 | ".align 4 \n\t"
|
|---|
| 167 | #endif
|
|---|
| 168 | "1: \n\t"
|
|---|
| 169 | "pxor %%mm1, %%mm1 \n\t" // 0
|
|---|
| 170 | "movq (%1, %%"REG_a"), %%mm0 \n\t" // block[i]
|
|---|
| 171 | "pcmpgtw %%mm0, %%mm1 \n\t" // block[i] <= 0 ? 0xFF : 0x00
|
|---|
| 172 | "pxor %%mm1, %%mm0 \n\t"
|
|---|
| 173 | "psubw %%mm1, %%mm0 \n\t" // ABS(block[i])
|
|---|
| 174 | "movq (%3, %%"REG_a"), %%mm6 \n\t" // bias[0]
|
|---|
| 175 | "paddusw %%mm6, %%mm0 \n\t" // ABS(block[i]) + bias[0]
|
|---|
| 176 | "movq (%2, %%"REG_a"), %%mm5 \n\t" // qmat[i]
|
|---|
| 177 | "pmulhw %%mm5, %%mm0 \n\t" // (ABS(block[i])*qmat[0] + bias[0]*qmat[0])>>16
|
|---|
| 178 | "por %%mm0, %%mm4 \n\t"
|
|---|
| 179 | "pxor %%mm1, %%mm0 \n\t"
|
|---|
| 180 | "psubw %%mm1, %%mm0 \n\t" // out=((ABS(block[i])*qmat[0] - bias[0]*qmat[0])>>16)*sign(block[i])
|
|---|
| 181 | "movq %%mm0, (%5, %%"REG_a") \n\t"
|
|---|
| 182 | "pcmpeqw %%mm7, %%mm0 \n\t" // out==0 ? 0xFF : 0x00
|
|---|
| 183 | "movq (%4, %%"REG_a"), %%mm1 \n\t"
|
|---|
| 184 | "movq %%mm7, (%1, %%"REG_a") \n\t" // 0
|
|---|
| 185 | "pandn %%mm1, %%mm0 \n\t"
|
|---|
| 186 | PMAXW(%%mm0, %%mm3)
|
|---|
| 187 | "add $8, %%"REG_a" \n\t"
|
|---|
| 188 | " js 1b \n\t"
|
|---|
| 189 | PMAX(%%mm3, %%mm0)
|
|---|
| 190 | "movd %%mm3, %%"REG_a" \n\t"
|
|---|
| 191 | "movzb %%al, %%"REG_a" \n\t" // last_non_zero_p1
|
|---|
| 192 | : "+a" (last_non_zero_p1)
|
|---|
| 193 | : "r" (block+64), "r" (qmat+64), "r" (bias+64),
|
|---|
| 194 | "r" (inv_zigzag_direct16+64), "r" (temp_block+64)
|
|---|
| 195 | );
|
|---|
| 196 | // note the asm is split cuz gcc doesnt like that many operands ...
|
|---|
| 197 | asm volatile(
|
|---|
| 198 | "movd %1, %%mm1 \n\t" // max_qcoeff
|
|---|
| 199 | SPREADW(%%mm1)
|
|---|
| 200 | "psubusw %%mm1, %%mm4 \n\t"
|
|---|
| 201 | "packuswb %%mm4, %%mm4 \n\t"
|
|---|
| 202 | "movd %%mm4, %0 \n\t" // *overflow
|
|---|
| 203 | : "=g" (*overflow)
|
|---|
| 204 | : "g" (s->max_qcoeff)
|
|---|
| 205 | );
|
|---|
| 206 | }
|
|---|
| 207 |
|
|---|
| 208 | if(s->mb_intra) block[0]= level;
|
|---|
| 209 | else block[0]= temp_block[0];
|
|---|
| 210 |
|
|---|
| 211 | if(s->dsp.idct_permutation_type == FF_SIMPLE_IDCT_PERM){
|
|---|
| 212 | if(last_non_zero_p1 <= 1) goto end;
|
|---|
| 213 | block[0x08] = temp_block[0x01]; block[0x10] = temp_block[0x08];
|
|---|
| 214 | block[0x20] = temp_block[0x10];
|
|---|
| 215 | if(last_non_zero_p1 <= 4) goto end;
|
|---|
| 216 | block[0x18] = temp_block[0x09]; block[0x04] = temp_block[0x02];
|
|---|
| 217 | block[0x09] = temp_block[0x03];
|
|---|
| 218 | if(last_non_zero_p1 <= 7) goto end;
|
|---|
| 219 | block[0x14] = temp_block[0x0A]; block[0x28] = temp_block[0x11];
|
|---|
| 220 | block[0x12] = temp_block[0x18]; block[0x02] = temp_block[0x20];
|
|---|
| 221 | if(last_non_zero_p1 <= 11) goto end;
|
|---|
| 222 | block[0x1A] = temp_block[0x19]; block[0x24] = temp_block[0x12];
|
|---|
| 223 | block[0x19] = temp_block[0x0B]; block[0x01] = temp_block[0x04];
