fdct_altivec.c 17.9 KB
Newer Older
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498
/* ffmpeg/libavcodec/ppc/fdct_altivec.c, this file is part of the
 * AltiVec optimized library for the FFMPEG Multimedia System
 * Copyright (C) 2003  James Klicman <james@klicman.org>
 *
 * This library 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.
 *
 * This library 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 this library; if not, write to the Free Software
 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
 */


#include "../common.h"
#include "../dsputil.h"
#include "dsputil_altivec.h"
#include "gcc_fixes.h"


#define vs16(v) ((vector signed short)(v))
#define vs32(v) ((vector signed int)(v))
#define vu8(v)  ((vector unsigned char)(v))
#define vu16(v) ((vector unsigned short)(v))
#define vu32(v) ((vector unsigned int)(v))


#define C1     0.98078525066375732421875000 /* cos(1*PI/16) */
#define C2     0.92387950420379638671875000 /* cos(2*PI/16) */
#define C3     0.83146959543228149414062500 /* cos(3*PI/16) */
#define C4     0.70710676908493041992187500 /* cos(4*PI/16) */
#define C5     0.55557024478912353515625000 /* cos(5*PI/16) */
#define C6     0.38268342614173889160156250 /* cos(6*PI/16) */
#define C7     0.19509032368659973144531250 /* cos(7*PI/16) */
#define SQRT_2 1.41421353816986083984375000 /* sqrt(2)      */


#define W0 -(2 * C2)
#define W1 (2 * C6)
#define W2 (SQRT_2 * C6)
#define W3 (SQRT_2 * C3)
#define W4 (SQRT_2 * (-C1 + C3 + C5 - C7))
#define W5 (SQRT_2 * ( C1 + C3 - C5 + C7))
#define W6 (SQRT_2 * ( C1 + C3 + C5 - C7))
#define W7 (SQRT_2 * ( C1 + C3 - C5 - C7))
#define W8 (SQRT_2 * ( C7 - C3))
#define W9 (SQRT_2 * (-C1 - C3))
#define WA (SQRT_2 * (-C3 - C5))
#define WB (SQRT_2 * ( C5 - C3))


static vector float fdctconsts[3] = {
    (vector float)AVV( W0, W1, W2, W3 ),
    (vector float)AVV( W4, W5, W6, W7 ),
    (vector float)AVV( W8, W9, WA, WB )
};

#define LD_W0 vec_splat(cnsts0, 0)
#define LD_W1 vec_splat(cnsts0, 1)
#define LD_W2 vec_splat(cnsts0, 2)
#define LD_W3 vec_splat(cnsts0, 3)
#define LD_W4 vec_splat(cnsts1, 0)
#define LD_W5 vec_splat(cnsts1, 1)
#define LD_W6 vec_splat(cnsts1, 2)
#define LD_W7 vec_splat(cnsts1, 3)
#define LD_W8 vec_splat(cnsts2, 0)
#define LD_W9 vec_splat(cnsts2, 1)
#define LD_WA vec_splat(cnsts2, 2)
#define LD_WB vec_splat(cnsts2, 3)


