FFmpeg: libavcodec/mips/mpegaudiodsp_mips

00001 /*
00002  * Copyright (c) 2012
00003  *      MIPS Technologies, Inc., California.
00004  *
00005  * Redistribution and use in source and binary forms, with or without
00006  * modification, are permitted provided that the following conditions
00007  * are met:
00008  * 1. Redistributions of source code must retain the above copyright
00009  *    notice, this list of conditions and the following disclaimer.
00010  * 2. Redistributions in binary form must reproduce the above copyright
00011  *    notice, this list of conditions and the following disclaimer in the
00012  *    documentation and/or other materials provided with the distribution.
00013  * 3. Neither the name of the MIPS Technologies, Inc., nor the names of its
00014  *    contributors may be used to endorse or promote products derived from
00015  *    this software without specific prior written permission.
00016  *
00017  * THIS SOFTWARE IS PROVIDED BY THE MIPS TECHNOLOGIES, INC. ``AS IS'' AND
00018  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
00019  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
00020  * ARE DISCLAIMED.  IN NO EVENT SHALL THE MIPS TECHNOLOGIES, INC. BE LIABLE
00021  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
00022  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
00023  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
00024  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
00025  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
00026  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
00027  * SUCH DAMAGE.
00028  *
00029  * Author:  Bojan Zivkovic (bojan@mips.com)
00030  *
00031  * MPEG Audio decoder optimized for MIPS fixed-point architecture
00032  *
00033  * This file is part of FFmpeg.
00034  *
00035  * FFmpeg is free software; you can redistribute it and/or
00036  * modify it under the terms of the GNU Lesser General Public
00037  * License as published by the Free Software Foundation; either
00038  * version 2.1 of the License, or (at your option) any later version.
00039  *
00040  * FFmpeg is distributed in the hope that it will be useful,
00041  * but WITHOUT ANY WARRANTY; without even the implied warranty of
00042  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
00043  * Lesser General Public License for more details.
00044  *
00045  * You should have received a copy of the GNU Lesser General Public
00046  * License along with FFmpeg; if not, write to the Free Software
00047  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
00048  */
00049 
00055 #include <string.h>
00056 
00057 #include "libavcodec/mpegaudiodsp.h"
00058 
00059 static void ff_mpadsp_apply_window_mips_fixed(int32_t *synth_buf, int32_t *window,
00060                                int *dither_state, int16_t *samples, int incr)
00061 {
00062     register const int32_t *w, *w2, *p;
00063     int j;
00064     int16_t *samples2;
00065     int w_asm, p_asm, w_asm1, p_asm1, w_asm2, p_asm2;
00066     int w2_asm, w2_asm1, *p_temp1, *p_temp2;
00067     int sum1 = 0;
00068     int const min_asm = -32768, max_asm = 32767;
00069     int temp1, temp2 = 0, temp3 = 0;
00070     int64_t sum;
00071 
00072     /* copy to avoid wrap */
00073     memcpy(synth_buf + 512, synth_buf, 32 * sizeof(*synth_buf));
00074     samples2 = samples + 31 * incr;
00075     w = window;
00076     w2 = window + 31;
00077     sum = *dither_state;
00078     p = synth_buf + 16;
00079     p_temp1 = synth_buf + 16;
00080     p_temp2 = synth_buf + 48;
00081     temp1 = sum;
00082 
00087     __asm__ volatile (
00088          "mthi   $zero                                                    \n\t"
00089          "mtlo   %[temp1]                                                 \n\t"
00090          "lw     %[w_asm],  0(%[w])                                       \n\t"
00091          "lw     %[p_asm],  0(%[p])                                       \n\t"
00092          "lw     %[w_asm1], 64*4(%[w])                                    \n\t"
00093          "lw     %[p_asm1], 64*4(%[p])                                    \n\t"
00094          "lw     %[w_asm2], 128*4(%[w])                                   \n\t"
00095          "lw     %[p_asm2], 128*4(%[p])                                   \n\t"
00096          "madd   %[w_asm],  %[p_asm]                                      \n\t"
00097          "madd   %[w_asm1], %[p_asm1]                                     \n\t"
00098          "madd   %[w_asm2], %[p_asm2]                                     \n\t"
00099          "lw     %[w_asm],  192*4(%[w])                                   \n\t"
00100          "lw     %[p_asm],  192*4(%[p])                                   \n\t"
00101          "lw     %[w_asm1], 256*4(%[w])                                   \n\t"
00102          "lw     %[p_asm1], 256*4(%[p])                                   \n\t"
00103          "lw     %[w_asm2], 320*4(%[w])                                   \n\t"
00104          "lw     %[p_asm2], 320*4(%[p])                                   \n\t"
00105          "madd   %[w_asm],  %[p_asm]                                      \n\t"
00106          "madd   %[w_asm1], %[p_asm1]                                     \n\t"
00107          "madd   %[w_asm2], %[p_asm2]                                     \n\t"
00108          "lw     %[w_asm],  384*4(%[w])                                   \n\t"
00109          "lw     %[p_asm],  384*4(%[p])                                   \n\t"
00110          "lw     %[w_asm1], 448*4(%[w])                                   \n\t"
00111          "lw     %[p_asm1], 448*4(%[p])                                   \n\t"
00112          "lw     %[w_asm2], 32*4(%[w])                                    \n\t"
00113          "lw     %[p_asm2], 32*4(%[p])                                    \n\t"
00114          "madd   %[w_asm],  %[p_asm]                                      \n\t"
00115          "madd   %[w_asm1], %[p_asm1]                                     \n\t"
00116          "msub   %[w_asm2], %[p_asm2]                                     \n\t"
00117          "lw     %[w_asm],  96*4(%[w])                                    \n\t"
00118          "lw     %[p_asm],  96*4(%[p])                                    \n\t"
00119          "lw     %[w_asm1], 160*4(%[w])                                   \n\t"
00120          "lw     %[p_asm1], 160*4(%[p])                                   \n\t"
00121          "lw     %[w_asm2], 224*4(%[w])                                   \n\t"
00122          "lw     %[p_asm2], 224*4(%[p])                                   \n\t"
00123          "msub   %[w_asm],  %[p_asm]                                      \n\t"
00124          "msub   %[w_asm1], %[p_asm1]                                     \n\t"
00125          "msub   %[w_asm2], %[p_asm2]                                     \n\t"
00126          "lw     %[w_asm],  288*4(%[w])                                   \n\t"
00127          "lw     %[p_asm],  288*4(%[p])                                   \n\t"
00128          "lw     %[w_asm1], 352*4(%[w])                                   \n\t"
00129          "lw     %[p_asm1], 352*4(%[p])                                   \n\t"
00130          "msub   %[w_asm],  %[p_asm]                                      \n\t"
00131          "lw     %[w_asm],  480*4(%[w])                                   \n\t"
00132          "lw     %[p_asm],  480*4(%[p])                                   \n\t"
00133          "lw     %[w_asm2], 416*4(%[w])                                   \n\t"
00134          "lw     %[p_asm2], 416*4(%[p])                                   \n\t"
00135          "msub   %[w_asm],  %[p_asm]                                      \n\t"
00136          "msub   %[w_asm1], %[p_asm1]                                     \n\t"
00137          "msub   %[w_asm2], %[p_asm2]                                     \n\t"
00138 
00139          /*round_sample function from the original code is eliminated,
00140           * changed with appropriate assembly instructions
00141           * code example:
00142 