|
|---|
| 224 | block[0x0C] = temp_block[0x05];
|
|---|
| 225 | if(last_non_zero_p1 <= 16) goto end;
|
|---|
| 226 | block[0x11] = temp_block[0x0C]; block[0x29] = temp_block[0x13];
|
|---|
| 227 | block[0x16] = temp_block[0x1A]; block[0x0A] = temp_block[0x21];
|
|---|
| 228 | block[0x30] = temp_block[0x28]; block[0x22] = temp_block[0x30];
|
|---|
| 229 | block[0x38] = temp_block[0x29]; block[0x06] = temp_block[0x22];
|
|---|
| 230 | if(last_non_zero_p1 <= 24) goto end;
|
|---|
| 231 | block[0x1B] = temp_block[0x1B]; block[0x21] = temp_block[0x14];
|
|---|
| 232 | block[0x1C] = temp_block[0x0D]; block[0x05] = temp_block[0x06];
|
|---|
| 233 | block[0x0D] = temp_block[0x07]; block[0x15] = temp_block[0x0E];
|
|---|
| 234 | block[0x2C] = temp_block[0x15]; block[0x13] = temp_block[0x1C];
|
|---|
| 235 | if(last_non_zero_p1 <= 32) goto end;
|
|---|
| 236 | block[0x0B] = temp_block[0x23]; block[0x34] = temp_block[0x2A];
|
|---|
| 237 | block[0x2A] = temp_block[0x31]; block[0x32] = temp_block[0x38];
|
|---|
| 238 | block[0x3A] = temp_block[0x39]; block[0x26] = temp_block[0x32];
|
|---|
| 239 | block[0x39] = temp_block[0x2B]; block[0x03] = temp_block[0x24];
|
|---|
| 240 | if(last_non_zero_p1 <= 40) goto end;
|
|---|
| 241 | block[0x1E] = temp_block[0x1D]; block[0x25] = temp_block[0x16];
|
|---|
| 242 | block[0x1D] = temp_block[0x0F]; block[0x2D] = temp_block[0x17];
|
|---|
| 243 | block[0x17] = temp_block[0x1E]; block[0x0E] = temp_block[0x25];
|
|---|
| 244 | block[0x31] = temp_block[0x2C]; block[0x2B] = temp_block[0x33];
|
|---|
| 245 | if(last_non_zero_p1 <= 48) goto end;
|
|---|
| 246 | block[0x36] = temp_block[0x3A]; block[0x3B] = temp_block[0x3B];
|
|---|
| 247 | block[0x23] = temp_block[0x34]; block[0x3C] = temp_block[0x2D];
|
|---|
| 248 | block[0x07] = temp_block[0x26]; block[0x1F] = temp_block[0x1F];
|
|---|
| 249 | block[0x0F] = temp_block[0x27]; block[0x35] = temp_block[0x2E];
|
|---|
| 250 | if(last_non_zero_p1 <= 56) goto end;
|
|---|
| 251 | block[0x2E] = temp_block[0x35]; block[0x33] = temp_block[0x3C];
|
|---|
| 252 | block[0x3E] = temp_block[0x3D]; block[0x27] = temp_block[0x36];
|
|---|
| 253 | block[0x3D] = temp_block[0x2F]; block[0x2F] = temp_block[0x37];
|
|---|
| 254 | block[0x37] = temp_block[0x3E]; block[0x3F] = temp_block[0x3F];
|
|---|
| 255 | }else if(s->dsp.idct_permutation_type == FF_LIBMPEG2_IDCT_PERM){
|
|---|
| 256 | if(last_non_zero_p1 <= 1) goto end;
|
|---|
| 257 | block[0x04] = temp_block[0x01];
|
|---|
| 258 | block[0x08] = temp_block[0x08]; block[0x10] = temp_block[0x10];
|
|---|
| 259 | if(last_non_zero_p1 <= 4) goto end;
|
|---|
| 260 | block[0x0C] = temp_block[0x09]; block[0x01] = temp_block[0x02];
|
|---|
| 261 | block[0x05] = temp_block[0x03];
|
|---|
| 262 | if(last_non_zero_p1 <= 7) goto end;
|
|---|
| 263 | block[0x09] = temp_block[0x0A]; block[0x14] = temp_block[0x11];
|
|---|
| 264 | block[0x18] = temp_block[0x18]; block[0x20] = temp_block[0x20];
|
|---|