#define FDCTROW(b0,b1,b2,b3,b4,b5,b6,b7) /* {{{ */                  \
    x0 = vec_add(b0, b7);               /* x0 = b0 + b7; */         \
    x7 = vec_sub(b0, b7);               /* x7 = b0 - b7; */         \
    x1 = vec_add(b1, b6);               /* x1 = b1 + b6; */         \
    x6 = vec_sub(b1, b6);               /* x6 = b1 - b6; */         \
    x2 = vec_add(b2, b5);               /* x2 = b2 + b5; */         \
    x5 = vec_sub(b2, b5);               /* x5 = b2 - b5; */         \
    x3 = vec_add(b3, b4);               /* x3 = b3 + b4; */         \
    x4 = vec_sub(b3, b4);               /* x4 = b3 - b4; */         \
                                                                    \
    b7 = vec_add(x0, x3);               /* b7 = x0 + x3; */         \
    b1 = vec_add(x1, x2);               /* b1 = x1 + x2; */         \
    b0 = vec_add(b7, b1);               /* b0 = b7 + b1; */         \
    b4 = vec_sub(b7, b1);               /* b4 = b7 - b1; */         \
                                                                    \
    b2 = vec_sub(x0, x3);               /* b2 = x0 - x3; */         \
    b6 = vec_sub(x1, x2);               /* b6 = x1 - x2; */         \
    b5 = vec_add(b6, b2);               /* b5 = b6 + b2; */         \
    cnst = LD_W2;                                                   \
    b5 = vec_madd(cnst, b5, mzero);     /* b5 = b5 * W2; */         \
    cnst = LD_W1;                                                   \
    b2 = vec_madd(cnst, b2, b5);        /* b2 = b5 + b2 * W1; */    \
    cnst = LD_W0;                                                   \
    b6 = vec_madd(cnst, b6, b5);        /* b6 = b5 + b6 * W0; */    \
                                                                    \
    x0 = vec_add(x4, x7);               /* x0 = x4 + x7; */         \
    x1 = vec_add(x5, x6);               /* x1 = x5 + x6; */         \
    x2 = vec_add(x4, x6);               /* x2 = x4 + x6; */         \
    x3 = vec_add(x5, x7);               /* x3 = x5 + x7; */         \
    x8 = vec_add(x2, x3);               /* x8 = x2 + x3; */         \
    cnst = LD_W3;                                                   \
    x8 = vec_madd(cnst, x8, mzero);     /* x8 = x8 * W3; */         \
                                                                    \
    cnst = LD_W8;                                                   \
    x0 = vec_madd(cnst, x0, mzero);     /* x0 *= W8; */             \
    cnst = LD_W9;                                                   \
    x1 = vec_madd(cnst, x1, mzero);     /* x1 *= W9; */             \
    cnst = LD_WA;                                                   \
    x2 = vec_madd(cnst, x2, x8);        /* x2 = x2 * WA + x8; */    \
    cnst = LD_WB;                                                   \
    x3 = vec_madd(cnst, x3, x8);        /* x3 = x3 * WB + x8; */    \
                                                                    \
    cnst = LD_W4;                                                   \
    b7 = vec_madd(cnst, x4, x0);        /* b7 = x4 * W4 + x0; */    \
    cnst = LD_W5;                                                   \
    b5 = vec_madd(cnst, x5, x1);        /* b5 = x5 * W5 + x1; */    \
    cnst = LD_W6;                                                   \
    b3 = vec_madd(cnst, x6, x1);        /* b3 = x6 * W6 + x1; */    \
    cnst = LD_W7;                                                   \
    b1 = vec_madd(cnst, x7, x0);        /* b1 = x7 * W7 + x0; */    \
                                                                    \
    b7 = vec_add(b7, x2);               /* b7 = b7 + x2; */         \
    b5 = vec_add(b5, x3);               /* b5 = b5 + x3; */         \
    b3 = vec_add(b3, x2);               /* b3 = b3 + x2; */         \
    b1 = vec_add(b1, x3);               /* b1 = b1 + x3; */         \
    /* }}} */

