DXR will be turned off on Tuesday, December 29th. It will redirect to Searchfox.
See the announcement on Discourse.

DXR is a code search and navigation tool aimed at making sense of large projects. It supports full-text and regex searches as well as structural queries.

Mercurial (d38398e5144e)

VCS Links

Line Code
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 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 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218
/*
 *  Copyright (c) 2013 The WebM project authors. All Rights Reserved.
 *
 *  Use of this source code is governed by a BSD-style license
 *  that can be found in the LICENSE file in the root of the source
 *  tree. An additional intellectual property rights grant can be found
 *  in the file PATENTS.  All contributing project authors may
 *  be found in the AUTHORS file in the root of the source tree.
 */

#include <assert.h>
#include <stdio.h>

#include "./vpx_config.h"
#include "vpx_dsp/mips/inv_txfm_dspr2.h"
#include "vpx_dsp/txfm_common.h"

#if HAVE_DSPR2
static void idct32_rows_dspr2(const int16_t *input, int16_t *output,
                              uint32_t no_rows) {
  int step1_0, step1_1, step1_2, step1_3, step1_4, step1_5, step1_6;
  int step1_7, step1_8, step1_9, step1_10, step1_11, step1_12, step1_13;
  int step1_14, step1_15, step1_16, step1_17, step1_18, step1_19, step1_20;
  int step1_21, step1_22, step1_23, step1_24, step1_25, step1_26, step1_27;
  int step1_28, step1_29, step1_30, step1_31;
  int step2_0, step2_1, step2_2, step2_3, step2_4, step2_5, step2_6;
  int step2_7, step2_8, step2_9, step2_10, step2_11, step2_12, step2_13;
  int step2_14, step2_15, step2_16, step2_17, step2_18, step2_19, step2_20;
  int step2_21, step2_22, step2_23, step2_24, step2_25, step2_26, step2_27;
  int step2_28, step2_29, step2_30, step2_31;
  int step3_8, step3_9, step3_10, step3_11, step3_12, step3_13, step3_14;
  int step3_15, step3_16, step3_17, step3_18, step3_19, step3_20, step3_21;
  int step3_22, step3_23, step3_24, step3_25, step3_26, step3_27, step3_28;
  int step3_29, step3_30, step3_31;
  int temp0, temp1, temp2, temp3;
  int load1, load2, load3, load4;
  int result1, result2;
  int i;
  const int const_2_power_13 = 8192;
  const int32_t *input_int;

  for (i = no_rows; i--;) {
    input_int = (const int32_t *)input;

    if (!(input_int[0] | input_int[1] | input_int[2] | input_int[3] |
          input_int[4] | input_int[5] | input_int[6] | input_int[7] |
          input_int[8] | input_int[9] | input_int[10] | input_int[11] |
          input_int[12] | input_int[13] | input_int[14] | input_int[15])) {
      input += 32;

      __asm__ __volatile__(
          "sh     $zero,     0(%[output])     \n\t"
          "sh     $zero,    64(%[output])     \n\t"
          "sh     $zero,   128(%[output])     \n\t"
          "sh     $zero,   192(%[output])     \n\t"
          "sh     $zero,   256(%[output])     \n\t"
          "sh     $zero,   320(%[output])     \n\t"
          "sh     $zero,   384(%[output])     \n\t"
          "sh     $zero,   448(%[output])     \n\t"
          "sh     $zero,   512(%[output])     \n\t"
          "sh     $zero,   576(%[output])     \n\t"
          "sh     $zero,   640(%[output])     \n\t"
          "sh     $zero,   704(%[output])     \n\t"
          "sh     $zero,   768(%[output])     \n\t"
          "sh     $zero,   832(%[output])     \n\t"
          "sh     $zero,   896(%[output])     \n\t"
          "sh     $zero,   960(%[output])     \n\t"
          "sh     $zero,  1024(%[output])     \n\t"
          "sh     $zero,  1088(%[output])     \n\t"
          "sh     $zero,  1152(%[output])     \n\t"
          "sh     $zero,  1216(%[output])     \n\t"
          "sh     $zero,  1280(%[output])     \n\t"
          "sh     $zero,  1344(%[output])     \n\t"
          "sh     $zero,  1408(%[output])     \n\t"
          "sh     $zero,  1472(%[output])     \n\t"
          "sh     $zero,  1536(%[output])     \n\t"
          "sh     $zero,  1600(%[output])     \n\t"
          "sh     $zero,  1664(%[output])     \n\t"
          "sh     $zero,  1728(%[output])     \n\t"
          "sh     $zero,  1792(%[output])     \n\t"
          "sh     $zero,  1856(%[output])     \n\t"
          "sh     $zero,  1920(%[output])     \n\t"
          "sh     $zero,  1984(%[output])     \n\t"

          :
          : [output] "r"(output));

      output += 1;

      continue;
    }

    /* prefetch row */
    prefetch_load((const uint8_t *)(input + 32));
    prefetch_load((const uint8_t *)(input + 48));

    __asm__ __volatile__(
        "lh       %[load1],             2(%[input])                     \n\t"
        "lh       %[load2],             62(%[input])                    \n\t"
        "lh       %[load3],             34(%[input])                    \n\t"
        "lh       %[load4],             30(%[input])                    \n\t"

        "mtlo     %[const_2_power_13],  $ac1                            \n\t"
        "mthi     $zero,                $ac1                            \n\t"
        "mtlo     %[const_2_power_13],  $ac3                            \n\t"
        "mthi     $zero,                $ac3                            \n\t"

        "madd     $ac1,                 %[load1],       %[cospi_31_64]  \n\t"
        "msub     $ac1,                 %[load2],       %[cospi_1_64]   \n\t"
        "extp     %[temp0],             $ac1,           31              \n\t"

        "madd     $ac3,                 %[load1],       %[cospi_1_64]   \n\t"
        "madd     $ac3,                 %[load2],       %[cospi_31_64]  \n\t"
        "extp     %[temp3],             $ac3,           31              \n\t"

        "mtlo     %[const_2_power_13],  $ac1                            \n\t"
        "mthi     $zero,                $ac1                            \n\t"
        "mtlo     %[const_2_power_13],  $ac2                            \n\t"
        "mthi     $zero,                $ac2                            \n\t"

        "madd     $ac2,                 %[load3],       %[cospi_15_64]  \n\t"
        "msub     $ac2,                 %[load4],       %[cospi_17_64]  \n\t"
        "extp     %[temp1],             $ac2,           31              \n\t"

        "madd     $ac1,                 %[load3],       %[cospi_17_64]  \n\t"
        "madd     $ac1,                 %[load4],       %[cospi_15_64]  \n\t"
        "extp     %[temp2],             $ac1,           31              \n\t"

        "mtlo     %[const_2_power_13],  $ac1                            \n\t"
        "mthi     $zero,                $ac1                            \n\t"
        "mtlo     %[const_2_power_13],  $ac3                            \n\t"
        "mthi     $zero,                $ac3                            \n\t"

        "sub      %[load1],             %[temp3],       %[temp2]        \n\t"
        "sub      %[load2],             %[temp0],       %[temp1]        \n\t"

        "madd     $ac1,                 %[load1],       %[cospi_28_64]  \n\t"
        "msub     $ac1,                 %[load2],       %[cospi_4_64]   \n\t"
        "madd     $ac3,                 %[load1],       %[cospi_4_64]   \n\t"
        "madd     $ac3,                 %[load2],       %[cospi_28_64]  \n\t"

        "extp     %[step1_17],          $ac1,           31              \n\t"
        "extp     %[step1_30],          $ac3,           31              \n\t"
        "add      %[step1_16],          %[temp0],       %[temp1]        \n\t"
        "add      %[step1_31],          %[temp2],       %[temp3]        \n\t"

        : [load1] "=&r"(load1), [load2] "=&r"(load2), [load3] "=&r"(load3),
          [load4] "=&r"(load4), [temp0] "=&r"(temp0), [temp1] "=&r"(temp1),
          [temp2] "=&r"(temp2), [temp3] "=&r"(temp3),
          [step1_16] "=&r"(step1_16), [step1_17] "=&r"(step1_17),
          [step1_30] "=&r"(step1_30), [step1_31] "=&r"(step1_31)
        : [const_2_power_13] "r"(const_2_power_13), [input] "r"(input),
          [cospi_31_64] "r"(cospi_31_64), [cospi_1_64] "r"(cospi_1_64),
          [cospi_4_64] "r"(cospi_4_64), [cospi_17_64] "r"(cospi_17_64),
          [cospi_15_64] "r"(cospi_15_64), [cospi_28_64] "r"(cospi_28_64));

    __asm__ __volatile__(
        "lh       %[load1],             18(%[input])                    \n\t"
        "lh       %[load2],             46(%[input])                    \n\t"
        "lh       %[load3],             50(%[input])                    \n\t"
        "lh       %[load4],             14(%[input])                    \n\t"

        "mtlo     %[const_2_power_13],  $ac1                            \n\t"
        "mthi     $zero,                $ac1                            \n\t"
        "mtlo     %[const_2_power_13],  $ac3                            \n\t"
        "mthi     $zero,                $ac3                            \n\t"

        "madd     $ac1,                 %[load1],       %[cospi_23_64]  \n\t"
        "msub     $ac1,                 %[load2],       %[cospi_9_64]   \n\t"
        "extp     %[temp0],             $ac1,           31              \n\t"

