FFmpeg: libavcodec/mips/compute_antialias

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  * Compute antialias function optimised 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 #ifndef AVCODEC_MIPS_COMPUTE_ANTIALIAS_FIXED_H
00056 #define AVCODEC_MIPS_COMPUTE_ANTIALIAS_FIXED_H
00057 
00058 #if HAVE_INLINE_ASM
00059 static void compute_antialias_mips_fixed(MPADecodeContext *s,
00060                                         GranuleDef *g)
00061 {
00062     int32_t *ptr, *csa;
00063     int n, i;
00064     int MAX_lo = 0xffffffff;
00065 
00066     /* we antialias only "long" bands */
00067     if (g->block_type == 2) {
00068         if (!g->switch_point)
00069             return;
00070         /* XXX: check this for 8000Hz case */
00071         n = 1;
00072     } else {
00073         n = SBLIMIT - 1;
00074     }
00075 
00076 
00077     ptr = g->sb_hybrid + 18;
00078 
00079     for(i = n;i > 0;i--) {
00080         int tmp0, tmp1, tmp2, tmp00, tmp11;
00081         int temp_reg1, temp_reg2, temp_reg3, temp_reg4, temp_reg5, temp_reg6;
00082         csa = &csa_table[0][0];
00083 
00087         __asm__ volatile (
00088             "lw   %[tmp0],      -1*4(%[ptr])                            \n\t"
00089             "lw   %[tmp1],      0*4(%[ptr])                             \n\t"
00090             "lw   %[temp_reg1], 0*4(%[csa])                             \n\t"
00091             "lw   %[temp_reg2], 2*4(%[csa])                             \n\t"
00092             "add  %[tmp2],      %[tmp0],      %[tmp1]                   \n\t"
00093             "lw   %[temp_reg3], 3*4(%[csa])                             \n\t"
00094             "mult $ac0,         %[tmp2],      %[temp_reg1]              \n\t"
00095             "mult $ac1,         %[tmp2],      %[temp_reg1]              \n\t"
00096             "lw   %[tmp00],     -2*4(%[ptr])                            \n\t"
00097             "lw   %[tmp11],     1*4(%[ptr])                             \n\t"
00098             "lw   %[temp_reg4], 4*4(%[csa])                             \n\t"
00099             "mtlo %[MAX_lo],    $ac0                                    \n\t"
00100             "mtlo $zero,        $ac1                                    \n\t"
00101             "msub $ac0,         %[tmp1],      %[temp_reg2]              \n\t"
00102             "madd $ac1,         %[tmp0],      %[temp_reg3]              \n\t"
00103             "add  %[tmp2],      %[tmp00],     %[tmp11]                  \n\t"
00104             "lw   %[temp_reg5], 6*4(%[csa])                             \n\t"
00105             "mult $ac2,         %[tmp2],      %[temp_reg4]              \n\t"
00106             "mult $ac3,         %[tmp2],      %[temp_reg4]              \n\t"
00107             "mfhi %[temp_reg1], $ac0                                    \n\t"
00108             "mfhi %[temp_reg2], $ac1                                    \n\t"
00109             "lw   %[temp_reg6], 7*4(%[csa])                             \n\t"
00110             "mtlo %[MAX_lo],    $ac2                                    \n\t"
00111             "msub $ac2,         %[tmp11],     %[temp_reg5]              \n\t"
00112             "mtlo $zero,        $ac3                                    \n\t"
00113             "madd $ac3,         %[tmp00],     %[temp_reg6]              \n\t"
00114             "sll  %[temp_reg1], %[temp_reg1], 2                         \n\t"
00115             "sw   %[temp_reg1], -1*4(%[ptr])                            \n\t"
00116             "mfhi %[temp_reg4], $ac2                                    \n\t"
00117             "sll  %[temp_reg2], %[temp_reg2], 2                         \n\t"
00118             "mfhi %[temp_reg5], $ac3                                    \n\t"
00119             "sw   %[temp_reg2], 0*4(%[ptr])                             \n\t"
00120             "lw   %[tmp0],      -3*4(%[ptr])                            \n\t"
00121             "lw   %[tmp1],      2*4(%[ptr])                             \n\t"
00122             "lw   %[temp_reg1], 8*4(%[csa])                             \n\t"
00123             "sll  %[temp_reg4], %[temp_reg4], 2                         \n\t"
00124             "add  %[tmp2],      %[tmp0],      %[tmp1]                   \n\t"
00125             "sll  %[temp_reg5], %[temp_reg5], 2                         \n\t"
00126             "mult $ac0,         %[tmp2],      %[temp_reg1]              \n\t"
00127             "mult $ac1,         %[tmp2],      %[temp_reg1]              \n\t"
00128             "sw   %[temp_reg4], -2*4(%[ptr])                            \n\t"
00129             "sw   %[temp_reg5], 1*4(%[ptr])                             \n\t"