#define FDCTCOL(b0,b1,b2,b3,b4,b5,b6,b7) /* {{{ */                  \
    x0 = vec_add(b0, b7);               /* x0 = b0 + b7; */         \
    x7 = vec_sub(b0, b7);               /* x7 = b0 - b7; */         \
    x1 = vec_add(b1, b6);               /* x1 = b1 + b6; */         \
    x6 = vec_sub(b1, b6);               /* x6 = b1 - b6; */         \
    x2 = vec_add(b2, b5);               /* x2 = b2 + b5; */         \
    x5 = vec_sub(b2, b5);               /* x5 = b2 - b5; */         \
    x3 = vec_add(b3, b4);               /* x3 = b3 + b4; */         \
    x4 = vec_sub(b3, b4);               /* x4 = b3 - b4; */         \
                                                                    \
    b7 = vec_add(x0, x3);               /* b7 = x0 + x3; */         \
    b1 = vec_add(x1, x2);               /* b1 = x1 + x2; */         \
    b0 = vec_add(b7, b1);               /* b0 = b7 + b1; */         \
    b4 = vec_sub(b7, b1);               /* b4 = b7 - b1; */         \
                                                                    \
    b2 = vec_sub(x0, x3);               /* b2 = x0 - x3; */         \
    b6 = vec_sub(x1, x2);               /* b6 = x1 - x2; */         \
    b5 = vec_add(b6, b2);               /* b5 = b6 + b2; */         \
    cnst = LD_W2;                                                   \
    b5 = vec_madd(cnst, b5, mzero);     /* b5 = b5 * W2; */         \
    cnst = LD_W1;                                                   \
    b2 = vec_madd(cnst, b2, b5);        /* b2 = b5 + b2 * W1; */    \
    cnst = LD_W0;                                                   \
    b6 = vec_madd(cnst, b6, b5);        /* b6 = b5 + b6 * W0; */    \
                                                                    \
    x0 = vec_add(x4, x7);               /* x0 = x4 + x7; */         \
    x1 = vec_add(x5, x6);               /* x1 = x5 + x6; */         \
    x2 = vec_add(x4, x6);               /* x2 = x4 + x6; */         \
    x3 = vec_add(x5, x7);               /* x3 = x5 + x7; */         \
    x8 = vec_add(x2, x3);               /* x8 = x2 + x3; */         \
    cnst = LD_W3;                                                   \
    x8 = vec_madd(cnst, x8, mzero);     /* x8 = x8 * W3; */         \
                                                                    \
    cnst = LD_W8;                                                   \
    x0 = vec_madd(cnst, x0, mzero);     /* x0 *= W8; */             \
    cnst = LD_W9;                                                   \
    x1 = vec_madd(cnst, x1, mzero);     /* x1 *= W9; */             \
    cnst = LD_WA;                                                   \
    x2 = vec_madd(cnst, x2, x8);        /* x2 = x2 * WA + x8; */    \
    cnst = LD_WB;                                                   \
    x3 = vec_madd(cnst, x3, x8);        /* x3 = x3 * WB + x8; */    \
                                                                    \
    cnst = LD_W4;                                                   \
    b7 = vec_madd(cnst, x4, x0);        /* b7 = x4 * W4 + x0; */    \
    cnst = LD_W5;                                                   \
    b5 = vec_madd(cnst, x5, x1);        /* b5 = x5 * W5 + x1; */    \
    cnst = LD_W6;                                                   \
    b3 = vec_madd(cnst, x6, x1);        /* b3 = x6 * W6 + x1; */    \
    cnst = LD_W7;                                                   \
    b1 = vec_madd(cnst, x7, x0);        /* b1 = x7 * W7 + x0; */    \
                                                                    \
    b7 = vec_add(b7, x2);               /* b7 += x2; */             \
    b5 = vec_add(b5, x3);               /* b5 += x3; */             \
    b3 = vec_add(b3, x2);               /* b3 += x2; */             \
    b1 = vec_add(b1, x3);               /* b1 += x3; */             \
    /* }}} */



/* two dimensional discrete cosine transform */

void fdct_altivec(int16_t *block)
{
POWERPC_PERF_DECLARE(altivec_fdct, 1);
#ifdef ALTIVEC_USE_REFERENCE_C_CODE
POWERPC_PERF_START_COUNT(altivec_fdct, 1);
    void ff_jpeg_fdct_islow(int16_t *block);
    ff_jpeg_fdct_islow(block);
POWERPC_PERF_STOP_COUNT(altivec_fdct, 1);
#else /* ALTIVEC_USE_REFERENCE_C_CODE */
    vector signed short *bp;
    vector float *cp;
    vector float b00, b10, b20, b30, b40, b50, b60, b70;
    vector float b01, b11, b21, b31, b41, b51, b61, b71;
    vector float mzero, cnst, cnsts0, cnsts1, cnsts2;
    vector float x0, x1, x2, x3, x4, x5, x6, x7, x8;

