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 floating-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_FLOAT_H
00056 #define AVCODEC_MIPS_COMPUTE_ANTIALIAS_FLOAT_H
00057 
00058 #if HAVE_INLINE_ASM
00059 static void compute_antialias_mips_float(MPADecodeContext *s,
00060                                         GranuleDef *g)
00061 {
00062     float *ptr, *ptr_end;
00063     float *csa = &csa_table[0][0];
00064     int n;
00065     /* temporary variables */
00066     float in1, in2, in3, in4, in5, in6, in7, in8;
00067     float out1, out2, out3, out4;
00068 
00069     ptr = g->sb_hybrid + 18;
00070     /* we antialias only "long" bands */
00071     if (g->block_type == 2) {
00072         if (!g->switch_point)
00073             return;
00074         /* XXX: check this for 8000Hz case */
00075         n = 1;
00076         ptr_end = ptr + 18;
00077     } else {
00078         n = 31;
00079         ptr_end = ptr + 558;
00080     }
00081 
00086     __asm__ volatile (
00087         "compute_antialias_float_loop%=:                                \t\n"
00088         "lwc1    %[in1],  -1*4(%[ptr])                                  \t\n"
00089         "lwc1    %[in2],  0(%[csa])                                     \t\n"
00090         "lwc1    %[in3],  1*4(%[csa])                                   \t\n"
00091         "lwc1    %[in4],  0(%[ptr])                                     \t\n"
00092         "lwc1    %[in5],  -2*4(%[ptr])                                  \t\n"
00093         "lwc1    %[in6],  4*4(%[csa])                                   \t\n"
00094         "mul.s   %[out1], %[in1],  %[in2]                               \t\n"
00095         "mul.s   %[out2], %[in1],  %[in3]                               \t\n"
00096         "lwc1    %[in7],  5*4(%[csa])                                   \t\n"
00097         "lwc1    %[in8],  1*4(%[ptr])                                   \t\n"
00098         "nmsub.s %[out1], %[out1], %[in3], %[in4]                       \t\n"
00099         "madd.s  %[out2], %[out2], %[in2], %[in4]                       \t\n"
00100         "mul.s   %[out3], %[in5],  %[in6]                               \t\n"
00101         "mul.s   %[out4], %[in5],  %[in7]                               \t\n"
00102         "lwc1    %[in1],  -3*4(%[ptr])                                  \t\n"
00103         "swc1    %[out1], -1*4(%[ptr])                                  \t\n"
00104         "swc1    %[out2], 0(%[ptr])                                     \t\n"
00105         "nmsub.s %[out3], %[out3], %[in7], %[in8]                       \t\n"
00106         "madd.s  %[out4], %[out4], %[in6], %[in8]                       \t\n"
00107         "lwc1    %[in2],  8*4(%[csa])                                   \t\n"
00108         "swc1    %[out3], -2*4(%[ptr])                                  \t\n"
00109         "swc1    %[out4], 1*4(%[ptr])                                   \t\n"
00110         "lwc1    %[in3],  9*4(%[csa])                                   \t\n"
00111         "lwc1    %[in4],  2*4(%[ptr])                                   \t\n"
00112         "mul.s   %[out1], %[in1],  %[in2]                               \t\n"
00113         "lwc1    %[in5],  -4*4(%[ptr])                                  \t\n"
00114         "lwc1    %[in6],  12*4(%[csa])                                  \t\n"
00115         "mul.s   %[out2], %[in1],  %[in3]                               \t\n"
00116         "lwc1    %[in7],  13*4(%[csa])                                  \t\n"
00117         "nmsub.s %[out1], %[out1], %[in3], %[in4]                       \t\n"
00118         "lwc1    %[in8],  3*4(%[ptr])                                   \t\n"
00119         "mul.s   %[out3], %[in5],  %[in6]                               \t\n"
00120         "madd.s  %[out2], %[out2], %[in2], %[in4]                       \t\n"
00121         "mul.s   %[out4], %[in5],  %[in7]                               \t\n"
00122         "swc1    %[out1], -3*4(%[ptr])                                  \t\n"
00123         "lwc1    %[in1],  -5*4(%[ptr])                                  \t\n"
00124         "nmsub.s %[out3], %[out3], %[in7], %[in8]                       \t\n"
00125         "swc1    %[out2], 2*4(%[ptr])                                   \t\n"