00130             "lw   %[temp_reg2], 10*4(%[csa])                            \n\t"
00131             "mtlo %[MAX_lo],    $ac0                                    \n\t"
00132             "lw   %[temp_reg3], 11*4(%[csa])                            \n\t"
00133             "msub $ac0,         %[tmp1],      %[temp_reg2]              \n\t"
00134             "mtlo $zero,        $ac1                                    \n\t"
00135             "madd $ac1,         %[tmp0],      %[temp_reg3]              \n\t"
00136             "lw   %[tmp00],     -4*4(%[ptr])                            \n\t"
00137             "lw   %[tmp11],     3*4(%[ptr])                             \n\t"
00138             "mfhi %[temp_reg1], $ac0                                    \n\t"
00139             "lw   %[temp_reg4], 12*4(%[csa])                            \n\t"
00140             "mfhi %[temp_reg2], $ac1                                    \n\t"
00141             "add  %[tmp2],      %[tmp00],     %[tmp11]                  \n\t"
00142             "mult $ac2,         %[tmp2],      %[temp_reg4]              \n\t"
00143             "mult $ac3,         %[tmp2],      %[temp_reg4]              \n\t"
00144             "lw   %[temp_reg5], 14*4(%[csa])                            \n\t"
00145             "lw   %[temp_reg6], 15*4(%[csa])                            \n\t"
00146             "sll  %[temp_reg1], %[temp_reg1], 2                         \n\t"
00147             "mtlo %[MAX_lo],    $ac2                                    \n\t"
00148             "msub $ac2,         %[tmp11],     %[temp_reg5]              \n\t"
00149             "mtlo $zero,        $ac3                                    \n\t"
00150             "madd $ac3,         %[tmp00],     %[temp_reg6]              \n\t"
00151             "sll  %[temp_reg2], %[temp_reg2], 2                         \n\t"
00152             "sw   %[temp_reg1], -3*4(%[ptr])                            \n\t"
00153             "mfhi %[temp_reg4], $ac2                                    \n\t"
00154             "sw   %[temp_reg2], 2*4(%[ptr])                             \n\t"
00155             "mfhi %[temp_reg5], $ac3                                    \n\t"
00156             "lw   %[tmp0],      -5*4(%[ptr])                            \n\t"
00157             "lw   %[tmp1],      4*4(%[ptr])                             \n\t"
00158             "lw   %[temp_reg1], 16*4(%[csa])                            \n\t"
00159             "lw   %[temp_reg2], 18*4(%[csa])                            \n\t"
00160             "add  %[tmp2],      %[tmp0],      %[tmp1]                   \n\t"
00161             "lw   %[temp_reg3], 19*4(%[csa])                            \n\t"
00162             "mult $ac0,         %[tmp2],      %[temp_reg1]              \n\t"
00163             "mult $ac1,         %[tmp2],      %[temp_reg1]              \n\t"
00164             "sll  %[temp_reg4], %[temp_reg4], 2                         \n\t"
00165             "sll  %[temp_reg5], %[temp_reg5], 2                         \n\t"
00166             "sw   %[temp_reg4], -4*4(%[ptr])                            \n\t"
00167             "mtlo %[MAX_lo],    $ac0                                    \n\t"
00168             "msub $ac0,         %[tmp1],      %[temp_reg2]              \n\t"
00169             "mtlo $zero,        $ac1                                    \n\t"
00170             "madd $ac1,         %[tmp0],      %[temp_reg3]              \n\t"
00171             "sw   %[temp_reg5], 3*4(%[ptr])                             \n\t"
00172             "lw   %[tmp00],     -6*4(%[ptr])                            \n\t"
00173             "mfhi %[temp_reg1], $ac0                                    \n\t"
00174             "lw   %[tmp11],     5*4(%[ptr])                             \n\t"
00175             "mfhi %[temp_reg2], $ac1                                    \n\t"
00176             "lw   %[temp_reg4], 20*4(%[csa])                            \n\t"
00177             "add  %[tmp2],      %[tmp00],     %[tmp11]                  \n\t"
00178             "lw   %[temp_reg5], 22*4(%[csa])                            \n\t"
00179             "mult $ac2,         %[tmp2],      %[temp_reg4]              \n\t"
00180             "mult $ac3,         %[tmp2],      %[temp_reg4]              \n\t"
00181             "lw   %[temp_reg6], 23*4(%[csa])                            \n\t"
00182             "sll  %[temp_reg1], %[temp_reg1], 2                         \n\t"
00183             "sll  %[temp_reg2], %[temp_reg2], 2                         \n\t"
00184             "mtlo %[MAX_lo],    $ac2                                    \n\t"
00185             "msub $ac2,         %[tmp11],     %[temp_reg5]              \n\t"