        "madd     $ac3,                 %[load1],       %[cospi_9_64]   \n\t"
        "madd     $ac3,                 %[load2],       %[cospi_23_64]  \n\t"
        "extp     %[temp3],             $ac3,           31              \n\t"

        "mtlo     %[const_2_power_13],  $ac1                            \n\t"
        "mthi     $zero,                $ac1                            \n\t"
        "mtlo     %[const_2_power_13],  $ac2                            \n\t"
        "mthi     $zero,                $ac2                            \n\t"

        "madd     $ac2,                 %[load3],       %[cospi_7_64]   \n\t"
        "msub     $ac2,                 %[load4],       %[cospi_25_64]  \n\t"
        "extp     %[temp1],             $ac2,           31              \n\t"

        "madd     $ac1,                 %[load3],       %[cospi_25_64]  \n\t"
        "madd     $ac1,                 %[load4],       %[cospi_7_64]   \n\t"
        "extp     %[temp2],             $ac1,           31              \n\t"

        "mtlo     %[const_2_power_13],  $ac1                            \n\t"
        "mthi     $zero,                $ac1                            \n\t"
        "mtlo     %[const_2_power_13],  $ac3                            \n\t"
        "mthi     $zero,                $ac3                            \n\t"

        "sub      %[load1],             %[temp1],       %[temp0]        \n\t"
        "sub      %[load2],             %[temp2],       %[temp3]        \n\t"

        "msub     $ac1,                 %[load1],       %[cospi_28_64]  \n\t"
        "msub     $ac1,                 %[load2],       %[cospi_4_64]   \n\t"
        "msub     $ac3,                 %[load1],       %[cospi_4_64]   \n\t"
        "madd     $ac3,                 %[load2],       %[cospi_28_64]  \n\t"

        "extp     %[step1_18],          $ac1,           31              \n\t"
        "extp     %[step1_29],          $ac3,           31              \n\t"
        "add      %[step1_19],          %[temp0],       %[temp1]        \n\t"
        "add      %[step1_28],          %[temp2],       %[temp3]        \n\t"

        : [load1] "=&r"(load1), [load2] "=&r"(load2), [load3] "=&r"(load3),
          [load4] "=&r"(load4), [temp0] "=&r"(temp0), [temp1] "=&r"(temp1),
          [temp2] "=&r"(temp2), [temp3] "=&r"(temp3),
          [step1_18] "=&r"(step1_18), [step1_19] "=&r"(step1_19),
          [step1_28] "=&r"(step1_28), [step1_29] "=&r"(step1_29)
        : [const_2_power_13] "r"(const_2_power_13), [input] "r"(input),
          [cospi_23_64] "r"(cospi_23_64), [cospi_9_64] "r"(cospi_9_64),
          [cospi_4_64] "r"(cospi_4_64), [cospi_7_64] "r"(cospi_7_64),
          [cospi_25_64] "r"(cospi_25_64), [cospi_28_64] "r"(cospi_28_64));

    __asm__ __volatile__(
        "lh       %[load1],             10(%[input])                    \n\t"
        "lh       %[load2],             54(%[input])                    \n\t"
        "lh       %[load3],             42(%[input])                    \n\t"
        "lh       %[load4],             22(%[input])                    \n\t"

        "mtlo     %[const_2_power_13],  $ac1                            \n\t"
        "mthi     $zero,                $ac1                            \n\t"
        "mtlo     %[const_2_power_13],  $ac3                            \n\t"
        "mthi     $zero,                $ac3                            \n\t"

        "madd     $ac1,                 %[load1],       %[cospi_27_64]  \n\t"
        "msub     $ac1,                 %[load2],       %[cospi_5_64]   \n\t"
        "extp     %[temp0],             $ac1,           31              \n\t"

        "madd     $ac3,                 %[load1],       %[cospi_5_64]   \n\t"
        "madd     $ac3,                 %[load2],       %[cospi_27_64]  \n\t"
        "extp     %[temp3],             $ac3,           31              \n\t"

        "mtlo     %[const_2_power_13],  $ac1                            \n\t"
        "mthi     $zero,                $ac1                            \n\t"
        "mtlo     %[const_2_power_13],  $ac2                            \n\t"
        "mthi     $zero,                $ac2                            \n\t"

        "madd     $ac2,                 %[load3],       %[cospi_11_64]  \n\t"
        "msub     $ac2,                 %[load4],       %[cospi_21_64]  \n\t"
        "extp     %[temp1],             $ac2,           31              \n\t"

        "madd     $ac1,                 %[load3],       %[cospi_21_64]  \n\t"
        "madd     $ac1,                 %[load4],       %[cospi_11_64]  \n\t"
        "extp     %[temp2],             $ac1,           31              \n\t"

        "mtlo     %[const_2_power_13],  $ac1                            \n\t"
        "mthi     $zero,                $ac1                            \n\t"
        "mtlo     %[const_2_power_13],  $ac3                            \n\t"
        "mthi     $zero,                $ac3                            \n\t"

        "sub      %[load1],             %[temp0],       %[temp1]        \n\t"
        "sub      %[load2],             %[temp3],       %[temp2]        \n\t"

        "madd     $ac1,                 %[load2],       %[cospi_12_64]  \n\t"
        "msub     $ac1,                 %[load1],       %[cospi_20_64]  \n\t"
        "madd     $ac3,                 %[load1],       %[cospi_12_64]  \n\t"
        "madd     $ac3,                 %[load2],       %[cospi_20_64]  \n\t"

        "extp     %[step1_21],          $ac1,           31              \n\t"
        "extp     %[step1_26],          $ac3,           31              \n\t"
        "add      %[step1_20],          %[temp0],       %[temp1]        \n\t"
        "add      %[step1_27],          %[temp2],       %[temp3]        \n\t"

        : [load1] "=&r"(load1), [load2] "=&r"(load2), [load3] "=&r"(load3),
          [load4] "=&r"(load4), [temp0] "=&r"(temp0), [temp1] "=&r"(temp1),
          [temp2] "=&r"(temp2), [temp3] "=&r"(temp3),
          [step1_20] "=&r"(step1_20), [step1_21] "=&r"(step1_21),
          [step1_26] "=&r"(step1_26), [step1_27] "=&r"(step1_27)
        : [const_2_power_13] "r"(const_2_power_13), [input] "r"(input),
          [cospi_27_64] "r"(cospi_27_64), [cospi_5_64] "r"(cospi_5_64),
          [cospi_11_64] "r"(cospi_11_64), [cospi_21_64] "r"(cospi_21_64),
          [cospi_12_64] "r"(cospi_12_64), [cospi_20_64] "r"(cospi_20_64));

    __asm__ __volatile__(
        "lh       %[load1],             26(%[input])                    \n\t"
        "lh       %[load2],             38(%[input])                    \n\t"
        "lh       %[load3],             58(%[input])                    \n\t"
        "lh       %[load4],              6(%[input])                    \n\t"

        "mtlo     %[const_2_power_13],  $ac1                            \n\t"
        "mthi     $zero,                $ac1                            \n\t"
        "mtlo     %[const_2_power_13],  $ac3                            \n\t"
        "mthi     $zero,                $ac3                            \n\t"

        "madd     $ac1,                 %[load1],       %[cospi_19_64]  \n\t"
        "msub     $ac1,                 %[load2],       %[cospi_13_64]  \n\t"
        "extp     %[temp0],             $ac1,           31              \n\t"
        "madd     $ac3,                 %[load1],       %[cospi_13_64]  \n\t"
        "madd     $ac3,                 %[load2],       %[cospi_19_64]  \n\t"
        "extp     %[temp3],             $ac3,           31              \n\t"

        "mtlo     %[const_2_power_13],  $ac1                            \n\t"
        "mthi     $zero,                $ac1                            \n\t"
        "mtlo     %[const_2_power_13],  $ac2                            \n\t"
        "mthi     $zero,                $ac2                            \n\t"

        "madd     $ac2,                 %[load3],       %[cospi_3_64]   \n\t"
        "msub     $ac2,                 %[load4],       %[cospi_29_64]  \n\t"
        "extp     %[temp1],             $ac2,           31              \n\t"
        "madd     $ac1,                 %[load3],       %[cospi_29_64]  \n\t"
        "madd     $ac1,                 %[load4],       %[cospi_3_64]   \n\t"
        "extp     %[temp2],             $ac1,           31              \n\t"

        "mtlo     %[const_2_power_13],  $ac1                            \n\t"
        "mthi     $zero,                $ac1                            \n\t"
        "mtlo     %[const_2_power_13],  $ac3                            \n\t"
        "mthi     $zero,                $ac3                            \n\t"

        "sub      %[load1],             %[temp1],       %[temp0]        \n\t"
        "sub      %[load2],             %[temp2],       %[temp3]        \n\t"
        "msub     $ac1,                 %[load1],       %[cospi_12_64]  \n\t"
        "msub     $ac1,                 %[load2],       %[cospi_20_64]  \n\t"
        "msub     $ac3,                 %[load1],       %[cospi_20_64]  \n\t"
        "madd     $ac3,                 %[load2],       %[cospi_12_64]  \n\t"
        "extp     %[step1_22],          $ac1,           31              \n\t"
        "extp     %[step1_25],          $ac3,           31              \n\t"
        "add      %[step1_23],          %[temp0],       %[temp1]        \n\t"
        "add      %[step1_24],          %[temp2],       %[temp3]        \n\t"