| 265 | if(last_non_zero_p1 <= 11) goto end;
|
|---|
| 266 | block[0x1C] = temp_block[0x19];
|
|---|
| 267 | block[0x11] = temp_block[0x12]; block[0x0D] = temp_block[0x0B];
|
|---|
| 268 | block[0x02] = temp_block[0x04]; block[0x06] = temp_block[0x05];
|
|---|
| 269 | if(last_non_zero_p1 <= 16) goto end;
|
|---|
| 270 | block[0x0A] = temp_block[0x0C]; block[0x15] = temp_block[0x13];
|
|---|
| 271 | block[0x19] = temp_block[0x1A]; block[0x24] = temp_block[0x21];
|
|---|
| 272 | block[0x28] = temp_block[0x28]; block[0x30] = temp_block[0x30];
|
|---|
| 273 | block[0x2C] = temp_block[0x29]; block[0x21] = temp_block[0x22];
|
|---|
| 274 | if(last_non_zero_p1 <= 24) goto end;
|
|---|
| 275 | block[0x1D] = temp_block[0x1B]; block[0x12] = temp_block[0x14];
|
|---|
| 276 | block[0x0E] = temp_block[0x0D]; block[0x03] = temp_block[0x06];
|
|---|
| 277 | block[0x07] = temp_block[0x07]; block[0x0B] = temp_block[0x0E];
|
|---|
| 278 | block[0x16] = temp_block[0x15]; block[0x1A] = temp_block[0x1C];
|
|---|
| 279 | if(last_non_zero_p1 <= 32) goto end;
|
|---|
| 280 | block[0x25] = temp_block[0x23]; block[0x29] = temp_block[0x2A];
|
|---|
| 281 | block[0x34] = temp_block[0x31]; block[0x38] = temp_block[0x38];
|
|---|
| 282 | block[0x3C] = temp_block[0x39]; block[0x31] = temp_block[0x32];
|
|---|
| 283 | block[0x2D] = temp_block[0x2B]; block[0x22] = temp_block[0x24];
|
|---|
| 284 | if(last_non_zero_p1 <= 40) goto end;
|
|---|
| 285 | block[0x1E] = temp_block[0x1D]; block[0x13] = temp_block[0x16];
|
|---|
| 286 | block[0x0F] = temp_block[0x0F]; block[0x17] = temp_block[0x17];
|
|---|
| 287 | block[0x1B] = temp_block[0x1E]; block[0x26] = temp_block[0x25];
|
|---|
| 288 | block[0x2A] = temp_block[0x2C]; block[0x35] = temp_block[0x33];
|
|---|
| 289 | if(last_non_zero_p1 <= 48) goto end;
|
|---|
| 290 | block[0x39] = temp_block[0x3A]; block[0x3D] = temp_block[0x3B];
|
|---|
| 291 | block[0x32] = temp_block[0x34]; block[0x2E] = temp_block[0x2D];
|
|---|
| 292 | block[0x23] = temp_block[0x26]; block[0x1F] = temp_block[0x1F];
|
|---|
| 293 | block[0x27] = temp_block[0x27]; block[0x2B] = temp_block[0x2E];
|
|---|
| 294 | if(last_non_zero_p1 <= 56) goto end;
|
|---|
| 295 | block[0x36] = temp_block[0x35]; block[0x3A] = temp_block[0x3C];
|
|---|
| 296 | block[0x3E] = temp_block[0x3D]; block[0x33] = temp_block[0x36];
|
|---|
| 297 | block[0x2F] = temp_block[0x2F]; block[0x37] = temp_block[0x37];
|
|---|
| 298 | block[0x3B] = temp_block[0x3E]; block[0x3F] = temp_block[0x3F];
|
|---|
| 299 | }else{
|
|---|
| 300 | if(last_non_zero_p1 <= 1) goto end;
|
|---|
| 301 | block[0x01] = temp_block[0x01];
|
|---|
| 302 | block[0x08] = temp_block[0x08]; block[0x10] = temp_block[0x10];
|
|---|
| 303 | if(last_non_zero_p1 <= 4) goto end;
|
|---|
| 304 | block[0x09] = temp_block[0x09]; block[0x02] = temp_block[0x02];
|
|---|
| 305 | block[0x03] = temp_block[0x03];
|
|---|
| 306 | if(last_non_zero_p1 <= 7) goto end;
|
|---|
| 307 | block[0x0A] = temp_block[0x0A]; block[0x11] = temp_block[0x11];