00143          "extr.w  %[sum1],$ac0,24                                       \n\t"
00144          "mflo %[temp3],  $ac0                                          \n\t"
00145          "and  %[temp1],  %[temp3],  0x00ffffff                         \n\t"
00146          "slt  %[temp2],  %[sum1],   %[min_asm]                         \n\t"
00147          "movn %[sum1],   %[min_asm],%[temp2]                           \n\t"
00148          "slt  %[temp2],  %[max_asm],%[sum1]                            \n\t"
00149          "movn %[sum1],   %[max_asm],%[temp2]                           \n\t"
00150          "sh   %[sum1],   0(%[samples])                                 \n\t"
00151          */
00152 
00153          "extr.w %[sum1],   $ac0,       24                                \n\t"
00154          "mflo   %[temp3]                                                 \n\t"
00155          "addi   %[w],      %[w],       4                                 \n\t"
00156          "and    %[temp1],  %[temp3],   0x00ffffff                        \n\t"
00157          "slt    %[temp2],  %[sum1],    %[min_asm]                        \n\t"
00158          "movn   %[sum1],   %[min_asm], %[temp2]                          \n\t"
00159          "slt    %[temp2],  %[max_asm], %[sum1]                           \n\t"
00160          "movn   %[sum1],   %[max_asm], %[temp2]                          \n\t"
00161          "sh     %[sum1],   0(%[samples])                                 \n\t"
00162 
00163         : [w_asm] "=&r" (w_asm), [p_asm] "=&r" (p_asm), [w_asm1] "=&r" (w_asm1),
00164           [p_asm1] "=&r" (p_asm1), [temp1] "+r" (temp1), [temp2] "+r" (temp2),
00165           [w_asm2] "=&r" (w_asm2), [p_asm2] "=&r" (p_asm2),
00166           [sum1] "+r" (sum1), [w] "+r" (w), [temp3] "+r" (temp3)
00167         : [p] "r" (p), [samples] "r" (samples), [min_asm] "r" (min_asm),
00168           [max_asm] "r" (max_asm)
00169         : "hi","lo"
00170      );
00171 
00172      samples += incr;
00173 
00174     /* we calculate two samples at the same time to avoid one memory
00175        access per two sample */
00176 
00177     for(j = 1; j < 16; j++) {
00178         __asm__ volatile (
00179              "mthi   $0,         $ac1                                      \n\t"
00180              "mtlo   $0,         $ac1                                      \n\t"
00181              "mthi   $0                                                    \n\t"
00182              "mtlo   %[temp1]                                              \n\t"
00183              "addi   %[p_temp1], %[p_temp1],       4                       \n\t"
00184              "lw     %[w_asm],   0(%[w])                                   \n\t"
00185              "lw     %[p_asm],   0(%[p_temp1])                             \n\t"
00186              "lw     %[w2_asm],  0(%[w2])                                  \n\t"
00187              "lw     %[w_asm1],  64*4(%[w])                                \n\t"
00188              "lw     %[p_asm1],  64*4(%[p_temp1])                          \n\t"
00189              "lw     %[w2_asm1], 64*4(%[w2])                               \n\t"
00190              "madd   %[w_asm],   %[p_asm]                                  \n\t"
00191              "msub   $ac1,       %[w2_asm],        %[p_asm]                \n\t"
00192              "madd   %[w_asm1],  %[p_asm1]                                 \n\t"
00193              "msub   $ac1,       %[w2_asm1],       %[p_asm1]               \n\t"
00194              "lw     %[w_asm],   128*4(%[w])                               \n\t"
00195              "lw     %[p_asm],   128*4(%[p_temp1])                         \n\t"
00196              "lw     %[w2_asm],  128*4(%[w2])                              \n\t"
00197              "lw     %[w_asm1],  192*4(%[w])                               \n\t"
00198              "lw     %[p_asm1],  192*4(%[p_temp1])                         \n\t"
00199              "lw     %[w2_asm1], 192*4(%[w2])                              \n\t"
00200              "madd   %[w_asm],   %[p_asm]                                  \n\t"
00201              "msub   $ac1,       %[w2_asm],        %[p_asm]                \n\t"
00202              "madd   %[w_asm1],  %[p_asm1]                                 \n\t"
00203              "msub   $ac1,       %[w2_asm1],       %[p_asm1]               \n\t"
00204              "lw     %[w_asm],   256*4(%[w])                               \n\t"
00205              "lw     %[p_asm],   256*4(%[p_temp1])                         \n\t"
00206              "lw     %[w2_asm],  256*4(%[w2])                              \n\t"
00207              "lw     %[w_asm1],  320*4(%[w])                               \n\t"
00208              "lw     %[p_asm1],  320*4(%[p_temp1])                         \n\t"
00209              "lw     %[w2_asm1], 320*4(%[w2])                              \n\t"
00210              "madd   %[w_asm],   %[p_asm]                                  \n\t"
00211              "msub   $ac1,       %[w2_asm],        %[p_asm]                \n\t"
00212              "madd   %[w_asm1],  %[p_asm1]                                 \n\t"
00213              "msub   $ac1,       %[w2_asm1],       %[p_asm1]               \n\t"
00214              "lw     %[w_asm],   384*4(%[w])                               \n\t"
00215              "lw     %[p_asm],   384*4(%[p_temp1])                         \n\t"
00216              "lw     %[w2_asm],  384*4(%[w2])                              \n\t"
00217              "lw     %[w_asm1],  448*4(%[w])                               \n\t"
00218              "lw     %[p_asm1],  448*4(%[p_temp1])                         \n\t"
00219              "lw     %[w2_asm1], 448*4(%[w2])                              \n\t"
00220              "madd   %[w_asm],   %[p_asm]                                  \n\t"
00221              "msub   $ac1,       %[w2_asm],        %[p_asm]                \n\t"
00222              "madd   %[w_asm1],  %[p_asm1]                                 \n\t"
00223              "msub   $ac1,       %[w2_asm1],       %[p_asm1]               \n\t"
00224              "addi   %[p_temp2], %[p_temp2],       -4                      \n\t"
00225              "lw     %[w_asm],   32*4(%[w])                                \n\t"
00226              "lw     %[p_asm],   0(%[p_temp2])                             \n\t"
00227              "lw     %[w2_asm],  32*4(%[w2])                               \n\t"
00228              "lw     %[w_asm1],  96*4(%[w])                                \n\t"
00229              "lw     %[p_asm1],  64*4(%[p_temp2])                          \n\t"
00230              "lw     %[w2_asm1], 96*4(%[w2])                               \n\t"
00231              "msub   %[w_asm],   %[p_asm]                                  \n\t"
00232              "msub   $ac1,       %[w2_asm],        %[p_asm]                \n\t"
00233              "msub   %[w_asm1],  %[p_asm1]                                 \n\t"
00234              "msub   $ac1,       %[w2_asm1],       %[p_asm1]               \n\t"
00235              "lw     %[w_asm],   160*4(%[w])                               \n\t"
00236              "lw     %[p_asm],   128*4(%[p_temp2])                         \n\t"
00237              "lw     %[w2_asm],  160*4(%[w2])                              \n\t"
00238              "lw     %[w_asm1],  224*4(%[w])                               \n\t"
00239              "lw     %[p_asm1],  192*4(%[p_temp2])                         \n\t"
00240              "lw     %[w2_asm1], 224*4(%[w2])                              \n\t"
00241              "msub   %[w_asm],   %[p_asm]                                  \n\t"
00242              "msub   $ac1,       %[w2_asm],        %[p_asm]                \n\t"
00243              "msub   %[w_asm1],  %[p_asm1]                                 \n\t"
00244              "msub   $ac1,       %[w2_asm1],       %[p_asm1]               \n\t"
00245              "lw     %[w_asm],   288*4(%[w])                               \n\t"
00246              "lw     %[p_asm],   256*4(%[p_temp2])                         \n\t"
00247              "lw     %[w2_asm],  288*4(%[w2])                              \n\t"