        : [load1] "=&r"(load1), [load2] "=&r"(load2), [load3] "=&r"(load3),
          [load4] "=&r"(load4), [temp0] "=&r"(temp0), [temp1] "=&r"(temp1),
          [temp2] "=&r"(temp2), [temp3] "=&r"(temp3),
          [step1_22] "=&r"(step1_22), [step1_23] "=&r"(step1_23),
          [step1_24] "=&r"(step1_24), [step1_25] "=&r"(step1_25)
        : [const_2_power_13] "r"(const_2_power_13), [input] "r"(input),
          [cospi_19_64] "r"(cospi_19_64), [cospi_13_64] "r"(cospi_13_64),
          [cospi_3_64] "r"(cospi_3_64), [cospi_29_64] "r"(cospi_29_64),
          [cospi_12_64] "r"(cospi_12_64), [cospi_20_64] "r"(cospi_20_64));

    __asm__ __volatile__(
        "lh       %[load1],              4(%[input])                    \n\t"
        "lh       %[load2],             60(%[input])                    \n\t"
        "lh       %[load3],             36(%[input])                    \n\t"
        "lh       %[load4],             28(%[input])                    \n\t"

        "mtlo     %[const_2_power_13],  $ac1                            \n\t"
        "mthi     $zero,                $ac1                            \n\t"
        "mtlo     %[const_2_power_13],  $ac3                            \n\t"
        "mthi     $zero,                $ac3                            \n\t"

        "madd     $ac1,                 %[load1],       %[cospi_30_64]  \n\t"
        "msub     $ac1,                 %[load2],       %[cospi_2_64]   \n\t"
        "extp     %[temp0],             $ac1,           31              \n\t"
        "madd     $ac3,                 %[load1],       %[cospi_2_64]   \n\t"
        "madd     $ac3,                 %[load2],       %[cospi_30_64]  \n\t"
        "extp     %[temp3],             $ac3,           31              \n\t"

        "mtlo     %[const_2_power_13],  $ac1                            \n\t"
        "mthi     $zero,                $ac1                            \n\t"
        "mtlo     %[const_2_power_13],  $ac2                            \n\t"
        "mthi     $zero,                $ac2                            \n\t"

        "madd     $ac2,                 %[load3],       %[cospi_14_64]  \n\t"
        "msub     $ac2,                 %[load4],       %[cospi_18_64]  \n\t"
        "extp     %[temp1],             $ac2,           31              \n\t"
        "madd     $ac1,                 %[load3],       %[cospi_18_64]  \n\t"
        "madd     $ac1,                 %[load4],       %[cospi_14_64]  \n\t"
        "extp     %[temp2],             $ac1,           31              \n\t"

        "mtlo     %[const_2_power_13],  $ac1                            \n\t"
        "mthi     $zero,                $ac1                            \n\t"
        "mtlo     %[const_2_power_13],  $ac3                            \n\t"
        "mthi     $zero,                $ac3                            \n\t"

        "sub      %[load1],             %[temp0],       %[temp1]        \n\t"
        "sub      %[load2],             %[temp3],       %[temp2]        \n\t"
        "msub     $ac1,                 %[load1],       %[cospi_8_64]   \n\t"
        "madd     $ac1,                 %[load2],       %[cospi_24_64]  \n\t"
        "madd     $ac3,                 %[load1],       %[cospi_24_64]  \n\t"
        "madd     $ac3,                 %[load2],       %[cospi_8_64]   \n\t"
        "extp     %[step2_9],           $ac1,           31              \n\t"
        "extp     %[step2_14],          $ac3,           31              \n\t"
        "add      %[step2_8],           %[temp0],       %[temp1]        \n\t"
        "add      %[step2_15],          %[temp2],       %[temp3]        \n\t"

        : [load1] "=&r"(load1), [load2] "=&r"(load2), [load3] "=&r"(load3),
          [load4] "=&r"(load4), [temp0] "=&r"(temp0), [temp1] "=&r"(temp1),
          [temp2] "=&r"(temp2), [temp3] "=&r"(temp3), [step2_8] "=&r"(step2_8),
          [step2_9] "=&r"(step2_9), [step2_14] "=&r"(step2_14),
          [step2_15] "=&r"(step2_15)
        : [const_2_power_13] "r"(const_2_power_13), [input] "r"(input),
          [cospi_30_64] "r"(cospi_30_64), [cospi_2_64] "r"(cospi_2_64),
          [cospi_14_64] "r"(cospi_14_64), [cospi_18_64] "r"(cospi_18_64),
          [cospi_8_64] "r"(cospi_8_64), [cospi_24_64] "r"(cospi_24_64));

    __asm__ __volatile__(
        "lh       %[load1],             20(%[input])                    \n\t"
        "lh       %[load2],             44(%[input])                    \n\t"
        "lh       %[load3],             52(%[input])                    \n\t"
        "lh       %[load4],             12(%[input])                    \n\t"

        "mtlo     %[const_2_power_13],  $ac1                            \n\t"
        "mthi     $zero,                $ac1                            \n\t"
        "mtlo     %[const_2_power_13],  $ac3                            \n\t"
        "mthi     $zero,                $ac3                            \n\t"

        "madd     $ac1,                 %[load1],       %[cospi_22_64]  \n\t"
        "msub     $ac1,                 %[load2],       %[cospi_10_64]  \n\t"
        "extp     %[temp0],             $ac1,           31              \n\t"
        "madd     $ac3,                 %[load1],       %[cospi_10_64]  \n\t"
        "madd     $ac3,                 %[load2],       %[cospi_22_64]  \n\t"
        "extp     %[temp3],             $ac3,           31              \n\t"

        "mtlo     %[const_2_power_13],  $ac1                            \n\t"
        "mthi     $zero,                $ac1                            \n\t"
        "mtlo     %[const_2_power_13],  $ac2                            \n\t"
        "mthi     $zero,                $ac2                            \n\t"

        "madd     $ac2,                 %[load3],       %[cospi_6_64]   \n\t"
        "msub     $ac2,                 %[load4],       %[cospi_26_64]  \n\t"
        "extp     %[temp1],             $ac2,           31              \n\t"
        "madd     $ac1,                 %[load3],       %[cospi_26_64]  \n\t"
        "madd     $ac1,                 %[load4],       %[cospi_6_64]   \n\t"
        "extp     %[temp2],             $ac1,           31              \n\t"

        "mtlo     %[const_2_power_13],  $ac1                            \n\t"
        "mthi     $zero,                $ac1                            \n\t"
        "mtlo     %[const_2_power_13],  $ac3                            \n\t"
        "mthi     $zero,                $ac3                            \n\t"

        "sub      %[load1],             %[temp1],       %[temp0]        \n\t"
        "sub      %[load2],             %[temp2],       %[temp3]        \n\t"
        "msub     $ac1,                 %[load1],       %[cospi_24_64]  \n\t"
        "msub     $ac1,                 %[load2],       %[cospi_8_64]   \n\t"
        "madd     $ac3,                 %[load2],       %[cospi_24_64]  \n\t"
        "msub     $ac3,                 %[load1],       %[cospi_8_64]   \n\t"
        "extp     %[step2_10],          $ac1,           31              \n\t"
        "extp     %[step2_13],          $ac3,           31              \n\t"
        "add      %[step2_11],          %[temp0],       %[temp1]        \n\t"
        "add      %[step2_12],          %[temp2],       %[temp3]        \n\t"

        : [load1] "=&r"(load1), [load2] "=&r"(load2), [load3] "=&r"(load3),
          [load4] "=&r"(load4), [temp0] "=&r"(temp0), [temp1] "=&r"(temp1),
          [temp2] "=&r"(temp2), [temp3] "=&r"(temp3),
          [step2_10] "=&r"(step2_10), [step2_11] "=&r"(step2_11),
          [step2_12] "=&r"(step2_12), [step2_13] "=&r"(step2_13)
        : [const_2_power_13] "r"(const_2_power_13), [input] "r"(input),
          [cospi_22_64] "r"(cospi_22_64), [cospi_10_64] "r"(cospi_10_64),
          [cospi_6_64] "r"(cospi_6_64), [cospi_26_64] "r"(cospi_26_64),
          [cospi_8_64] "r"(cospi_8_64), [cospi_24_64] "r"(cospi_24_64));

    __asm__ __volatile__(
        "mtlo     %[const_2_power_13],  $ac0                            \n\t"
        "mthi     $zero,                $ac0                            \n\t"
        "sub      %[temp0],             %[step2_14],    %[step2_13]     \n\t"
        "sub      %[temp0],             %[temp0],       %[step2_9]      \n\t"
        "add      %[temp0],             %[temp0],       %[step2_10]     \n\t"
        "madd     $ac0,                 %[temp0],       %[cospi_16_64]  \n\t"
        "mtlo     %[const_2_power_13],  $ac1                            \n\t"
        "mthi     $zero,                $ac1                            \n\t"
        "sub      %[temp1],             %[step2_14],    %[step2_13]     \n\t"
        "add      %[temp1],             %[temp1],       %[step2_9]      \n\t"
        "sub      %[temp1],             %[temp1],       %[step2_10]     \n\t"
        "madd     $ac1,                 %[temp1],       %[cospi_16_64]  \n\t"
        "mtlo     %[const_2_power_13],  $ac2                            \n\t"
        "mthi     $zero,                $ac2                            \n\t"
        "sub      %[temp0],             %[step2_15],    %[step2_12]     \n\t"
        "sub      %[temp0],             %[temp0],       %[step2_8]      \n\t"
        "add      %[temp0],             %[temp0],       %[step2_11]     \n\t"
        "madd     $ac2,                 %[temp0],       %[cospi_16_64]  \n\t"
        "mtlo     %[const_2_power_13],  $ac3                            \n\t"
        "mthi     $zero,                $ac3                            \n\t"
        "sub      %[temp1],             %[step2_15],    %[step2_12]     \n\t"
        "add      %[temp1],             %[temp1],       %[step2_8]      \n\t"
        "sub      %[temp1],             %[temp1],       %[step2_11]     \n\t"
        "madd     $ac3,                 %[temp1],       %[cospi_16_64]  \n\t"