    POWERPC_PERF_START_COUNT(altivec_fdct, 1);


    /* setup constants {{{ */
    /* mzero = -0.0 */
    vu32(mzero) = vec_splat_u32(-1);
    vu32(mzero) = vec_sl(vu32(mzero), vu32(mzero));
    cp = fdctconsts;
    cnsts0 = vec_ld(0, cp); cp++;
    cnsts1 = vec_ld(0, cp); cp++;
    cnsts2 = vec_ld(0, cp);
    /* }}} */


    /* 8x8 matrix transpose (vector short[8]) {{{ */
#define MERGE_S16(hl,a,b) vec_merge##hl(vs16(a), vs16(b))

    bp = (vector signed short*)block;
    vs16(b00) = vec_ld(0,    bp);
    vs16(b40) = vec_ld(16*4, bp);
    vs16(b01) = MERGE_S16(h, b00, b40);
    vs16(b11) = MERGE_S16(l, b00, b40);
    bp++;
    vs16(b10) = vec_ld(0,    bp);
    vs16(b50) = vec_ld(16*4, bp);
    vs16(b21) = MERGE_S16(h, b10, b50);
    vs16(b31) = MERGE_S16(l, b10, b50);
    bp++;
    vs16(b20) = vec_ld(0,    bp);
    vs16(b60) = vec_ld(16*4, bp);
    vs16(b41) = MERGE_S16(h, b20, b60);
    vs16(b51) = MERGE_S16(l, b20, b60);
    bp++;
    vs16(b30) = vec_ld(0,    bp);
    vs16(b70) = vec_ld(16*4, bp);
    vs16(b61) = MERGE_S16(h, b30, b70);
    vs16(b71) = MERGE_S16(l, b30, b70);

    vs16(x0) = MERGE_S16(h, b01, b41);
    vs16(x1) = MERGE_S16(l, b01, b41);
    vs16(x2) = MERGE_S16(h, b11, b51);
    vs16(x3) = MERGE_S16(l, b11, b51);
    vs16(x4) = MERGE_S16(h, b21, b61);
    vs16(x5) = MERGE_S16(l, b21, b61);
    vs16(x6) = MERGE_S16(h, b31, b71);
    vs16(x7) = MERGE_S16(l, b31, b71);

    vs16(b00) = MERGE_S16(h, x0, x4);
    vs16(b10) = MERGE_S16(l, x0, x4);
    vs16(b20) = MERGE_S16(h, x1, x5);
    vs16(b30) = MERGE_S16(l, x1, x5);
    vs16(b40) = MERGE_S16(h, x2, x6);
    vs16(b50) = MERGE_S16(l, x2, x6);
    vs16(b60) = MERGE_S16(h, x3, x7);
    vs16(b70) = MERGE_S16(l, x3, x7);

#undef MERGE_S16
    /* }}} */


/* Some of the initial calculations can be done as vector short before
 * conversion to vector float.  The following code section takes advantage
 * of this.
 */
#if 1
    /* fdct rows {{{ */
    vs16(x0) = vec_add(vs16(b00), vs16(b70));
    vs16(x7) = vec_sub(vs16(b00), vs16(b70));
    vs16(x1) = vec_add(vs16(b10), vs16(b60));
    vs16(x6) = vec_sub(vs16(b10), vs16(b60));
    vs16(x2) = vec_add(vs16(b20), vs16(b50));
    vs16(x5) = vec_sub(vs16(b20), vs16(b50));
    vs16(x3) = vec_add(vs16(b30), vs16(b40));
    vs16(x4) = vec_sub(vs16(b30), vs16(b40));

    vs16(b70) = vec_add(vs16(x0), vs16(x3));
    vs16(b10) = vec_add(vs16(x1), vs16(x2));

    vs16(b00) = vec_add(vs16(b70), vs16(b10));
    vs16(b40) = vec_sub(vs16(b70), vs16(b10));