00186             "mtlo $zero,        $ac3                                    \n\t"
00187             "madd $ac3,         %[tmp00],     %[temp_reg6]              \n\t"
00188             "sw   %[temp_reg1], -5*4(%[ptr])                            \n\t"
00189             "sw   %[temp_reg2], 4*4(%[ptr])                             \n\t"
00190             "mfhi %[temp_reg4], $ac2                                    \n\t"
00191             "lw   %[tmp0],      -7*4(%[ptr])                            \n\t"
00192             "mfhi %[temp_reg5], $ac3                                    \n\t"
00193             "lw   %[tmp1],      6*4(%[ptr])                             \n\t"
00194             "lw   %[temp_reg1], 24*4(%[csa])                            \n\t"
00195             "lw   %[temp_reg2], 26*4(%[csa])                            \n\t"
00196             "add  %[tmp2],      %[tmp0],      %[tmp1]                   \n\t"
00197             "lw   %[temp_reg3], 27*4(%[csa])                            \n\t"
00198             "mult $ac0,         %[tmp2],      %[temp_reg1]              \n\t"
00199             "mult $ac1,         %[tmp2],      %[temp_reg1]              \n\t"
00200             "sll  %[temp_reg4], %[temp_reg4], 2                         \n\t"
00201             "sll  %[temp_reg5], %[temp_reg5], 2                         \n\t"
00202             "sw   %[temp_reg4], -6*4(%[ptr])                            \n\t"
00203             "mtlo %[MAX_lo],    $ac0                                    \n\t"
00204             "msub $ac0,         %[tmp1],      %[temp_reg2]              \n\t"
00205             "mtlo $zero,        $ac1                                    \n\t"
00206             "madd $ac1,         %[tmp0],      %[temp_reg3]              \n\t"
00207             "sw   %[temp_reg5], 5*4(%[ptr])                             \n\t"
00208             "lw   %[tmp00],     -8*4(%[ptr])                            \n\t"
00209             "mfhi %[temp_reg1], $ac0                                    \n\t"
00210             "lw   %[tmp11],     7*4(%[ptr])                             \n\t"
00211             "mfhi %[temp_reg2], $ac1                                    \n\t"
00212             "lw   %[temp_reg4], 28*4(%[csa])                            \n\t"
00213             "add  %[tmp2],      %[tmp00],     %[tmp11]                  \n\t"
00214             "lw   %[temp_reg5], 30*4(%[csa])                            \n\t"
00215             "mult $ac2,         %[tmp2],      %[temp_reg4]              \n\t"
00216             "mult $ac3,         %[tmp2],      %[temp_reg4]              \n\t"
00217             "lw   %[temp_reg6], 31*4(%[csa])                            \n\t"
00218             "sll  %[temp_reg1], %[temp_reg1], 2                         \n\t"
00219             "sll  %[temp_reg2], %[temp_reg2], 2                         \n\t"
00220             "mtlo %[MAX_lo],    $ac2                                    \n\t"
00221             "msub $ac2,         %[tmp11],     %[temp_reg5]              \n\t"
00222             "mtlo $zero,        $ac3                                    \n\t"
00223             "madd $ac3,         %[tmp00],     %[temp_reg6]              \n\t"
00224             "sw   %[temp_reg1], -7*4(%[ptr])                            \n\t"
00225             "sw   %[temp_reg2], 6*4(%[ptr])                             \n\t"
00226             "mfhi %[temp_reg4], $ac2                                    \n\t"
00227             "mfhi %[temp_reg5], $ac3                                    \n\t"
00228             "sll  %[temp_reg4], %[temp_reg4], 2                         \n\t"
00229             "sll  %[temp_reg5], %[temp_reg5], 2                         \n\t"
00230             "sw   %[temp_reg4], -8*4(%[ptr])                            \n\t"
00231             "sw   %[temp_reg5], 7*4(%[ptr])                             \n\t"
00232 
00233             : [tmp0] "=&r" (tmp0), [tmp1] "=&r" (tmp1), [tmp2] "=&r" (tmp2),
00234               [tmp00] "=&r" (tmp00), [tmp11] "=&r" (tmp11),
00235               [temp_reg1] "=&r" (temp_reg1), [temp_reg2] "=&r" (temp_reg2),
00236               [temp_reg3] "=&r" (temp_reg3), [temp_reg4] "=&r" (temp_reg4),
00237               [temp_reg5] "=&r" (temp_reg5), [temp_reg6] "=&r" (temp_reg6)
00238             : [csa] "r" (csa), [ptr] "r" (ptr),
00239               [MAX_lo] "r" (MAX_lo)
00240          );
00241 
00242         ptr += 18;
00243     }
00244 }
00245 #define compute_antialias compute_antialias_mips_fixed
00246 #endif /* HAVE_INLINE_ASM */
00247 
00248 #endif /* AVCODEC_MIPS_COMPUTE_ANTIALIAS_FIXED_H */
libavcodec/mips/compute_antialias_fixed.h