        "add      %[step3_8],           %[step2_8],     %[step2_11]     \n\t"
        "add      %[step3_9],           %[step2_9],     %[step2_10]     \n\t"
        "add      %[step3_14],          %[step2_13],    %[step2_14]     \n\t"
        "add      %[step3_15],          %[step2_12],    %[step2_15]     \n\t"
        "extp     %[step3_10],          $ac0,           31              \n\t"
        "extp     %[step3_13],          $ac1,           31              \n\t"
        "extp     %[step3_11],          $ac2,           31              \n\t"
        "extp     %[step3_12],          $ac3,           31              \n\t"

        : [temp0] "=&r"(temp0), [temp1] "=&r"(temp1), [step3_8] "=&r"(step3_8),
          [step3_9] "=&r"(step3_9), [step3_10] "=&r"(step3_10),
          [step3_11] "=&r"(step3_11), [step3_12] "=&r"(step3_12),
          [step3_13] "=&r"(step3_13), [step3_14] "=&r"(step3_14),
          [step3_15] "=&r"(step3_15)
        : [const_2_power_13] "r"(const_2_power_13), [step2_8] "r"(step2_8),
          [step2_9] "r"(step2_9), [step2_10] "r"(step2_10),
          [step2_11] "r"(step2_11), [step2_12] "r"(step2_12),
          [step2_13] "r"(step2_13), [step2_14] "r"(step2_14),
          [step2_15] "r"(step2_15), [cospi_16_64] "r"(cospi_16_64));

    __asm__ __volatile__(
        "mtlo     %[const_2_power_13],  $ac0                            \n\t"
        "mthi     $zero,                $ac0                            \n\t"
        "mtlo     %[const_2_power_13],  $ac1                            \n\t"
        "mthi     $zero,                $ac1                            \n\t"
        "sub      %[temp0],             %[step1_17],    %[step1_18]     \n\t"
        "sub      %[temp1],             %[step1_30],    %[step1_29]     \n\t"
        "add      %[step3_17],          %[step1_17],    %[step1_18]     \n\t"
        "add      %[step3_30],          %[step1_30],    %[step1_29]     \n\t"

        "msub     $ac0,                 %[temp0],       %[cospi_8_64]   \n\t"
        "madd     $ac0,                 %[temp1],       %[cospi_24_64]  \n\t"
        "extp     %[step3_18],          $ac0,           31              \n\t"
        "madd     $ac1,                 %[temp0],       %[cospi_24_64]  \n\t"
        "madd     $ac1,                 %[temp1],       %[cospi_8_64]   \n\t"
        "extp     %[step3_29],          $ac1,           31              \n\t"

        : [temp0] "=&r"(temp0), [temp1] "=&r"(temp1),
          [step3_18] "=&r"(step3_18), [step3_29] "=&r"(step3_29),
          [step3_17] "=&r"(step3_17), [step3_30] "=&r"(step3_30)
        : [const_2_power_13] "r"(const_2_power_13), [step1_17] "r"(step1_17),
          [step1_18] "r"(step1_18), [step1_30] "r"(step1_30),
          [step1_29] "r"(step1_29), [cospi_24_64] "r"(cospi_24_64),
          [cospi_8_64] "r"(cospi_8_64));

    __asm__ __volatile__(
        "mtlo     %[const_2_power_13],  $ac0                            \n\t"
        "mthi     $zero,                $ac0                            \n\t"
        "mtlo     %[const_2_power_13],  $ac1                            \n\t"
        "mthi     $zero,                $ac1                            \n\t"
        "sub      %[temp0],             %[step1_16],    %[step1_19]     \n\t"
        "sub      %[temp1],             %[step1_31],    %[step1_28]     \n\t"
        "add      %[step3_16],          %[step1_16],    %[step1_19]     \n\t"
        "add      %[step3_31],          %[step1_31],    %[step1_28]     \n\t"

        "msub     $ac0,                 %[temp0],       %[cospi_8_64]   \n\t"
        "madd     $ac0,                 %[temp1],       %[cospi_24_64]  \n\t"
        "extp     %[step3_19],          $ac0,           31              \n\t"
        "madd     $ac1,                 %[temp0],       %[cospi_24_64]  \n\t"
        "madd     $ac1,                 %[temp1],       %[cospi_8_64]   \n\t"
        "extp     %[step3_28],          $ac1,           31              \n\t"

        : [temp0] "=&r"(temp0), [temp1] "=&r"(temp1),
          [step3_16] "=&r"(step3_16), [step3_31] "=&r"(step3_31),
          [step3_19] "=&r"(step3_19), [step3_28] "=&r"(step3_28)
        : [const_2_power_13] "r"(const_2_power_13), [step1_16] "r"(step1_16),
          [step1_19] "r"(step1_19), [step1_31] "r"(step1_31),
          [step1_28] "r"(step1_28), [cospi_24_64] "r"(cospi_24_64),
          [cospi_8_64] "r"(cospi_8_64));

    __asm__ __volatile__(
        "mtlo     %[const_2_power_13],  $ac0                            \n\t"
        "mthi     $zero,                $ac0                            \n\t"
        "mtlo     %[const_2_power_13],  $ac1                            \n\t"
        "mthi     $zero,                $ac1                            \n\t"
        "sub      %[temp0],             %[step1_23],    %[step1_20]     \n\t"
        "sub      %[temp1],             %[step1_24],    %[step1_27]     \n\t"
        "add      %[step3_23],          %[step1_23],    %[step1_20]     \n\t"
        "add      %[step3_24],          %[step1_24],    %[step1_27]     \n\t"

        "msub     $ac0,                 %[temp0],       %[cospi_8_64]   \n\t"
        "madd     $ac0,                 %[temp1],       %[cospi_24_64]  \n\t"
        "extp     %[step3_27],          $ac0,           31              \n\t"
        "msub     $ac1,                 %[temp0],       %[cospi_24_64]  \n\t"
        "msub     $ac1,                 %[temp1],       %[cospi_8_64]   \n\t"
        "extp     %[step3_20],          $ac1,           31              \n\t"

        : [temp0] "=&r"(temp0), [temp1] "=&r"(temp1),
          [step3_23] "=&r"(step3_23), [step3_24] "=&r"(step3_24),
          [step3_20] "=&r"(step3_20), [step3_27] "=&r"(step3_27)
        : [const_2_power_13] "r"(const_2_power_13), [step1_23] "r"(step1_23),
          [step1_20] "r"(step1_20), [step1_24] "r"(step1_24),
          [step1_27] "r"(step1_27), [cospi_24_64] "r"(cospi_24_64),
          [cospi_8_64] "r"(cospi_8_64));

    __asm__ __volatile__(
        "mtlo     %[const_2_power_13],  $ac0                            \n\t"
        "mthi     $zero,                $ac0                            \n\t"
        "mtlo     %[const_2_power_13],  $ac1                            \n\t"
        "mthi     $zero,                $ac1                            \n\t"
        "sub      %[temp0],             %[step1_22],    %[step1_21]     \n\t"
        "sub      %[temp1],             %[step1_25],    %[step1_26]     \n\t"
        "add      %[step3_22],          %[step1_22],    %[step1_21]     \n\t"
        "add      %[step3_25],          %[step1_25],    %[step1_26]     \n\t"

        "msub     $ac0,                 %[temp0],       %[cospi_24_64]  \n\t"
        "msub     $ac0,                 %[temp1],       %[cospi_8_64]   \n\t"
        "extp     %[step3_21],          $ac0,           31              \n\t"
        "msub     $ac1,                 %[temp0],       %[cospi_8_64]   \n\t"
        "madd     $ac1,                 %[temp1],       %[cospi_24_64]  \n\t"
        "extp     %[step3_26],          $ac1,           31              \n\t"

        : [temp0] "=&r"(temp0), [temp1] "=&r"(temp1),
          [step3_22] "=&r"(step3_22), [step3_25] "=&r"(step3_25),
          [step3_21] "=&r"(step3_21), [step3_26] "=&r"(step3_26)
        : [const_2_power_13] "r"(const_2_power_13), [step1_22] "r"(step1_22),
          [step1_21] "r"(step1_21), [step1_25] "r"(step1_25),
          [step1_26] "r"(step1_26), [cospi_24_64] "r"(cospi_24_64),
          [cospi_8_64] "r"(cospi_8_64));

    __asm__ __volatile__(
        "add      %[step2_16],          %[step3_16],    %[step3_23]     \n\t"
        "add      %[step2_17],          %[step3_17],    %[step3_22]     \n\t"
        "add      %[step2_18],          %[step3_18],    %[step3_21]     \n\t"
        "add      %[step2_19],          %[step3_19],    %[step3_20]     \n\t"
        "sub      %[step2_20],          %[step3_19],    %[step3_20]     \n\t"
        "sub      %[step2_21],          %[step3_18],    %[step3_21]     \n\t"
        "sub      %[step2_22],          %[step3_17],    %[step3_22]     \n\t"
        "sub      %[step2_23],          %[step3_16],    %[step3_23]     \n\t"