00248              "lw     %[w_asm1],  352*4(%[w])                               \n\t"
00249              "lw     %[p_asm1],  320*4(%[p_temp2])                         \n\t"
00250              "lw     %[w2_asm1], 352*4(%[w2])                              \n\t"
00251              "msub   %[w_asm],   %[p_asm]                                  \n\t"
00252              "msub   $ac1,       %[w2_asm],        %[p_asm]                \n\t"
00253              "msub   %[w_asm1],  %[p_asm1]                                 \n\t"
00254              "msub   $ac1,       %[w2_asm1],       %[p_asm1]               \n\t"
00255              "lw     %[w_asm],   416*4(%[w])                               \n\t"
00256              "lw     %[p_asm],   384*4(%[p_temp2])                         \n\t"
00257              "lw     %[w2_asm],  416*4(%[w2])                              \n\t"
00258              "lw     %[w_asm1],  480*4(%[w])                               \n\t"
00259              "lw     %[p_asm1],  448*4(%[p_temp2])                         \n\t"
00260              "lw     %[w2_asm1], 480*4(%[w2])                              \n\t"
00261              "msub   %[w_asm],   %[p_asm]                                  \n\t"
00262              "msub   %[w_asm1],  %[p_asm1]                                 \n\t"
00263              "msub   $ac1,       %[w2_asm],        %[p_asm]                \n\t"
00264              "msub   $ac1,       %[w2_asm1],       %[p_asm1]               \n\t"
00265              "addi   %[w],       %[w],             4                       \n\t"
00266              "addi   %[w2],      %[w2],            -4                      \n\t"
00267              "mflo   %[temp2]                                              \n\t"
00268              "extr.w %[sum1],    $ac0,             24                      \n\t"
00269              "li     %[temp3],   1                                         \n\t"
00270              "and    %[temp1],   %[temp2],         0x00ffffff              \n\t"
00271              "madd   $ac1,       %[temp1],         %[temp3]                \n\t"
00272              "slt    %[temp2],   %[sum1],          %[min_asm]              \n\t"
00273              "movn   %[sum1],    %[min_asm],       %[temp2]                \n\t"
00274              "slt    %[temp2],   %[max_asm],       %[sum1]                 \n\t"
00275              "movn   %[sum1],    %[max_asm],       %[temp2]                \n\t"
00276              "sh     %[sum1],    0(%[samples])                             \n\t"
00277              "mflo   %[temp3],   $ac1                                      \n\t"
00278              "extr.w %[sum1],    $ac1,             24                      \n\t"
00279              "and    %[temp1],   %[temp3],         0x00ffffff              \n\t"
00280              "slt    %[temp2],   %[sum1],          %[min_asm]              \n\t"
00281              "movn   %[sum1],    %[min_asm],       %[temp2]                \n\t"
00282              "slt    %[temp2],   %[max_asm],       %[sum1]                 \n\t"
00283              "movn   %[sum1],    %[max_asm],       %[temp2]                \n\t"
00284              "sh     %[sum1],    0(%[samples2])                            \n\t"
00285 
00286             : [w_asm] "=&r" (w_asm), [p_asm] "=&r" (p_asm), [w_asm1] "=&r" (w_asm1),
00287               [p_asm1] "=&r" (p_asm1), [w2_asm1] "=&r" (w2_asm1),
00288               [w2_asm] "=&r" (w2_asm), [temp1] "+r" (temp1), [temp2] "+r" (temp2),
00289               [p_temp1] "+r" (p_temp1), [p_temp2] "+r" (p_temp2), [sum1] "+r" (sum1),
00290               [w] "+r" (w), [w2] "+r" (w2), [samples] "+r" (samples),
00291               [samples2] "+r" (samples2), [temp3] "+r" (temp3)
00292             : [min_asm] "r" (min_asm), [max_asm] "r" (max_asm)
00293             : "hi", "lo"
00294         );
00295 
00296         samples += incr;
00297         samples2 -= incr;
00298     }
00299 
00300     p = synth_buf + 32;
00301 
00302     __asm__ volatile (
00303         "mthi   $0                                                        \n\t"
00304         "mtlo   %[temp1]                                                  \n\t"
00305         "lw     %[w_asm],  32*4(%[w])                                     \n\t"
00306         "lw     %[p_asm],  0(%[p])                                        \n\t"
00307         "lw     %[w_asm1], 96*4(%[w])                                     \n\t"
00308         "lw     %[p_asm1], 64*4(%[p])                                     \n\t"
00309         "lw     %[w_asm2], 160*4(%[w])                                    \n\t"
00310         "lw     %[p_asm2], 128*4(%[p])                                    \n\t"
00311         "msub   %[w_asm],  %[p_asm]                                       \n\t"
00312         "msub   %[w_asm1], %[p_asm1]                                      \n\t"
00313         "msub   %[w_asm2], %[p_asm2]                                      \n\t"
00314         "lw     %[w_asm],  224*4(%[w])                                    \n\t"
00315         "lw     %[p_asm],  192*4(%[p])                                    \n\t"
00316         "lw     %[w_asm1], 288*4(%[w])                                    \n\t"
00317         "lw     %[p_asm1], 256*4(%[p])                                    \n\t"
00318         "lw     %[w_asm2], 352*4(%[w])                                    \n\t"
00319         "lw     %[p_asm2], 320*4(%[p])                                    \n\t"
00320         "msub   %[w_asm],  %[p_asm]                                       \n\t"
00321         "msub   %[w_asm1], %[p_asm1]                                      \n\t"
00322         "msub   %[w_asm2], %[p_asm2]                                      \n\t"
00323         "lw     %[w_asm],  416*4(%[w])                                    \n\t"
00324         "lw     %[p_asm],  384*4(%[p])                                    \n\t"
00325         "lw     %[w_asm1], 480*4(%[w])                                    \n\t"
00326         "lw     %[p_asm1], 448*4(%[p])                                    \n\t"
00327         "msub   %[w_asm],  %[p_asm]                                       \n\t"
00328         "msub   %[w_asm1], %[p_asm1]                                      \n\t"
00329         "extr.w %[sum1],   $ac0,       24                                 \n\t"
00330         "mflo   %[temp2]                                                  \n\t"
00331         "and    %[temp1],  %[temp2],   0x00ffffff                         \n\t"
00332         "slt    %[temp2],  %[sum1],    %[min_asm]                         \n\t"
00333         "movn   %[sum1],   %[min_asm], %[temp2]                           \n\t"
00334         "slt    %[temp2],  %[max_asm], %[sum1]                            \n\t"
00335         "movn   %[sum1],   %[max_asm], %[temp2]                           \n\t"
00336         "sh     %[sum1],   0(%[samples])                                  \n\t"
00337 
00338         : [w_asm] "=&r" (w_asm), [p_asm] "=&r" (p_asm), [w_asm1] "=&r" (w_asm1),
00339           [p_asm1] "=&r" (p_asm1), [temp1] "+r" (temp1), [temp2] "+r" (temp2),
00340           [w_asm2] "=&r" (w_asm2), [p_asm2] "=&r" (p_asm2), [sum1] "+r" (sum1)
00341         : [w] "r" (w), [p] "r" (p), [samples] "r" (samples), [min_asm] "r" (min_asm),
00342           [max_asm] "r" (max_asm)
00343         : "hi", "lo"
00344      );
00345 
00346     *dither_state= temp1;
00347 }
00348 
00349 static void imdct36_mips_fixed(int *out, int *buf, int *in, int *win)
00350 {
00351     int j;
00352     int t0, t1, t2, t3, s0, s1, s2, s3;
00353     int tmp[18], *tmp1, *in1;
00354     /* temporary variables */
00355     int temp_reg1, temp_reg2, temp_reg3, temp_reg4, temp_reg5, temp_reg6;
00356     int t4, t5, t6, t8, t7;
00357 
00358    /* values defined in macros and tables are
00359     * eliminated - they are directly loaded in appropriate variables
00360     */
00361     int const C_1  =  4229717092; /* cos(pi*1/18)*2  */
00362     int const C_2  =  4035949074; /* cos(pi*2/18)*2  */
00363     int const C_3  =  575416510;  /* -cos(pi*3/18)*2 */
00364     int const C_3A =  3719550786; /* cos(pi*3/18)*2  */