#define CTF0(n) \
    vs32(b##n##1) = vec_unpackl(vs16(b##n##0)); \
    vs32(b##n##0) = vec_unpackh(vs16(b##n##0)); \
    b##n##1 = vec_ctf(vs32(b##n##1), 0); \
    b##n##0 = vec_ctf(vs32(b##n##0), 0);

    CTF0(0);
    CTF0(4);

    vs16(b20) = vec_sub(vs16(x0), vs16(x3));
    vs16(b60) = vec_sub(vs16(x1), vs16(x2));

    CTF0(2);
    CTF0(6);

#undef CTF0

    x0 = vec_add(b60, b20);
    x1 = vec_add(b61, b21);

    cnst = LD_W2;
    x0 = vec_madd(cnst, x0, mzero);
    x1 = vec_madd(cnst, x1, mzero);
    cnst = LD_W1;
    b20 = vec_madd(cnst, b20, x0);
    b21 = vec_madd(cnst, b21, x1);
    cnst = LD_W0;
    b60 = vec_madd(cnst, b60, x0);
    b61 = vec_madd(cnst, b61, x1);

#define CTFX(x,b) \
    vs32(b##0) = vec_unpackh(vs16(x)); \
    vs32(b##1) = vec_unpackl(vs16(x)); \
    b##0 = vec_ctf(vs32(b##0), 0); \
    b##1 = vec_ctf(vs32(b##1), 0); \

    CTFX(x4, b7);
    CTFX(x5, b5);
    CTFX(x6, b3);
    CTFX(x7, b1);

#undef CTFX


    x0 = vec_add(b70, b10);
    x1 = vec_add(b50, b30);
    x2 = vec_add(b70, b30);
    x3 = vec_add(b50, b10);
    x8 = vec_add(x2, x3);
    cnst = LD_W3;
    x8 = vec_madd(cnst, x8, mzero);

    cnst = LD_W8;
    x0 = vec_madd(cnst, x0, mzero);
    cnst = LD_W9;
    x1 = vec_madd(cnst, x1, mzero);
    cnst = LD_WA;
    x2 = vec_madd(cnst, x2, x8);
    cnst = LD_WB;
    x3 = vec_madd(cnst, x3, x8);

    cnst = LD_W4;
    b70 = vec_madd(cnst, b70, x0);
    cnst = LD_W5;
    b50 = vec_madd(cnst, b50, x1);
    cnst = LD_W6;
    b30 = vec_madd(cnst, b30, x1);
    cnst = LD_W7;
    b10 = vec_madd(cnst, b10, x0);

    b70 = vec_add(b70, x2);
    b50 = vec_add(b50, x3);
    b30 = vec_add(b30, x2);
    b10 = vec_add(b10, x3);


    x0 = vec_add(b71, b11);
    x1 = vec_add(b51, b31);
    x2 = vec_add(b71, b31);
    x3 = vec_add(b51, b11);
    x8 = vec_add(x2, x3);
    cnst = LD_W3;
    x8 = vec_madd(cnst, x8, mzero);

    cnst = LD_W8;
    x0 = vec_madd(cnst, x0, mzero);
    cnst = LD_W9;
    x1 = vec_madd(cnst, x1, mzero);
    cnst = LD_WA;
    x2 = vec_madd(cnst, x2, x8);
    cnst = LD_WB;
    x3 = vec_madd(cnst, x3, x8);

    cnst = LD_W4;
    b71 = vec_madd(cnst, b71, x0);
    cnst = LD_W5;
    b51 = vec_madd(cnst, b51, x1);
    cnst = LD_W6;
    b31 = vec_madd(cnst, b31, x1);
    cnst = LD_W7;
    b11 = vec_madd(cnst, b11, x0);