        : [step2_16] "=&r"(step2_16), [step2_17] "=&r"(step2_17),
          [step2_18] "=&r"(step2_18), [step2_19] "=&r"(step2_19),
          [step2_20] "=&r"(step2_20), [step2_21] "=&r"(step2_21),
          [step2_22] "=&r"(step2_22), [step2_23] "=&r"(step2_23)
        : [step3_16] "r"(step3_16), [step3_23] "r"(step3_23),
          [step3_17] "r"(step3_17), [step3_22] "r"(step3_22),
          [step3_18] "r"(step3_18), [step3_21] "r"(step3_21),
          [step3_19] "r"(step3_19), [step3_20] "r"(step3_20));

    __asm__ __volatile__(
        "sub      %[step2_24],          %[step3_31],    %[step3_24]     \n\t"
        "sub      %[step2_25],          %[step3_30],    %[step3_25]     \n\t"
        "sub      %[step2_26],          %[step3_29],    %[step3_26]     \n\t"
        "sub      %[step2_27],          %[step3_28],    %[step3_27]     \n\t"
        "add      %[step2_28],          %[step3_28],    %[step3_27]     \n\t"
        "add      %[step2_29],          %[step3_29],    %[step3_26]     \n\t"
        "add      %[step2_30],          %[step3_30],    %[step3_25]     \n\t"
        "add      %[step2_31],          %[step3_31],    %[step3_24]     \n\t"

        : [step2_24] "=&r"(step2_24), [step2_28] "=&r"(step2_28),
          [step2_25] "=&r"(step2_25), [step2_29] "=&r"(step2_29),
          [step2_26] "=&r"(step2_26), [step2_30] "=&r"(step2_30),
          [step2_27] "=&r"(step2_27), [step2_31] "=&r"(step2_31)
        : [step3_31] "r"(step3_31), [step3_24] "r"(step3_24),
          [step3_30] "r"(step3_30), [step3_25] "r"(step3_25),
          [step3_29] "r"(step3_29), [step3_26] "r"(step3_26),
          [step3_28] "r"(step3_28), [step3_27] "r"(step3_27));

    __asm__ __volatile__(
        "lh       %[load1],             0(%[input])                     \n\t"
        "lh       %[load2],             32(%[input])                    \n\t"
        "lh       %[load3],             16(%[input])                    \n\t"
        "lh       %[load4],             48(%[input])                    \n\t"

        "mtlo     %[const_2_power_13],  $ac1                            \n\t"
        "mthi     $zero,                $ac1                            \n\t"
        "mtlo     %[const_2_power_13],  $ac2                            \n\t"
        "mthi     $zero,                $ac2                            \n\t"
        "add      %[result1],           %[load1],       %[load2]        \n\t"
        "sub      %[result2],           %[load1],       %[load2]        \n\t"
        "madd     $ac1,                 %[result1],     %[cospi_16_64]  \n\t"
        "madd     $ac2,                 %[result2],     %[cospi_16_64]  \n\t"
        "extp     %[temp0],             $ac1,           31              \n\t"
        "extp     %[temp1],             $ac2,           31              \n\t"

        "mtlo     %[const_2_power_13],  $ac3                            \n\t"
        "mthi     $zero,                $ac3                            \n\t"
        "madd     $ac3,                 %[load3],       %[cospi_24_64]  \n\t"
        "msub     $ac3,                 %[load4],       %[cospi_8_64]   \n\t"
        "extp     %[temp2],             $ac3,           31              \n\t"
        "mtlo     %[const_2_power_13],  $ac1                            \n\t"
        "mthi     $zero,                $ac1                            \n\t"
        "madd     $ac1,                 %[load3],       %[cospi_8_64]   \n\t"
        "madd     $ac1,                 %[load4],       %[cospi_24_64]  \n\t"
        "extp     %[temp3],             $ac1,           31              \n\t"
        "add      %[step1_0],           %[temp0],       %[temp3]        \n\t"
        "add      %[step1_1],           %[temp1],       %[temp2]        \n\t"
        "sub      %[step1_2],           %[temp1],       %[temp2]        \n\t"
        "sub      %[step1_3],           %[temp0],       %[temp3]        \n\t"

        : [load1] "=&r"(load1), [load2] "=&r"(load2), [load3] "=&r"(load3),
          [load4] "=&r"(load4), [result1] "=&r"(result1),
          [result2] "=&r"(result2), [temp0] "=&r"(temp0), [temp1] "=&r"(temp1),
          [temp2] "=&r"(temp2), [temp3] "=&r"(temp3), [step1_0] "=&r"(step1_0),
          [step1_1] "=&r"(step1_1), [step1_2] "=&r"(step1_2),
          [step1_3] "=&r"(step1_3)
        : [const_2_power_13] "r"(const_2_power_13), [input] "r"(input),
          [cospi_24_64] "r"(cospi_24_64), [cospi_8_64] "r"(cospi_8_64),
          [cospi_16_64] "r"(cospi_16_64));

    __asm__ __volatile__(
        "lh       %[load1],             8(%[input])                     \n\t"
        "lh       %[load2],             56(%[input])                    \n\t"
        "lh       %[load3],             40(%[input])                    \n\t"
        "lh       %[load4],             24(%[input])                    \n\t"

        "mtlo     %[const_2_power_13],  $ac1                            \n\t"
        "mthi     $zero,                $ac1                            \n\t"
        "mtlo     %[const_2_power_13],  $ac3                            \n\t"
        "mthi     $zero,                $ac3                            \n\t"

        "madd     $ac1,                 %[load1],       %[cospi_28_64]  \n\t"
        "msub     $ac1,                 %[load2],       %[cospi_4_64]   \n\t"
        "extp     %[temp0],             $ac1,           31              \n\t"
        "madd     $ac3,                 %[load1],       %[cospi_4_64]   \n\t"
        "madd     $ac3,                 %[load2],       %[cospi_28_64]  \n\t"
        "extp     %[temp3],             $ac3,           31              \n\t"

        "mtlo     %[const_2_power_13],  $ac1                            \n\t"
        "mthi     $zero,                $ac1                            \n\t"
        "mtlo     %[const_2_power_13],  $ac2                            \n\t"
        "mthi     $zero,                $ac2                            \n\t"

        "madd     $ac2,                 %[load3],       %[cospi_12_64]  \n\t"
        "msub     $ac2,                 %[load4],       %[cospi_20_64]  \n\t"
        "extp     %[temp1],             $ac2,           31              \n\t"
        "madd     $ac1,                 %[load3],       %[cospi_20_64]  \n\t"
        "madd     $ac1,                 %[load4],       %[cospi_12_64]  \n\t"
        "extp     %[temp2],             $ac1,           31              \n\t"

        "mtlo     %[const_2_power_13],  $ac1                            \n\t"
        "mthi     $zero,                $ac1                            \n\t"
        "mtlo     %[const_2_power_13],  $ac3                            \n\t"
        "mthi     $zero,                $ac3                            \n\t"

        "sub      %[load1],             %[temp3],       %[temp2]        \n\t"
        "sub      %[load1],             %[load1],       %[temp0]        \n\t"
        "add      %[load1],             %[load1],       %[temp1]        \n\t"
        "sub      %[load2],             %[temp0],       %[temp1]        \n\t"
        "sub      %[load2],             %[load2],       %[temp2]        \n\t"
        "add      %[load2],             %[load2],       %[temp3]        \n\t"
        "madd     $ac1,                 %[load1],       %[cospi_16_64]  \n\t"
        "madd     $ac3,                 %[load2],       %[cospi_16_64]  \n\t"

        "extp     %[step1_5],           $ac1,           31              \n\t"
        "extp     %[step1_6],           $ac3,           31              \n\t"
        "add      %[step1_4],           %[temp0],       %[temp1]        \n\t"
        "add      %[step1_7],           %[temp3],       %[temp2]        \n\t"

        : [load1] "=&r"(load1), [load2] "=&r"(load2), [load3] "=&r"(load3),
          [load4] "=&r"(load4), [temp0] "=&r"(temp0), [temp1] "=&r"(temp1),
          [temp2] "=&r"(temp2), [temp3] "=&r"(temp3), [step1_4] "=&r"(step1_4),
          [step1_5] "=&r"(step1_5), [step1_6] "=&r"(step1_6),
          [step1_7] "=&r"(step1_7)
        : [const_2_power_13] "r"(const_2_power_13), [input] "r"(input),
          [cospi_20_64] "r"(cospi_20_64), [cospi_12_64] "r"(cospi_12_64),
          [cospi_4_64] "r"(cospi_4_64), [cospi_28_64] "r"(cospi_28_64),
          [cospi_16_64] "r"(cospi_16_64));

    __asm__ __volatile__(
        "add      %[step2_0],          %[step1_0],    %[step1_7]     \n\t"
        "add      %[step2_1],          %[step1_1],    %[step1_6]     \n\t"
        "add      %[step2_2],          %[step1_2],    %[step1_5]     \n\t"
        "add      %[step2_3],          %[step1_3],    %[step1_4]     \n\t"
        "sub      %[step2_4],          %[step1_3],    %[step1_4]     \n\t"
        "sub      %[step2_5],          %[step1_2],    %[step1_5]     \n\t"
        "sub      %[step2_6],          %[step1_1],    %[step1_6]     \n\t"
        "sub      %[step2_7],          %[step1_0],    %[step1_7]     \n\t"