00365     int const C_4  =  1004831466; /* -cos(pi*4/18)*2 */
00366     int const C_5  =  1534215534; /* -cos(pi*5/18)*2 */
00367     int const C_7  = -1468965330; /* -cos(pi*7/18)*2 */
00368     int const C_8  = -745813244;  /* -cos(pi*8/18)*2 */
00369 
00370    /*
00371     * instructions of the first two loops are reorganized and loops are unrolled,
00372     * in order to eliminate unnecessary readings and writings in array
00373     */
00374 
00375     __asm__ volatile (
00376         "lw   %[t1], 17*4(%[in])                                         \n\t"
00377         "lw   %[t2], 16*4(%[in])                                         \n\t"
00378         "lw   %[t3], 15*4(%[in])                                         \n\t"
00379         "lw   %[t4], 14*4(%[in])                                         \n\t"
00380         "addu %[t1], %[t1],      %[t2]                                   \n\t"
00381         "addu %[t2], %[t2],      %[t3]                                   \n\t"
00382         "addu %[t3], %[t3],      %[t4]                                   \n\t"
00383         "lw   %[t5], 13*4(%[in])                                         \n\t"
00384         "addu %[t1], %[t1],      %[t3]                                   \n\t"
00385         "sw   %[t2], 16*4(%[in])                                         \n\t"
00386         "lw   %[t6], 12*4(%[in])                                         \n\t"
00387         "sw   %[t1], 17*4(%[in])                                         \n\t"
00388         "addu %[t4], %[t4],      %[t5]                                   \n\t"
00389         "addu %[t5], %[t5],      %[t6]                                   \n\t"
00390         "lw   %[t7], 11*4(%[in])                                         \n\t"
00391         "addu %[t3], %[t3],      %[t5]                                   \n\t"
00392         "sw   %[t4], 14*4(%[in])                                         \n\t"
00393         "lw   %[t8], 10*4(%[in])                                         \n\t"
00394         "sw   %[t3], 15*4(%[in])                                         \n\t"
00395         "addu %[t6], %[t6],      %[t7]                                   \n\t"
00396         "addu %[t7], %[t7],      %[t8]                                   \n\t"
00397         "sw   %[t6], 12*4(%[in])                                         \n\t"
00398         "addu %[t5], %[t5],      %[t7]                                   \n\t"
00399         "lw   %[t1], 9*4(%[in])                                          \n\t"
00400         "lw   %[t2], 8*4(%[in])                                          \n\t"
00401         "sw   %[t5], 13*4(%[in])                                         \n\t"
00402         "addu %[t8], %[t8],      %[t1]                                   \n\t"
00403         "addu %[t1], %[t1],      %[t2]                                   \n\t"
00404         "sw   %[t8], 10*4(%[in])                                         \n\t"
00405         "addu %[t7], %[t7],      %[t1]                                   \n\t"
00406         "lw   %[t3], 7*4(%[in])                                          \n\t"
00407         "lw   %[t4], 6*4(%[in])                                          \n\t"
00408         "sw   %[t7], 11*4(%[in])                                         \n\t"
00409         "addu %[t2], %[t2],      %[t3]                                   \n\t"
00410         "addu %[t3], %[t3],      %[t4]                                   \n\t"
00411         "sw   %[t2], 8*4(%[in])                                          \n\t"
00412         "addu %[t1], %[t1],      %[t3]                                   \n\t"
00413         "lw   %[t5], 5*4(%[in])                                          \n\t"
00414         "lw   %[t6], 4*4(%[in])                                          \n\t"
00415         "sw   %[t1], 9*4(%[in])                                          \n\t"
00416         "addu %[t4], %[t4],      %[t5]                                   \n\t"
00417         "addu %[t5], %[t5],      %[t6]                                   \n\t"
00418         "sw   %[t4], 6*4(%[in])                                          \n\t"
00419         "addu %[t3], %[t3],      %[t5]                                   \n\t"
00420         "lw   %[t7], 3*4(%[in])                                          \n\t"
00421         "lw   %[t8], 2*4(%[in])                                          \n\t"
00422         "sw   %[t3], 7*4(%[in])                                          \n\t"
00423         "addu %[t6], %[t6],      %[t7]                                   \n\t"
00424         "addu %[t7], %[t7],      %[t8]                                   \n\t"
00425         "sw   %[t6], 4*4(%[in])                                          \n\t"
00426         "addu %[t5], %[t5],      %[t7]                                   \n\t"
00427         "lw   %[t1], 1*4(%[in])                                          \n\t"
00428         "lw   %[t2], 0*4(%[in])                                          \n\t"
00429         "sw   %[t5], 5*4(%[in])                                          \n\t"
00430         "addu %[t8], %[t8],      %[t1]                                   \n\t"
00431         "addu %[t1], %[t1],      %[t2]                                   \n\t"
00432         "sw   %[t8], 2*4(%[in])                                          \n\t"
00433         "addu %[t7], %[t7],      %[t1]                                   \n\t"
00434         "sw   %[t7], 3*4(%[in])                                          \n\t"
00435         "sw   %[t1], 1*4(%[in])                                          \n\t"
00436 
00437         : [in] "+r" (in), [t1] "=&r" (t1), [t2] "=&r" (t2), [t3] "=&r" (t3),
00438           [t4] "=&r" (t4), [t5] "=&r" (t5), [t6] "=&r" (t6),
00439           [t7] "=&r" (t7), [t8] "=&r" (t8)
00440     );
00441 
00442     for(j = 0; j < 2; j++) {
00443 
00444         tmp1 = tmp + j;
00445         in1 = in + j;
00446 
00472         __asm__ volatile (
00473             "lw    %[t7],        4*4(%[in1])                               \n\t"
00474             "lw    %[t8],        8*4(%[in1])                               \n\t"
00475             "lw    %[t6],        16*4(%[in1])                              \n\t"
00476             "lw    %[t4],        0*4(%[in1])                               \n\t"
00477             "addu  %[temp_reg2], %[t7],        %[t8]                       \n\t"
00478             "addu  %[t2],        %[t6],        %[t8]                       \n\t"
00479             "multu %[C_2],       %[temp_reg2]                              \n\t"
00480             "lw    %[t5],        12*4(%[in1])                              \n\t"
00481             "sub   %[t2],        %[t2],        %[t7]                       \n\t"
00482             "sub   %[t1],        %[t4],        %[t5]                       \n\t"
00483             "sra   %[t3],        %[t5],        1                           \n\t"
00484             "sra   %[temp_reg1], %[t2],        1                           \n\t"
00485             "addu  %[t3],        %[t3],        %[t4]                       \n\t"
00486             "sub   %[temp_reg1], %[t1],        %[temp_reg1]                \n\t"
00487             "sra   %[temp_reg2], %[temp_reg2], 31                          \n\t"
00488             "sw    %[temp_reg1], 6*4(%[tmp1])                              \n\t"
00489             "move  %[t0],        $0                                        \n\t"
00490             "movn  %[t0],        %[C_2],       %[temp_reg2]                \n\t"
00491             "mfhi  %[temp_reg1]                                            \n\t"
00492             "addu  %[t1],        %[t1],        %[t2]                       \n\t"
00493             "sw    %[t1],        16*4(%[tmp1])                             \n\t"
00494             "sub   %[temp_reg4], %[t8],        %[t6]                       \n\t"
00495             "add   %[temp_reg2], %[t7],        %[t6]                       \n\t"
00496             "mult  $ac1,         %[C_8],       %[temp_reg4]                \n\t"
00497             "multu $ac2,         %[C_4],       %[temp_reg2]                \n\t"