|
|---|
| 308 | block[0x18] = temp_block[0x18]; block[0x20] = temp_block[0x20];
|
|---|
| 309 | if(last_non_zero_p1 <= 11) goto end;
|
|---|
| 310 | block[0x19] = temp_block[0x19];
|
|---|
| 311 | block[0x12] = temp_block[0x12]; block[0x0B] = temp_block[0x0B];
|
|---|
| 312 | block[0x04] = temp_block[0x04]; block[0x05] = temp_block[0x05];
|
|---|
| 313 | if(last_non_zero_p1 <= 16) goto end;
|
|---|
| 314 | block[0x0C] = temp_block[0x0C]; block[0x13] = temp_block[0x13];
|
|---|
| 315 | block[0x1A] = temp_block[0x1A]; block[0x21] = temp_block[0x21];
|
|---|
| 316 | block[0x28] = temp_block[0x28]; block[0x30] = temp_block[0x30];
|
|---|
| 317 | block[0x29] = temp_block[0x29]; block[0x22] = temp_block[0x22];
|
|---|
| 318 | if(last_non_zero_p1 <= 24) goto end;
|
|---|
| 319 | block[0x1B] = temp_block[0x1B]; block[0x14] = temp_block[0x14];
|
|---|
| 320 | block[0x0D] = temp_block[0x0D]; block[0x06] = temp_block[0x06];
|
|---|
| 321 | block[0x07] = temp_block[0x07]; block[0x0E] = temp_block[0x0E];
|
|---|
| 322 | block[0x15] = temp_block[0x15]; block[0x1C] = temp_block[0x1C];
|
|---|
| 323 | if(last_non_zero_p1 <= 32) goto end;
|
|---|
| 324 | block[0x23] = temp_block[0x23]; block[0x2A] = temp_block[0x2A];
|
|---|
| 325 | block[0x31] = temp_block[0x31]; block[0x38] = temp_block[0x38];
|
|---|
| 326 | block[0x39] = temp_block[0x39]; block[0x32] = temp_block[0x32];
|
|---|
| 327 | block[0x2B] = temp_block[0x2B]; block[0x24] = temp_block[0x24];
|
|---|
| 328 | if(last_non_zero_p1 <= 40) goto end;
|
|---|
| 329 | block[0x1D] = temp_block[0x1D]; block[0x16] = temp_block[0x16];
|
|---|
| 330 | block[0x0F] = temp_block[0x0F]; block[0x17] = temp_block[0x17];
|
|---|
| 331 | block[0x1E] = temp_block[0x1E]; block[0x25] = temp_block[0x25];
|
|---|
| 332 | block[0x2C] = temp_block[0x2C]; block[0x33] = temp_block[0x33];
|
|---|
| 333 | if(last_non_zero_p1 <= 48) goto end;
|
|---|
| 334 | block[0x3A] = temp_block[0x3A]; block[0x3B] = temp_block[0x3B];
|
|---|
| 335 | block[0x34] = temp_block[0x34]; block[0x2D] = temp_block[0x2D];
|
|---|
| 336 | block[0x26] = temp_block[0x26]; block[0x1F] = temp_block[0x1F];
|
|---|
| 337 | block[0x27] = temp_block[0x27]; block[0x2E] = temp_block[0x2E];
|
|---|
| 338 | if(last_non_zero_p1 <= 56) goto end;
|
|---|
| 339 | block[0x35] = temp_block[0x35]; block[0x3C] = temp_block[0x3C];
|
|---|
| 340 | block[0x3D] = temp_block[0x3D]; block[0x36] = temp_block[0x36];
|
|---|
| 341 | block[0x2F] = temp_block[0x2F]; block[0x37] = temp_block[0x37];
|
|---|
| 342 | block[0x3E] = temp_block[0x3E]; block[0x3F] = temp_block[0x3F];
|
|---|
| 343 | }
|
|---|
| 344 | end:
|
|---|
| 345 | /*
|
|---|
| 346 | for(i=0; i<last_non_zero_p1; i++)
|
|---|
| 347 | {
|
|---|
| 348 | int j= zigzag_direct_noperm[i];
|
|---|
| 349 | block[block_permute_op(j)]= temp_block[j];
|
|---|
| 350 | }
|
|---|
| 351 | */
|
|---|
| 352 |
|
|---|
| 353 | return last_non_zero_p1 - 1;
|
|---|
| 354 | }
|
|---|