        : [step2_0] "=&r"(step2_0), [step2_4] "=&r"(step2_4),
          [step2_1] "=&r"(step2_1), [step2_5] "=&r"(step2_5),
          [step2_2] "=&r"(step2_2), [step2_6] "=&r"(step2_6),
          [step2_3] "=&r"(step2_3), [step2_7] "=&r"(step2_7)
        : [step1_0] "r"(step1_0), [step1_7] "r"(step1_7),
          [step1_1] "r"(step1_1), [step1_6] "r"(step1_6),
          [step1_2] "r"(step1_2), [step1_5] "r"(step1_5),
          [step1_3] "r"(step1_3), [step1_4] "r"(step1_4));

    // stage 7
    __asm__ __volatile__(
        "add      %[step1_0],          %[step2_0],    %[step3_15]     \n\t"
        "add      %[step1_1],          %[step2_1],    %[step3_14]     \n\t"
        "add      %[step1_2],          %[step2_2],    %[step3_13]     \n\t"
        "add      %[step1_3],          %[step2_3],    %[step3_12]     \n\t"
        "sub      %[step1_12],         %[step2_3],    %[step3_12]     \n\t"
        "sub      %[step1_13],         %[step2_2],    %[step3_13]     \n\t"
        "sub      %[step1_14],         %[step2_1],    %[step3_14]     \n\t"
        "sub      %[step1_15],         %[step2_0],    %[step3_15]     \n\t"

        : [step1_0] "=&r"(step1_0), [step1_12] "=&r"(step1_12),
          [step1_1] "=&r"(step1_1), [step1_13] "=&r"(step1_13),
          [step1_2] "=&r"(step1_2), [step1_14] "=&r"(step1_14),
          [step1_3] "=&r"(step1_3), [step1_15] "=&r"(step1_15)
        : [step2_0] "r"(step2_0), [step3_15] "r"(step3_15),
          [step2_1] "r"(step2_1), [step3_14] "r"(step3_14),
          [step2_2] "r"(step2_2), [step3_13] "r"(step3_13),
          [step2_3] "r"(step2_3), [step3_12] "r"(step3_12));

    __asm__ __volatile__(
        "add      %[step1_4],          %[step2_4],    %[step3_11]     \n\t"
        "add      %[step1_5],          %[step2_5],    %[step3_10]     \n\t"
        "add      %[step1_6],          %[step2_6],    %[step3_9]      \n\t"
        "add      %[step1_7],          %[step2_7],    %[step3_8]      \n\t"
        "sub      %[step1_8],          %[step2_7],    %[step3_8]      \n\t"
        "sub      %[step1_9],          %[step2_6],    %[step3_9]      \n\t"
        "sub      %[step1_10],         %[step2_5],    %[step3_10]     \n\t"
        "sub      %[step1_11],         %[step2_4],    %[step3_11]     \n\t"

        : [step1_4] "=&r"(step1_4), [step1_8] "=&r"(step1_8),
          [step1_5] "=&r"(step1_5), [step1_9] "=&r"(step1_9),
          [step1_6] "=&r"(step1_6), [step1_10] "=&r"(step1_10),
          [step1_7] "=&r"(step1_7), [step1_11] "=&r"(step1_11)
        : [step2_4] "r"(step2_4), [step3_11] "r"(step3_11),
          [step2_5] "r"(step2_5), [step3_10] "r"(step3_10),
          [step2_6] "r"(step2_6), [step3_9] "r"(step3_9),
          [step2_7] "r"(step2_7), [step3_8] "r"(step3_8));

    __asm__ __volatile__(
        "sub      %[temp0],             %[step2_27],    %[step2_20]     \n\t"
        "add      %[temp1],             %[step2_27],    %[step2_20]     \n\t"
        "sub      %[temp2],             %[step2_26],    %[step2_21]     \n\t"
        "add      %[temp3],             %[step2_26],    %[step2_21]     \n\t"

        "mtlo     %[const_2_power_13],  $ac0                            \n\t"
        "mthi     $zero,                $ac0                            \n\t"
        "mtlo     %[const_2_power_13],  $ac1                            \n\t"
        "mthi     $zero,                $ac1                            \n\t"
        "mtlo     %[const_2_power_13],  $ac2                            \n\t"
        "mthi     $zero,                $ac2                            \n\t"
        "mtlo     %[const_2_power_13],  $ac3                            \n\t"
        "mthi     $zero,                $ac3                            \n\t"

        "madd     $ac0,                 %[temp0],       %[cospi_16_64]  \n\t"
        "madd     $ac1,                 %[temp1],       %[cospi_16_64]  \n\t"
        "madd     $ac2,                 %[temp2],       %[cospi_16_64]  \n\t"
        "madd     $ac3,                 %[temp3],       %[cospi_16_64]  \n\t"

        "extp     %[step1_20],          $ac0,           31              \n\t"
        "extp     %[step1_27],          $ac1,           31              \n\t"
        "extp     %[step1_21],          $ac2,           31              \n\t"
        "extp     %[step1_26],          $ac3,           31              \n\t"

        : [temp0] "=&r"(temp0), [temp1] "=&r"(temp1), [temp2] "=&r"(temp2),
          [temp3] "=&r"(temp3), [step1_20] "=&r"(step1_20),
          [step1_27] "=&r"(step1_27), [step1_21] "=&r"(step1_21),
          [step1_26] "=&r"(step1_26)
        : [const_2_power_13] "r"(const_2_power_13), [step2_20] "r"(step2_20),
          [step2_27] "r"(step2_27), [step2_21] "r"(step2_21),
          [step2_26] "r"(step2_26), [cospi_16_64] "r"(cospi_16_64));

    __asm__ __volatile__(
        "sub      %[temp0],             %[step2_25],    %[step2_22]     \n\t"
        "add      %[temp1],             %[step2_25],    %[step2_22]     \n\t"
        "sub      %[temp2],             %[step2_24],    %[step2_23]     \n\t"
        "add      %[temp3],             %[step2_24],    %[step2_23]     \n\t"

        "mtlo     %[const_2_power_13],  $ac0                            \n\t"
        "mthi     $zero,                $ac0                            \n\t"
        "mtlo     %[const_2_power_13],  $ac1                            \n\t"
        "mthi     $zero,                $ac1                            \n\t"
        "mtlo     %[const_2_power_13],  $ac2                            \n\t"
        "mthi     $zero,                $ac2                            \n\t"
        "mtlo     %[const_2_power_13],  $ac3                            \n\t"
        "mthi     $zero,                $ac3                            \n\t"

        "madd     $ac0,                 %[temp0],       %[cospi_16_64]  \n\t"
        "madd     $ac1,                 %[temp1],       %[cospi_16_64]  \n\t"
        "madd     $ac2,                 %[temp2],       %[cospi_16_64]  \n\t"
        "madd     $ac3,                 %[temp3],       %[cospi_16_64]  \n\t"

        "extp     %[step1_22],          $ac0,           31              \n\t"
        "extp     %[step1_25],          $ac1,           31              \n\t"
        "extp     %[step1_23],          $ac2,           31              \n\t"
        "extp     %[step1_24],          $ac3,           31              \n\t"

        : [temp0] "=&r"(temp0), [temp1] "=&r"(temp1), [temp2] "=&r"(temp2),
          [temp3] "=&r"(temp3), [step1_22] "=&r"(step1_22),
          [step1_25] "=&r"(step1_25), [step1_23] "=&r"(step1_23),
          [step1_24] "=&r"(step1_24)
        : [const_2_power_13] "r"(const_2_power_13), [step2_22] "r"(step2_22),
          [step2_25] "r"(step2_25), [step2_23] "r"(step2_23),
          [step2_24] "r"(step2_24), [cospi_16_64] "r"(cospi_16_64));

    // final stage
    __asm__ __volatile__(
        "add      %[temp0],            %[step1_0],    %[step2_31]     \n\t"
        "add      %[temp1],            %[step1_1],    %[step2_30]     \n\t"
        "add      %[temp2],            %[step1_2],    %[step2_29]     \n\t"
        "add      %[temp3],            %[step1_3],    %[step2_28]     \n\t"
        "sub      %[load1],            %[step1_3],    %[step2_28]     \n\t"
        "sub      %[load2],            %[step1_2],    %[step2_29]     \n\t"
        "sub      %[load3],            %[step1_1],    %[step2_30]     \n\t"
        "sub      %[load4],            %[step1_0],    %[step2_31]     \n\t"
        "sh       %[temp0],          0(%[output])                     \n\t"
        "sh       %[temp1],         64(%[output])                     \n\t"
        "sh       %[temp2],        128(%[output])                     \n\t"
        "sh       %[temp3],        192(%[output])                     \n\t"
        "sh       %[load1],       1792(%[output])                     \n\t"
        "sh       %[load2],       1856(%[output])                     \n\t"
        "sh       %[load3],       1920(%[output])                     \n\t"
        "sh       %[load4],       1984(%[output])                     \n\t"

        : [temp0] "=&r"(temp0), [load1] "=&r"(load1), [temp1] "=&r"(temp1),
          [load2] "=&r"(load2), [temp2] "=&r"(temp2), [load3] "=&r"(load3),
          [temp3] "=&r"(temp3), [load4] "=&r"(load4)
        : [step1_0] "r"(step1_0), [step2_31] "r"(step2_31),
          [step1_1] "r"(step1_1), [step2_30] "r"(step2_30),
          [step1_2] "r"(step1_2), [step2_29] "r"(step2_29),
          [step1_3] "r"(step1_3), [step2_28] "r"(step2_28),
          [output] "r"(output));