00498             "sub   %[t0],        %[temp_reg1], %[t0]                       \n\t"
00499             "sra   %[temp_reg1], %[temp_reg2], 31                          \n\t"
00500             "move  %[t2],        $0                                        \n\t"
00501             "movn  %[t2],        %[C_4],       %[temp_reg1]                \n\t"
00502             "mfhi  %[t1],        $ac1                                      \n\t"
00503             "mfhi  %[temp_reg1], $ac2                                      \n\t"
00504             "lw    %[t6],        10*4(%[in1])                              \n\t"
00505             "lw    %[t8],        14*4(%[in1])                              \n\t"
00506             "lw    %[t7],        2*4(%[in1])                               \n\t"
00507             "lw    %[t4],        6*4(%[in1])                               \n\t"
00508             "sub   %[temp_reg3], %[t3],        %[t0]                       \n\t"
00509             "add   %[temp_reg4], %[t3],        %[t0]                       \n\t"
00510             "sub   %[temp_reg1], %[temp_reg1], %[temp_reg2]                \n\t"
00511             "add   %[temp_reg4], %[temp_reg4], %[t1]                       \n\t"
00512             "sub   %[t2],        %[temp_reg1], %[t2]                       \n\t"
00513             "sw    %[temp_reg4], 2*4(%[tmp1])                              \n\t"
00514             "sub   %[temp_reg3], %[temp_reg3], %[t2]                       \n\t"
00515             "add   %[temp_reg1], %[t3],        %[t2]                       \n\t"
00516             "sw    %[temp_reg3], 10*4(%[tmp1])                             \n\t"
00517             "sub   %[temp_reg1], %[temp_reg1], %[t1]                       \n\t"
00518             "addu  %[temp_reg2], %[t6],        %[t8]                       \n\t"
00519             "sw    %[temp_reg1], 14*4(%[tmp1])                             \n\t"
00520             "sub   %[temp_reg2], %[temp_reg2], %[t7]                       \n\t"
00521             "addu  %[temp_reg3], %[t7],        %[t6]                       \n\t"
00522             "multu $ac3,         %[C_3],       %[temp_reg2]                \n\t"
00523             "multu %[C_1],       %[temp_reg3]                              \n\t"
00524             "sra   %[temp_reg1], %[temp_reg2], 31                          \n\t"
00525             "move  %[t1],        $0                                        \n\t"
00526             "sra   %[temp_reg3], %[temp_reg3], 31                          \n\t"
00527             "movn  %[t1],        %[C_3],       %[temp_reg1]                \n\t"
00528             "mfhi  %[temp_reg1], $ac3                                      \n\t"
00529             "mfhi  %[temp_reg4]                                            \n\t"
00530             "move  %[t2],        $0                                        \n\t"
00531             "movn  %[t2],        %[C_1],       %[temp_reg3]                \n\t"
00532             "sub   %[temp_reg3], %[t6],        %[t8]                       \n\t"
00533             "sub   %[t2],        %[temp_reg4], %[t2]                       \n\t"
00534             "multu $ac1,         %[C_7],       %[temp_reg3]                \n\t"
00535             "sub   %[temp_reg1], %[temp_reg1], %[temp_reg2]                \n\t"
00536             "sra   %[temp_reg4], %[temp_reg3], 31                          \n\t"
00537             "sub   %[t1],        %[temp_reg1], %[t1]                       \n\t"
00538             "move  %[t3],        $0                                        \n\t"
00539             "sw    %[t1],        4*4(%[tmp1])                              \n\t"
00540             "movn  %[t3],        %[C_7],       %[temp_reg4]                \n\t"
00541             "multu $ac2,         %[C_3A],      %[t4]                       \n\t"
00542             "add   %[temp_reg2], %[t7],        %[t8]                       \n\t"
00543             "move  %[t1],        $0                                        \n\t"
00544             "mfhi  %[temp_reg4], $ac1                                      \n\t"
00545             "multu $ac3,%[C_5],  %[temp_reg2]                              \n\t"
00546             "move  %[t0],        $0                                        \n\t"
00547             "sra   %[temp_reg1], %[temp_reg2], 31                          \n\t"
00548             "movn  %[t1],%[C_5], %[temp_reg1]                              \n\t"
00549             "sub   %[temp_reg4], %[temp_reg4], %[temp_reg3]                \n\t"
00550             "mfhi  %[temp_reg1], $ac3                                      \n\t"
00551             "sra   %[temp_reg3], %[t4],        31                          \n\t"
00552             "movn  %[t0],        %[C_3A],      %[temp_reg3]                \n\t"
00553             "mfhi  %[temp_reg3], $ac2                                      \n\t"
00554             "sub   %[t3],        %[temp_reg4], %[t3]                       \n\t"
00555             "add   %[temp_reg4], %[t3],        %[t2]                       \n\t"
00556             "sub   %[temp_reg1], %[temp_reg1], %[temp_reg2]                \n\t"
00557             "sub   %[t1],        %[temp_reg1], %[t1]                       \n\t"
00558             "sub   %[t0],        %[temp_reg3], %[t0]                       \n\t"
00559             "add   %[temp_reg1], %[t2],        %[t1]                       \n\t"
00560             "add   %[temp_reg4], %[temp_reg4], %[t0]                       \n\t"
00561             "sub   %[temp_reg2], %[t3],        %[t1]                       \n\t"
00562             "sw    %[temp_reg4], 0*4(%[tmp1])                              \n\t"
00563             "sub   %[temp_reg1], %[temp_reg1], %[t0]                       \n\t"
00564             "sub   %[temp_reg2], %[temp_reg2], %[t0]                       \n\t"
00565             "sw    %[temp_reg1], 12*4(%[tmp1])                             \n\t"
00566             "sw    %[temp_reg2], 8*4(%[tmp1])                              \n\t"
00567 
00568             : [t7] "=&r" (t7), [temp_reg1] "=&r" (temp_reg1),
00569               [temp_reg2] "=&r" (temp_reg2), [temp_reg4] "=&r" (temp_reg4),
00570               [temp_reg3] "=&r" (temp_reg3), [t8] "=&r" (t8), [t0] "=&r" (t0),
00571               [t4] "=&r" (t4), [t5] "=&r" (t5), [t6] "=&r"(t6), [t2] "=&r" (t2),
00572               [t3] "=&r" (t3), [t1] "=&r" (t1)
00573             : [C_2] "r" (C_2), [in1] "r" (in1), [tmp1] "r" (tmp1), [C_8] "r" (C_8),
00574               [C_4] "r" (C_4), [C_3] "r" (C_3), [C_1] "r" (C_1), [C_7] "r" (C_7),
00575               [C_3A] "r" (C_3A), [C_5] "r" (C_5)
00576             : "hi", "lo"
00577          );
00578     }
00579 
00588     __asm__ volatile (
00589         "lw     %[t2],        1*4(%[tmp])                                  \n\t"
00590         "lw     %[t3],        3*4(%[tmp])                                  \n\t"
00591         "lw     %[t0],        0*4(%[tmp])                                  \n\t"
00592         "lw     %[t1],        2*4(%[tmp])                                  \n\t"
00593         "addu   %[temp_reg1], %[t3],        %[t2]                          \n\t"
00594         "li     %[temp_reg2], 0x807D2B1E                                   \n\t"
00595         "move   %[s1],        $0                                           \n\t"
00596         "multu  %[temp_reg2], %[temp_reg1]                                 \n\t"
00597         "sra    %[temp_reg1], %[temp_reg1], 31                             \n\t"
00598         "movn   %[s1],        %[temp_reg2], %[temp_reg1]                   \n\t"
00599         "sub    %[temp_reg3], %[t3],        %[t2]                          \n\t"
00600         "li     %[temp_reg4], 0x2de5151                                    \n\t"
00601         "mfhi   %[temp_reg2]                                               \n\t"
00602         "addu   %[s0],        %[t1],        %[t0]                          \n\t"
00603         "lw     %[temp_reg5], 9*4(%[win])                                  \n\t"
00604         "mult   $ac1,         %[temp_reg4], %[temp_reg3]                   \n\t"