    b71 = vec_add(b71, x2);
    b51 = vec_add(b51, x3);
    b31 = vec_add(b31, x2);
    b11 = vec_add(b11, x3);
    /* }}} */
#else
    /* convert to float {{{ */
#define CTF(n) \
    vs32(b##n##1) = vec_unpackl(vs16(b##n##0)); \
    vs32(b##n##0) = vec_unpackh(vs16(b##n##0)); \
    b##n##1 = vec_ctf(vs32(b##n##1), 0); \
    b##n##0 = vec_ctf(vs32(b##n##0), 0); \

    CTF(0);
    CTF(1);
    CTF(2);
    CTF(3);
    CTF(4);
    CTF(5);
    CTF(6);
    CTF(7);

#undef CTF
    /* }}} */

    FDCTROW(b00, b10, b20, b30, b40, b50, b60, b70);
    FDCTROW(b01, b11, b21, b31, b41, b51, b61, b71);
#endif


    /* 8x8 matrix transpose (vector float[8][2]) {{{ */
    x0 = vec_mergel(b00, b20);
    x1 = vec_mergeh(b00, b20);
    x2 = vec_mergel(b10, b30);
    x3 = vec_mergeh(b10, b30);

    b00 = vec_mergeh(x1, x3);
    b10 = vec_mergel(x1, x3);
    b20 = vec_mergeh(x0, x2);
    b30 = vec_mergel(x0, x2);

    x4 = vec_mergel(b41, b61);
    x5 = vec_mergeh(b41, b61);
    x6 = vec_mergel(b51, b71);
    x7 = vec_mergeh(b51, b71);

    b41 = vec_mergeh(x5, x7);
    b51 = vec_mergel(x5, x7);
    b61 = vec_mergeh(x4, x6);
    b71 = vec_mergel(x4, x6);

    x0 = vec_mergel(b01, b21);
    x1 = vec_mergeh(b01, b21);
    x2 = vec_mergel(b11, b31);
    x3 = vec_mergeh(b11, b31);

    x4 = vec_mergel(b40, b60);
    x5 = vec_mergeh(b40, b60);
    x6 = vec_mergel(b50, b70);
    x7 = vec_mergeh(b50, b70);

    b40 = vec_mergeh(x1, x3);
    b50 = vec_mergel(x1, x3);
    b60 = vec_mergeh(x0, x2);
    b70 = vec_mergel(x0, x2);

    b01 = vec_mergeh(x5, x7);
    b11 = vec_mergel(x5, x7);
    b21 = vec_mergeh(x4, x6);
    b31 = vec_mergel(x4, x6);
    /* }}} */


    FDCTCOL(b00, b10, b20, b30, b40, b50, b60, b70);
    FDCTCOL(b01, b11, b21, b31, b41, b51, b61, b71);


    /* round, convert back to short {{{ */
#define CTS(n) \
    b##n##0 = vec_round(b##n##0); \
    b##n##1 = vec_round(b##n##1); \
    vs32(b##n##0) = vec_cts(b##n##0, 0); \
    vs32(b##n##1) = vec_cts(b##n##1, 0); \
    vs16(b##n##0) = vec_pack(vs32(b##n##0), vs32(b##n##1)); \
    vec_st(vs16(b##n##0), 0, bp);

    bp = (vector signed short*)block;
    CTS(0); bp++;
    CTS(1); bp++;
    CTS(2); bp++;
    CTS(3); bp++;
    CTS(4); bp++;
    CTS(5); bp++;
    CTS(6); bp++;
    CTS(7);

#undef CTS
    /* }}} */

POWERPC_PERF_STOP_COUNT(altivec_fdct, 1);
#endif /* ALTIVEC_USE_REFERENCE_C_CODE */
}

/* vim:set foldmethod=marker foldlevel=0: */