    __asm__ __volatile__(
        "add      %[temp0],            %[step1_4],    %[step1_27]     \n\t"
        "add      %[temp1],            %[step1_5],    %[step1_26]     \n\t"
        "add      %[temp2],            %[step1_6],    %[step1_25]     \n\t"
        "add      %[temp3],            %[step1_7],    %[step1_24]     \n\t"
        "sub      %[load1],            %[step1_7],    %[step1_24]     \n\t"
        "sub      %[load2],            %[step1_6],    %[step1_25]     \n\t"
        "sub      %[load3],            %[step1_5],    %[step1_26]     \n\t"
        "sub      %[load4],            %[step1_4],    %[step1_27]     \n\t"
        "sh       %[temp0],        256(%[output])                     \n\t"
        "sh       %[temp1],        320(%[output])                     \n\t"
        "sh       %[temp2],        384(%[output])                     \n\t"
        "sh       %[temp3],        448(%[output])                     \n\t"
        "sh       %[load1],       1536(%[output])                     \n\t"
        "sh       %[load2],       1600(%[output])                     \n\t"
        "sh       %[load3],       1664(%[output])                     \n\t"
        "sh       %[load4],       1728(%[output])                     \n\t"

        : [temp0] "=&r"(temp0), [load1] "=&r"(load1), [temp1] "=&r"(temp1),
          [load2] "=&r"(load2), [temp2] "=&r"(temp2), [load3] "=&r"(load3),
          [temp3] "=&r"(temp3), [load4] "=&r"(load4)
        : [step1_4] "r"(step1_4), [step1_27] "r"(step1_27),
          [step1_5] "r"(step1_5), [step1_26] "r"(step1_26),
          [step1_6] "r"(step1_6), [step1_25] "r"(step1_25),
          [step1_7] "r"(step1_7), [step1_24] "r"(step1_24),
          [output] "r"(output));

    __asm__ __volatile__(
        "add      %[temp0],            %[step1_8],     %[step1_23]     \n\t"
        "add      %[temp1],            %[step1_9],     %[step1_22]     \n\t"
        "add      %[temp2],            %[step1_10],    %[step1_21]     \n\t"
        "add      %[temp3],            %[step1_11],    %[step1_20]     \n\t"
        "sub      %[load1],            %[step1_11],    %[step1_20]     \n\t"
        "sub      %[load2],            %[step1_10],    %[step1_21]     \n\t"
        "sub      %[load3],            %[step1_9],     %[step1_22]     \n\t"
        "sub      %[load4],            %[step1_8],     %[step1_23]     \n\t"
        "sh       %[temp0],        512(%[output])                      \n\t"
        "sh       %[temp1],        576(%[output])                      \n\t"
        "sh       %[temp2],        640(%[output])                      \n\t"
        "sh       %[temp3],        704(%[output])                      \n\t"
        "sh       %[load1],       1280(%[output])                      \n\t"
        "sh       %[load2],       1344(%[output])                      \n\t"
        "sh       %[load3],       1408(%[output])                      \n\t"
        "sh       %[load4],       1472(%[output])                      \n\t"

        : [temp0] "=&r"(temp0), [load1] "=&r"(load1), [temp1] "=&r"(temp1),
          [load2] "=&r"(load2), [temp2] "=&r"(temp2), [load3] "=&r"(load3),
          [temp3] "=&r"(temp3), [load4] "=&r"(load4)
        : [step1_8] "r"(step1_8), [step1_23] "r"(step1_23),
          [step1_9] "r"(step1_9), [step1_22] "r"(step1_22),
          [step1_10] "r"(step1_10), [step1_21] "r"(step1_21),
          [step1_11] "r"(step1_11), [step1_20] "r"(step1_20),
          [output] "r"(output));

    __asm__ __volatile__(
        "add      %[temp0],            %[step1_12],    %[step2_19]     \n\t"
        "add      %[temp1],            %[step1_13],    %[step2_18]     \n\t"
        "add      %[temp2],            %[step1_14],    %[step2_17]     \n\t"
        "add      %[temp3],            %[step1_15],    %[step2_16]     \n\t"
        "sub      %[load1],            %[step1_15],    %[step2_16]     \n\t"
        "sub      %[load2],            %[step1_14],    %[step2_17]     \n\t"
        "sub      %[load3],            %[step1_13],    %[step2_18]     \n\t"
        "sub      %[load4],            %[step1_12],    %[step2_19]     \n\t"
        "sh       %[temp0],        768(%[output])                      \n\t"
        "sh       %[temp1],        832(%[output])                      \n\t"
        "sh       %[temp2],        896(%[output])                      \n\t"
        "sh       %[temp3],        960(%[output])                      \n\t"
        "sh       %[load1],       1024(%[output])                      \n\t"
        "sh       %[load2],       1088(%[output])                      \n\t"
        "sh       %[load3],       1152(%[output])                      \n\t"
        "sh       %[load4],       1216(%[output])                      \n\t"

        : [temp0] "=&r"(temp0), [load1] "=&r"(load1), [temp1] "=&r"(temp1),
          [load2] "=&r"(load2), [temp2] "=&r"(temp2), [load3] "=&r"(load3),
          [temp3] "=&r"(temp3), [load4] "=&r"(load4)
        : [step1_12] "r"(step1_12), [step2_19] "r"(step2_19),
          [step1_13] "r"(step1_13), [step2_18] "r"(step2_18),
          [step1_14] "r"(step1_14), [step2_17] "r"(step2_17),
          [step1_15] "r"(step1_15), [step2_16] "r"(step2_16),
          [output] "r"(output));

    input += 32;
    output += 1;
  }
}

void vpx_idct32x32_1024_add_dspr2(const int16_t *input, uint8_t *dest,
                                  int stride) {
  DECLARE_ALIGNED(32, int16_t, out[32 * 32]);
  int16_t *outptr = out;
  uint32_t pos = 45;

  /* bit positon for extract from acc */
  __asm__ __volatile__("wrdsp      %[pos],     1           \n\t"
                       :
                       : [pos] "r"(pos));

  // Rows
  idct32_rows_dspr2(input, outptr, 32);

  // Columns
  vpx_idct32_cols_add_blk_dspr2(out, dest, stride);
}

void vpx_idct32x32_34_add_dspr2(const int16_t *input, uint8_t *dest,
                                int stride) {
  DECLARE_ALIGNED(32, int16_t, out[32 * 32]);
  int16_t *outptr = out;
  uint32_t i;
  uint32_t pos = 45;

  /* bit positon for extract from acc */
  __asm__ __volatile__("wrdsp      %[pos],     1           \n\t"
                       :
                       : [pos] "r"(pos));

  // Rows
  idct32_rows_dspr2(input, outptr, 8);

  outptr += 8;
  __asm__ __volatile__(
      "sw     $zero,      0(%[outptr])     \n\t"
      "sw     $zero,      4(%[outptr])     \n\t"
      "sw     $zero,      8(%[outptr])     \n\t"
      "sw     $zero,     12(%[outptr])     \n\t"
      "sw     $zero,     16(%[outptr])     \n\t"
      "sw     $zero,     20(%[outptr])     \n\t"
      "sw     $zero,     24(%[outptr])     \n\t"
      "sw     $zero,     28(%[outptr])     \n\t"
      "sw     $zero,     32(%[outptr])     \n\t"
      "sw     $zero,     36(%[outptr])     \n\t"
      "sw     $zero,     40(%[outptr])     \n\t"
      "sw     $zero,     44(%[outptr])     \n\t"

      :
      : [outptr] "r"(outptr));

  for (i = 0; i < 31; ++i) {
    outptr += 32;

    __asm__ __volatile__(
        "sw     $zero,      0(%[outptr])     \n\t"
        "sw     $zero,      4(%[outptr])     \n\t"
        "sw     $zero,      8(%[outptr])     \n\t"
        "sw     $zero,     12(%[outptr])     \n\t"
        "sw     $zero,     16(%[outptr])     \n\t"
        "sw     $zero,     20(%[outptr])     \n\t"
        "sw     $zero,     24(%[outptr])     \n\t"
        "sw     $zero,     28(%[outptr])     \n\t"
        "sw     $zero,     32(%[outptr])     \n\t"
        "sw     $zero,     36(%[outptr])     \n\t"
        "sw     $zero,     40(%[outptr])     \n\t"
        "sw     $zero,     44(%[outptr])     \n\t"

        :
        : [outptr] "r"(outptr));
  }

  // Columns
  vpx_idct32_cols_add_blk_dspr2(out, dest, stride);
}

void vpx_idct32x32_1_add_dspr2(const int16_t *input, uint8_t *dest,
                               int stride) {
  int r, out;
  int32_t a1, absa1;
  int32_t vector_a1;
  int32_t t1, t2, t3, t4;
  int32_t vector_1, vector_2, vector_3, vector_4;
  uint32_t pos = 45;

  /* bit positon for extract from acc */
  __asm__ __volatile__("wrdsp      %[pos],     1           \n\t"

                       :
                       : [pos] "r"(pos));

  out = DCT_CONST_ROUND_SHIFT_TWICE_COSPI_16_64(input[0]);
  __asm__ __volatile__(
      "addi     %[out],    %[out],    32      \n\t"
      "sra      %[a1],     %[out],    6       \n\t"