00605         "lw     %[temp_reg6], 4*9*4(%[buf])                                \n\t"
00606         "sub    %[s2],        %[t1],        %[t0]                          \n\t"
00607         "lw     %[temp_reg3], 29*4(%[win])                                 \n\t"
00608         "subu   %[s1],        %[temp_reg2], %[s1]                          \n\t"
00609         "lw     %[temp_reg4], 28*4(%[win])                                 \n\t"
00610         "add    %[t0],        %[s0],        %[s1]                          \n\t"
00611         "extr.w %[s3],        $ac1,23                                      \n\t"
00612         "mult   $ac2,         %[t0],        %[temp_reg3]                   \n\t"
00613         "sub    %[t1],        %[s0],        %[s1]                          \n\t"
00614         "lw     %[temp_reg1], 4*8*4(%[buf])                                \n\t"
00615         "mult   %[t1],        %[temp_reg5]                                 \n\t"
00616         "lw     %[temp_reg2], 8*4(%[win])                                  \n\t"
00617         "mfhi   %[temp_reg3], $ac2                                         \n\t"
00618         "mult   $ac3,         %[t0],        %[temp_reg4]                   \n\t"
00619         "add    %[t0],        %[s2],        %[s3]                          \n\t"
00620         "mfhi   %[temp_reg5]                                               \n\t"
00621         "mult   $ac1,         %[t1],        %[temp_reg2]                   \n\t"
00622         "sub    %[t1],        %[s2],        %[s3]                          \n\t"
00623         "sw     %[temp_reg3], 4*9*4(%[buf])                                \n\t"
00624         "mfhi   %[temp_reg4], $ac3                                         \n\t"
00625         "lw     %[temp_reg3], 37*4(%[win])                                 \n\t"
00626         "mfhi   %[temp_reg2], $ac1                                         \n\t"
00627         "add    %[temp_reg5], %[temp_reg5], %[temp_reg6]                   \n\t"
00628         "lw     %[temp_reg6], 17*4(%[win])                                 \n\t"
00629         "sw     %[temp_reg5], 32*9*4(%[out])                               \n\t"
00630         "sw     %[temp_reg4], 4*8*4(%[buf])                                \n\t"
00631         "mult   %[t1],        %[temp_reg6]                                 \n\t"
00632         "add    %[temp_reg1], %[temp_reg1], %[temp_reg2]                   \n\t"
00633         "lw     %[temp_reg2], 0*4(%[win])                                  \n\t"
00634         "lw     %[temp_reg5], 4*17*4(%[buf])                               \n\t"
00635         "sw     %[temp_reg1], 8*32*4(%[out])                               \n\t"
00636         "mfhi   %[temp_reg6]                                               \n\t"
00637         "mult   $ac1,         %[t1],        %[temp_reg2]                   \n\t"
00638         "lw     %[temp_reg4], 20*4(%[win])                                 \n\t"
00639         "lw     %[temp_reg1], 0(%[buf])                                    \n\t"
00640         "mult   $ac2,         %[t0],        %[temp_reg3]                   \n\t"
00641         "mult   %[t0],        %[temp_reg4]                                 \n\t"
00642         "mfhi   %[temp_reg2], $ac1                                         \n\t"
00643         "lw     %[t0],        4*4(%[tmp])                                  \n\t"
00644         "add    %[temp_reg5], %[temp_reg5], %[temp_reg6]                   \n\t"
00645         "mfhi   %[temp_reg3], $ac2                                         \n\t"
00646         "mfhi   %[temp_reg4]                                               \n\t"
00647         "sw     %[temp_reg5], 17*32*4(%[out])                              \n\t"
00648         "lw     %[t1],        6*4(%[tmp])                                  \n\t"
00649         "add    %[temp_reg1], %[temp_reg1], %[temp_reg2]                   \n\t"
00650         "lw     %[t2],        5*4(%[tmp])                                  \n\t"
00651         "sw     %[temp_reg1], 0*32*4(%[out])                               \n\t"
00652         "addu   %[s0],        %[t1],        %[t0]                          \n\t"
00653         "sw     %[temp_reg3], 4*17*4(%[buf])                               \n\t"
00654         "lw     %[t3],        7*4(%[tmp])                                  \n\t"
00655         "sub    %[s2],        %[t1],        %[t0]                          \n\t"
00656         "sw     %[temp_reg4], 0(%[buf])                                    \n\t"
00657         "addu   %[temp_reg5], %[t3],        %[t2]                          \n\t"
00658         "li     %[temp_reg6], 0x8483EE0C                                   \n\t"
00659         "move   %[s1],        $0                                           \n\t"
00660         "multu  %[temp_reg6], %[temp_reg5]                                 \n\t"
00661         "sub    %[temp_reg1], %[t3],        %[t2]                          \n\t"
00662         "li     %[temp_reg2], 0xf746ea                                     \n\t"
00663         "sra    %[temp_reg5], %[temp_reg5], 31                             \n\t"
00664         "mult   $ac1,         %[temp_reg2], %[temp_reg1]                   \n\t"
00665         "movn   %[s1],        %[temp_reg6], %[temp_reg5]                   \n\t"
00666         "mfhi   %[temp_reg5]                                               \n\t"
00667         "lw     %[temp_reg3], 10*4(%[win])                                 \n\t"
00668         "lw     %[temp_reg4], 4*10*4(%[buf])                               \n\t"
00669         "extr.w %[s3],        $ac1,         23                             \n\t"
00670         "lw     %[temp_reg1], 4*7*4(%[buf])                                \n\t"
00671         "lw     %[temp_reg2], 7*4(%[win])                                  \n\t"
00672         "lw     %[temp_reg6], 30*4(%[win])                                 \n\t"
00673         "subu   %[s1],        %[temp_reg5], %[s1]                          \n\t"
00674         "sub    %[t1],        %[s0],        %[s1]                          \n\t"
00675         "add    %[t0],        %[s0],        %[s1]                          \n\t"
00676         "mult   $ac2,         %[t1],        %[temp_reg3]                   \n\t"
00677         "mult   $ac3,         %[t1],        %[temp_reg2]                   \n\t"
00678         "mult   %[t0],        %[temp_reg6]                                 \n\t"
00679         "lw     %[temp_reg5], 27*4(%[win])                                 \n\t"
00680         "mult   $ac1,         %[t0],        %[temp_reg5]                   \n\t"
00681         "mfhi   %[temp_reg3], $ac2                                         \n\t"
00682         "mfhi   %[temp_reg2], $ac3                                         \n\t"
00683         "mfhi   %[temp_reg6]                                               \n\t"
00684         "add    %[t0],        %[s2],        %[s3]                          \n\t"
00685         "sub    %[t1],        %[s2],        %[s3]                          \n\t"
00686         "add    %[temp_reg3], %[temp_reg3], %[temp_reg4]                   \n\t"
00687         "lw     %[temp_reg4], 16*4(%[win])                                 \n\t"
00688         "mfhi   %[temp_reg5], $ac1                                         \n\t"
00689         "sw     %[temp_reg3], 32*10*4(%[out])                              \n\t"
00690         "add    %[temp_reg1], %[temp_reg1], %[temp_reg2]                   \n\t"
00691         "lw     %[temp_reg3], 4*16*4(%[buf])                               \n\t"
00692         "sw     %[temp_reg6], 4*10*4(%[buf])                               \n\t"
00693         "sw     %[temp_reg1], 7*32*4(%[out])                               \n\t"
00694         "mult   $ac2,         %[t1],        %[temp_reg4]                   \n\t"
00695         "sw     %[temp_reg5], 4*7*4(%[buf])                                \n\t"
00696         "lw     %[temp_reg6], 1*4(%[win])                                  \n\t"
00697         "lw     %[temp_reg5], 4*1*4(%[buf])                                \n\t"
00698         "lw     %[temp_reg1], 36*4(%[win])                                 \n\t"
00699         "mult   $ac3,         %[t1],        %[temp_reg6]                   \n\t"
00700         "lw     %[temp_reg2], 21*4(%[win])                                 \n\t"
00701         "mfhi   %[temp_reg4], $ac2                                         \n\t"
00702         "mult   %[t0],        %[temp_reg1]                                 \n\t"
00703         "mult   $ac1,         %[t0],%[temp_reg2]                           \n\t"
00704         "lw     %[t0],        8*4(%[tmp])                                  \n\t"
00705         "mfhi   %[temp_reg6], $ac3                                         \n\t"
00706         "lw     %[t1],        10*4(%[tmp])                                 \n\t"
00707         "lw     %[t3],        11*4(%[tmp])                                 \n\t"
00708         "mfhi   %[temp_reg1]                                               \n\t"
00709         "add    %[temp_reg3], %[temp_reg3], %[temp_reg4]                   \n\t"
00710         "lw     %[t2],        9*4(%[tmp])                                  \n\t"
00711         "mfhi   %[temp_reg2], $ac1                                         \n\t"
00712         "add    %[temp_reg5], %[temp_reg5], %[temp_reg6]                   \n\t"
00713         "sw     %[temp_reg3], 16*32*4(%[out])                              \n\t"
00714         "sw     %[temp_reg5], 1*32*4(%[out])                               \n\t"
00715         "sw     %[temp_reg1], 4*16*4(%[buf])                               \n\t"
00716         "addu   %[temp_reg3], %[t3],        %[t2]                          \n\t"
00717         "li     %[temp_reg4], 0x8D3B7CD6                                   \n\t"
00718         "sw     %[temp_reg2], 4*1*4(%[buf])                                \n\t"
00719         "multu  %[temp_reg4],%[temp_reg3]                                  \n\t"
00720         "sra    %[temp_reg3], %[temp_reg3], 31                             \n\t"
00721         "move   %[s1],        $0                                           \n\t"
00722         "movn   %[s1],        %[temp_reg4], %[temp_reg3]                   \n\t"
00723         "addu   %[s0],        %[t1],        %[t0]                          \n\t"
00724         "mfhi   %[temp_reg3]                                               \n\t"
00725         "sub    %[s2],        %[t1],        %[t0]                          \n\t"
00726         "sub    %[temp_reg5], %[t3],        %[t2]                          \n\t"
00727         "li     %[temp_reg6], 0x976fd9                                     \n\t"
00728         "lw     %[temp_reg2], 11*4(%[win])                                 \n\t"
00729         "lw     %[temp_reg1], 4*11*4(%[buf])                               \n\t"
00730         "mult   $ac1,         %[temp_reg6], %[temp_reg5]                   \n\t"
00731         "subu   %[s1],        %[temp_reg3], %[s1]                          \n\t"
00732         "lw     %[temp_reg5], 31*4(%[win])                                 \n\t"
00733         "sub    %[t1],        %[s0],        %[s1]                          \n\t"
00734         "add    %[t0],        %[s0],        %[s1]                          \n\t"
00735         "mult   $ac2,         %[t1],        %[temp_reg2]                   \n\t"
00736         "mult   %[t0],        %[temp_reg5]                                 \n\t"
00737         "lw     %[temp_reg4], 6*4(%[win])                                  \n\t"
00738         "extr.w %[s3],        $ac1,         23                             \n\t"
00739         "lw     %[temp_reg3], 4*6*4(%[buf])                                \n\t"
00740         "mfhi   %[temp_reg2], $ac2                                         \n\t"
00741         "lw     %[temp_reg6], 26*4(%[win])                                 \n\t"
00742         "mfhi   %[temp_reg5]                                               \n\t"
00743         "mult   $ac3,         %[t1],        %[temp_reg4]                   \n\t"
00744         "mult   $ac1,         %[t0],        %[temp_reg6]                   \n\t"
00745         "add    %[t0],        %[s2],        %[s3]                          \n\t"
00746         "sub    %[t1],        %[s2],        %[s3]                          \n\t"
00747         "add    %[temp_reg2], %[temp_reg2], %[temp_reg1]                   \n\t"
00748         "mfhi   %[temp_reg4], $ac3                                         \n\t"
00749         "mfhi   %[temp_reg6], $ac1                                         \n\t"
00750         "sw     %[temp_reg5], 4*11*4(%[buf])                               \n\t"
00751         "sw     %[temp_reg2], 32*11*4(%[out])                              \n\t"
00752         "lw     %[temp_reg1], 4*15*4(%[buf])                               \n\t"
00753         "add    %[temp_reg3], %[temp_reg3], %[temp_reg4]                   \n\t"
00754         "lw     %[temp_reg2], 15*4(%[win])                                 \n\t"
00755         "sw     %[temp_reg3], 6*32*4(%[out])                               \n\t"
00756         "sw     %[temp_reg6], 4*6*4(%[buf])                                \n\t"
00757         "mult   %[t1],        %[temp_reg2]                                 \n\t"
00758         "lw     %[temp_reg3], 2*4(%[win])                                  \n\t"
00759         "lw     %[temp_reg4], 4*2*4(%[buf])                                \n\t"
00760         "lw     %[temp_reg5], 35*4(%[win])                                 \n\t"
00761         "mult   $ac1,         %[t1],        %[temp_reg3]                   \n\t"
00762         "mfhi   %[temp_reg2]                                               \n\t"
00763         "lw     %[temp_reg6], 22*4(%[win])                                 \n\t"
00764         "mult   $ac2,         %[t0],        %[temp_reg5]                   \n\t"
00765         "lw     %[t1],        14*4(%[tmp])                                 \n\t"
00766         "mult   $ac3,         %[t0],        %[temp_reg6]                   \n\t"
00767         "lw     %[t0],        12*4(%[tmp])                                 \n\t"
00768         "mfhi   %[temp_reg3], $ac1                                         \n\t"
00769         "add    %[temp_reg1], %[temp_reg1], %[temp_reg2]                   \n\t"
00770         "mfhi   %[temp_reg5], $ac2                                         \n\t"
00771         "sw     %[temp_reg1], 15*32*4(%[out])                              \n\t"
00772         "mfhi   %[temp_reg6], $ac3                                         \n\t"
00773         "lw     %[t2],        13*4(%[tmp])                                 \n\t"
00774         "lw     %[t3],        15*4(%[tmp])                                 \n\t"
00775         "add    %[temp_reg4], %[temp_reg4], %[temp_reg3]                   \n\t"
00776         "sw     %[temp_reg5], 4*15*4(%[buf])                               \n\t"
00777         "addu   %[temp_reg1], %[t3],        %[t2]                          \n\t"
00778         "li     %[temp_reg2], 0x9C42577C                                   \n\t"
00779         "move   %[s1],        $0                                           \n\t"
00780         "multu  %[temp_reg2], %[temp_reg1]                                 \n\t"
00781         "sw     %[temp_reg4], 2*32*4(%[out])                               \n\t"
00782         "sra    %[temp_reg1], %[temp_reg1], 31                             \n\t"
00783         "movn   %[s1],        %[temp_reg2], %[temp_reg1]                   \n\t"
00784         "sub    %[temp_reg3], %[t3],        %[t2]                          \n\t"
00785         "li     %[temp_reg4], 0x6f94a2                                     \n\t"
00786         "mfhi   %[temp_reg1]                                               \n\t"
00787         "addu   %[s0],        %[t1],        %[t0]                          \n\t"
00788         "sw     %[temp_reg6], 4*2*4(%[buf])                                \n\t"
00789         "mult   $ac1,         %[temp_reg4], %[temp_reg3]                   \n\t"
00790         "sub    %[s2],        %[t1],        %[t0]                          \n\t"
00791         "lw     %[temp_reg5], 12*4(%[win])                                 \n\t"