      : [out] "+r"(out), [a1] "=r"(a1)
      :);

  if (a1 < 0) {
    /* use quad-byte
     * input and output memory are four byte aligned */
    __asm__ __volatile__(
        "abs        %[absa1],     %[a1]         \n\t"
        "replv.qb   %[vector_a1], %[absa1]      \n\t"

        : [absa1] "=&r"(absa1), [vector_a1] "=&r"(vector_a1)
        : [a1] "r"(a1));

    for (r = 32; r--;) {
      __asm__ __volatile__(
          "lw             %[t1],          0(%[dest])                      \n\t"
          "lw             %[t2],          4(%[dest])                      \n\t"
          "lw             %[t3],          8(%[dest])                      \n\t"
          "lw             %[t4],          12(%[dest])                     \n\t"
          "subu_s.qb      %[vector_1],    %[t1],          %[vector_a1]    \n\t"
          "subu_s.qb      %[vector_2],    %[t2],          %[vector_a1]    \n\t"
          "subu_s.qb      %[vector_3],    %[t3],          %[vector_a1]    \n\t"
          "subu_s.qb      %[vector_4],    %[t4],          %[vector_a1]    \n\t"
          "sw             %[vector_1],    0(%[dest])                      \n\t"
          "sw             %[vector_2],    4(%[dest])                      \n\t"
          "sw             %[vector_3],    8(%[dest])                      \n\t"
          "sw             %[vector_4],    12(%[dest])                     \n\t"

          "lw             %[t1],          16(%[dest])                     \n\t"
          "lw             %[t2],          20(%[dest])                     \n\t"
          "lw             %[t3],          24(%[dest])                     \n\t"
          "lw             %[t4],          28(%[dest])                     \n\t"
          "subu_s.qb      %[vector_1],    %[t1],          %[vector_a1]    \n\t"
          "subu_s.qb      %[vector_2],    %[t2],          %[vector_a1]    \n\t"
          "subu_s.qb      %[vector_3],    %[t3],          %[vector_a1]    \n\t"
          "subu_s.qb      %[vector_4],    %[t4],          %[vector_a1]    \n\t"
          "sw             %[vector_1],    16(%[dest])                     \n\t"
          "sw             %[vector_2],    20(%[dest])                     \n\t"
          "sw             %[vector_3],    24(%[dest])                     \n\t"
          "sw             %[vector_4],    28(%[dest])                     \n\t"

          "add            %[dest],        %[dest],        %[stride]       \n\t"

          : [t1] "=&r"(t1), [t2] "=&r"(t2), [t3] "=&r"(t3), [t4] "=&r"(t4),
            [vector_1] "=&r"(vector_1), [vector_2] "=&r"(vector_2),
            [vector_3] "=&r"(vector_3), [vector_4] "=&r"(vector_4),
            [dest] "+&r"(dest)
          : [stride] "r"(stride), [vector_a1] "r"(vector_a1));
    }
  } else if (a1 > 255) {
    int32_t a11, a12, vector_a11, vector_a12;

    /* use quad-byte
     * input and output memory are four byte aligned */
    a11 = a1 >> 1;
    a12 = a1 - a11;
    __asm__ __volatile__(
        "replv.qb       %[vector_a11],  %[a11]     \n\t"
        "replv.qb       %[vector_a12],  %[a12]     \n\t"

        : [vector_a11] "=&r"(vector_a11), [vector_a12] "=&r"(vector_a12)
        : [a11] "r"(a11), [a12] "r"(a12));

    for (r = 32; r--;) {
      __asm__ __volatile__(
          "lw             %[t1],          0(%[dest])                      \n\t"
          "lw             %[t2],          4(%[dest])                      \n\t"
          "lw             %[t3],          8(%[dest])                      \n\t"
          "lw             %[t4],          12(%[dest])                     \n\t"
          "addu_s.qb      %[vector_1],    %[t1],          %[vector_a11]   \n\t"
          "addu_s.qb      %[vector_2],    %[t2],          %[vector_a11]   \n\t"
          "addu_s.qb      %[vector_3],    %[t3],          %[vector_a11]   \n\t"
          "addu_s.qb      %[vector_4],    %[t4],          %[vector_a11]   \n\t"
          "addu_s.qb      %[vector_1],    %[vector_1],    %[vector_a12]   \n\t"
          "addu_s.qb      %[vector_2],    %[vector_2],    %[vector_a12]   \n\t"
          "addu_s.qb      %[vector_3],    %[vector_3],    %[vector_a12]   \n\t"
          "addu_s.qb      %[vector_4],    %[vector_4],    %[vector_a12]   \n\t"
          "sw             %[vector_1],    0(%[dest])                      \n\t"
          "sw             %[vector_2],    4(%[dest])                      \n\t"
          "sw             %[vector_3],    8(%[dest])                      \n\t"
          "sw             %[vector_4],    12(%[dest])                     \n\t"

          "lw             %[t1],          16(%[dest])                     \n\t"
          "lw             %[t2],          20(%[dest])                     \n\t"
          "lw             %[t3],          24(%[dest])                     \n\t"
          "lw             %[t4],          28(%[dest])                     \n\t"
          "addu_s.qb      %[vector_1],    %[t1],          %[vector_a11]    \n\t"
          "addu_s.qb      %[vector_2],    %[t2],          %[vector_a11]    \n\t"
          "addu_s.qb      %[vector_3],    %[t3],          %[vector_a11]    \n\t"
          "addu_s.qb      %[vector_4],    %[t4],          %[vector_a11]    \n\t"
          "addu_s.qb      %[vector_1],    %[vector_1],    %[vector_a12]   \n\t"
          "addu_s.qb      %[vector_2],    %[vector_2],    %[vector_a12]   \n\t"
          "addu_s.qb      %[vector_3],    %[vector_3],    %[vector_a12]   \n\t"
          "addu_s.qb      %[vector_4],    %[vector_4],    %[vector_a12]   \n\t"
          "sw             %[vector_1],    16(%[dest])                     \n\t"
          "sw             %[vector_2],    20(%[dest])                     \n\t"
          "sw             %[vector_3],    24(%[dest])                     \n\t"
          "sw             %[vector_4],    28(%[dest])                     \n\t"

          "add            %[dest],        %[dest],        %[stride]       \n\t"

          : [t1] "=&r"(t1), [t2] "=&r"(t2), [t3] "=&r"(t3), [t4] "=&r"(t4),
            [vector_1] "=&r"(vector_1), [vector_2] "=&r"(vector_2),
            [vector_3] "=&r"(vector_3), [vector_4] "=&r"(vector_4),
            [dest] "+&r"(dest)
          : [stride] "r"(stride), [vector_a11] "r"(vector_a11),
            [vector_a12] "r"(vector_a12));
    }
  } else {
    /* use quad-byte
     * input and output memory are four byte aligned */
    __asm__ __volatile__("replv.qb       %[vector_a1],   %[a1]     \n\t"

                         : [vector_a1] "=&r"(vector_a1)
                         : [a1] "r"(a1));

    for (r = 32; r--;) {
      __asm__ __volatile__(
          "lw             %[t1],          0(%[dest])                      \n\t"
          "lw             %[t2],          4(%[dest])                      \n\t"
          "lw             %[t3],          8(%[dest])                      \n\t"
          "lw             %[t4],          12(%[dest])                     \n\t"
          "addu_s.qb      %[vector_1],    %[t1],          %[vector_a1]    \n\t"
          "addu_s.qb      %[vector_2],    %[t2],          %[vector_a1]    \n\t"
          "addu_s.qb      %[vector_3],    %[t3],          %[vector_a1]    \n\t"
          "addu_s.qb      %[vector_4],    %[t4],          %[vector_a1]    \n\t"
          "sw             %[vector_1],    0(%[dest])                      \n\t"
          "sw             %[vector_2],    4(%[dest])                      \n\t"
          "sw             %[vector_3],    8(%[dest])                      \n\t"
          "sw             %[vector_4],    12(%[dest])                     \n\t"

          "lw             %[t1],          16(%[dest])                     \n\t"
          "lw             %[t2],          20(%[dest])                     \n\t"
          "lw             %[t3],          24(%[dest])                     \n\t"
          "lw             %[t4],          28(%[dest])                     \n\t"
          "addu_s.qb      %[vector_1],    %[t1],          %[vector_a1]    \n\t"
          "addu_s.qb      %[vector_2],    %[t2],          %[vector_a1]    \n\t"
          "addu_s.qb      %[vector_3],    %[t3],          %[vector_a1]    \n\t"
          "addu_s.qb      %[vector_4],    %[t4],          %[vector_a1]    \n\t"
          "sw             %[vector_1],    16(%[dest])                     \n\t"
          "sw             %[vector_2],    20(%[dest])                     \n\t"
          "sw             %[vector_3],    24(%[dest])                     \n\t"
          "sw             %[vector_4],    28(%[dest])                     \n\t"

          "add            %[dest],        %[dest],        %[stride]       \n\t"

          : [t1] "=&r"(t1), [t2] "=&r"(t2), [t3] "=&r"(t3), [t4] "=&r"(t4),
            [vector_1] "=&r"(vector_1), [vector_2] "=&r"(vector_2),
            [vector_3] "=&r"(vector_3), [vector_4] "=&r"(vector_4),
            [dest] "+&r"(dest)
          : [stride] "r"(stride), [vector_a1] "r"(vector_a1));
    }
  }
}
#endif  // #if HAVE_DSPR2