00792         "lw     %[temp_reg6], 4*12*4(%[buf])                               \n\t"
00793         "subu   %[s1],        %[temp_reg1], %[s1]                          \n\t"
00794         "sub    %[t1],        %[s0],        %[s1]                          \n\t"
00795         "lw     %[temp_reg3], 32*4(%[win])                                 \n\t"
00796         "mult   $ac2,         %[t1],        %[temp_reg5]                   \n\t"
00797         "add    %[t0],        %[s0],        %[s1]                          \n\t"
00798         "extr.w %[s3],        $ac1,         23                             \n\t"
00799         "lw     %[temp_reg2], 5*4(%[win])                                  \n\t"
00800         "mult   %[t0],        %[temp_reg3]                                 \n\t"
00801         "mfhi   %[temp_reg5], $ac2                                         \n\t"
00802         "lw     %[temp_reg4], 25*4(%[win])                                 \n\t"
00803         "lw     %[temp_reg1], 4*5*4(%[buf])                                \n\t"
00804         "mult   $ac3,         %[t1],        %[temp_reg2]                   \n\t"
00805         "mult   $ac1,         %[t0],        %[temp_reg4]                   \n\t"
00806         "mfhi   %[temp_reg3]                                               \n\t"
00807         "add    %[t0],        %[s2],        %[s3]                          \n\t"
00808         "add    %[temp_reg5], %[temp_reg5], %[temp_reg6]                   \n\t"
00809         "mfhi   %[temp_reg2], $ac3                                         \n\t"
00810         "mfhi   %[temp_reg4], $ac1                                         \n\t"
00811         "sub    %[t1],        %[s2],        %[s3]                          \n\t"
00812         "sw     %[temp_reg5], 32*12*4(%[out])                              \n\t"
00813         "sw     %[temp_reg3], 4*12*4(%[buf])                               \n\t"
00814         "lw     %[temp_reg6], 14*4(%[win])                                 \n\t"
00815         "lw     %[temp_reg5], 4*14*4(%[buf])                               \n\t"
00816         "add    %[temp_reg1], %[temp_reg1], %[temp_reg2]                   \n\t"
00817         "sw     %[temp_reg4], 4*5*4(%[buf])                                \n\t"
00818         "sw     %[temp_reg1], 5*32*4(%[out])                               \n\t"
00819         "mult   %[t1],        %[temp_reg6]                                 \n\t"
00820         "lw     %[temp_reg4], 34*4(%[win])                                 \n\t"
00821         "lw     %[temp_reg2], 3*4(%[win])                                  \n\t"
00822         "lw     %[temp_reg1], 4*3*4(%[buf])                                \n\t"
00823         "mult   $ac2,         %[t0],        %[temp_reg4]                   \n\t"
00824         "mfhi   %[temp_reg6]                                               \n\t"
00825         "mult   $ac1,         %[t1],        %[temp_reg2]                   \n\t"
00826         "lw     %[temp_reg3], 23*4(%[win])                                 \n\t"
00827         "lw     %[s0],        16*4(%[tmp])                                 \n\t"
00828         "mfhi   %[temp_reg4], $ac2                                         \n\t"
00829         "lw     %[t1],        17*4(%[tmp])                                 \n\t"
00830         "mult   $ac3,         %[t0],        %[temp_reg3]                   \n\t"
00831         "move   %[s1],        $0                                           \n\t"
00832         "add    %[temp_reg5], %[temp_reg5], %[temp_reg6]                   \n\t"
00833         "mfhi   %[temp_reg2], $ac1                                         \n\t"
00834         "sw     %[temp_reg5], 14*32*4(%[out])                              \n\t"
00835         "sw     %[temp_reg4], 4*14*4(%[buf])                               \n\t"
00836         "mfhi   %[temp_reg3], $ac3                                         \n\t"
00837         "li     %[temp_reg5], 0xB504F334                                   \n\t"
00838         "add    %[temp_reg1], %[temp_reg1], %[temp_reg2]                   \n\t"
00839         "multu  %[temp_reg5], %[t1]                                        \n\t"
00840         "lw     %[temp_reg2], 4*13*4(%[buf])                               \n\t"
00841         "sw     %[temp_reg1], 3*32*4(%[out])                               \n\t"
00842         "sra    %[t1],        %[t1],        31                             \n\t"
00843         "mfhi   %[temp_reg6]                                               \n\t"
00844         "movn   %[s1],        %[temp_reg5], %[t1]                          \n\t"
00845         "sw     %[temp_reg3], 4*3*4(%[buf])                                \n\t"
00846         "lw     %[temp_reg1], 13*4(%[win])                                 \n\t"
00847         "lw     %[temp_reg4], 4*4*4(%[buf])                                \n\t"
00848         "lw     %[temp_reg3], 4*4(%[win])                                  \n\t"
00849         "lw     %[temp_reg5], 33*4(%[win])                                 \n\t"
00850         "subu   %[s1],        %[temp_reg6], %[s1]                          \n\t"
00851         "lw     %[temp_reg6], 24*4(%[win])                                 \n\t"
00852         "sub    %[t1],        %[s0],        %[s1]                          \n\t"
00853         "add    %[t0],        %[s0],        %[s1]                          \n\t"
00854         "mult   $ac1,         %[t1],        %[temp_reg1]                   \n\t"
00855         "mult   $ac2,         %[t1],        %[temp_reg3]                   \n\t"
00856         "mult   $ac3,         %[t0],        %[temp_reg5]                   \n\t"
00857         "mult   %[t0],        %[temp_reg6]                                 \n\t"
00858         "mfhi   %[temp_reg1], $ac1                                         \n\t"
00859         "mfhi   %[temp_reg3], $ac2                                         \n\t"
00860         "mfhi   %[temp_reg5], $ac3                                         \n\t"
00861         "mfhi   %[temp_reg6]                                               \n\t"
00862         "add    %[temp_reg2], %[temp_reg2], %[temp_reg1]                   \n\t"
00863         "add    %[temp_reg4], %[temp_reg4], %[temp_reg3]                   \n\t"
00864         "sw     %[temp_reg2], 13*32*4(%[out])                              \n\t"
00865         "sw     %[temp_reg4], 4*32*4(%[out])                               \n\t"
00866         "sw     %[temp_reg5], 4*13*4(%[buf])                               \n\t"
00867         "sw     %[temp_reg6], 4*4*4(%[buf])                                \n\t"
00868 
00869         : [t0] "=&r" (t0), [t1] "=&r" (t1), [t2] "=&r" (t2), [t3] "=&r" (t3),
00870           [s0] "=&r" (s0), [s2] "=&r" (s2), [temp_reg1] "=&r" (temp_reg1),
00871           [temp_reg2] "=&r" (temp_reg2), [s1] "=&r" (s1), [s3] "=&r" (s3),
00872           [temp_reg3] "=&r" (temp_reg3), [temp_reg4] "=&r" (temp_reg4),
00873           [temp_reg5] "=&r" (temp_reg5), [temp_reg6] "=&r" (temp_reg6),
00874           [out] "+r" (out)
00875         : [tmp] "r" (tmp), [win] "r" (win), [buf] "r" (buf)
00876         : "hi", "lo"
00877     );
00878 }
00879 
00880 static void ff_imdct36_blocks_mips_fixed(int *out, int *buf, int *in,
00881                                int count, int switch_point, int block_type)
00882 {
00883     int j;
00884     for (j=0 ; j < count; j++) {
00885         /* apply window & overlap with previous buffer */
00886 
00887         /* select window */
00888         int win_idx = (switch_point && j < 2) ? 0 : block_type;
00889         int *win = ff_mdct_win_fixed[win_idx + (4 & -(j & 1))];
00890 
00891         imdct36_mips_fixed(out, buf, in, win);
00892 
00893         in  += 18;
00894         buf += ((j&3) != 3 ? 1 : (72-3));
00895         out++;
00896     }
00897 }
00898 
00899 void ff_mpadsp_init_mipsdspr1(MPADSPContext *s)
00900 {
00901     s->apply_window_fixed   = ff_mpadsp_apply_window_mips_fixed;
00902     s->imdct36_blocks_fixed = ff_imdct36_blocks_mips_fixed;
00903 }
libavcodec/mips/mpegaudiodsp_mips_fixed.c