00126         "madd.s  %[out4], %[out4], %[in6], %[in8]                       \t\n"
00127         "lwc1    %[in2],  16*4(%[csa])                                  \t\n"
00128         "lwc1    %[in3],  17*4(%[csa])                                  \t\n"
00129         "swc1    %[out3], -4*4(%[ptr])                                  \t\n"
00130         "lwc1    %[in4],  4*4(%[ptr])                                   \t\n"
00131         "swc1    %[out4], 3*4(%[ptr])                                   \t\n"
00132         "mul.s   %[out1], %[in1],  %[in2]                               \t\n"
00133         "mul.s   %[out2], %[in1],  %[in3]                               \t\n"
00134         "lwc1    %[in5],  -6*4(%[ptr])                                  \t\n"
00135         "lwc1    %[in6],  20*4(%[csa])                                  \t\n"
00136         "lwc1    %[in7],  21*4(%[csa])                                  \t\n"
00137         "nmsub.s %[out1], %[out1], %[in3], %[in4]                       \t\n"
00138         "madd.s  %[out2], %[out2], %[in2], %[in4]                       \t\n"
00139         "lwc1    %[in8],  5*4(%[ptr])                                   \t\n"
00140         "mul.s   %[out3], %[in5],  %[in6]                               \t\n"
00141         "mul.s   %[out4], %[in5],  %[in7]                               \t\n"
00142         "swc1    %[out1], -5*4(%[ptr])                                  \t\n"
00143         "swc1    %[out2], 4*4(%[ptr])                                   \t\n"
00144         "lwc1    %[in1],  -7*4(%[ptr])                                  \t\n"
00145         "nmsub.s %[out3], %[out3], %[in7], %[in8]                       \t\n"
00146         "madd.s  %[out4], %[out4], %[in6], %[in8]                       \t\n"
00147         "lwc1    %[in2],  24*4(%[csa])                                  \t\n"
00148         "lwc1    %[in3],  25*4(%[csa])                                  \t\n"
00149         "lwc1    %[in4],  6*4(%[ptr])                                   \t\n"
00150         "swc1    %[out3], -6*4(%[ptr])                                  \t\n"
00151         "swc1    %[out4], 5*4(%[ptr])                                   \t\n"
00152         "mul.s   %[out1], %[in1],  %[in2]                               \t\n"
00153         "lwc1    %[in5],  -8*4(%[ptr])                                  \t\n"
00154         "mul.s   %[out2], %[in1],  %[in3]                               \t\n"
00155         "lwc1    %[in6],  28*4(%[csa])                                  \t\n"
00156         "lwc1    %[in7],  29*4(%[csa])                                  \t\n"
00157         "nmsub.s %[out1], %[out1], %[in3], %[in4]                       \t\n"
00158         "lwc1    %[in8],  7*4(%[ptr])                                   \t\n"
00159         "madd.s  %[out2], %[out2], %[in2], %[in4]                       \t\n"
00160         "mul.s   %[out3], %[in5],  %[in6]                               \t\n"
00161         "mul.s   %[out4], %[in5],  %[in7]                               \t\n"
00162         "swc1    %[out1], -7*4(%[ptr])                                  \t\n"
00163         "swc1    %[out2], 6*4(%[ptr])                                   \t\n"
00164         "addiu   %[ptr],  %[ptr],  72                                   \t\n"
00165         "nmsub.s %[out3], %[out3], %[in7], %[in8]                       \t\n"
00166         "madd.s  %[out4], %[out4], %[in6], %[in8]                       \t\n"
00167         "swc1    %[out3], -26*4(%[ptr])                                 \t\n"
00168         "swc1    %[out4], -11*4(%[ptr])                                 \t\n"
00169         "bne     %[ptr],  %[ptr_end],  compute_antialias_float_loop%=   \t\n"
00170 
00171         : [ptr] "+r" (ptr),
00172           [in1] "=&f" (in1), [in2] "=&f" (in2),
00173           [in3] "=&f" (in3), [in4] "=&f" (in4),
00174           [in5] "=&f" (in5), [in6] "=&f" (in6),
00175           [in7] "=&f" (in7), [in8] "=&f" (in8),
00176           [out1] "=&f" (out1), [out2] "=&f" (out2),
00177           [out3] "=&f" (out3), [out4] "=&f" (out4)
00178         : [csa] "r" (csa), [ptr_end] "r" (ptr_end)
00179     );
00180 }
00181 #define compute_antialias compute_antialias_mips_float
00182 #endif /* HAVE_INLINE_ASM */
00183 
00184 #endif /* AVCODEC_MIPS_COMPUTE_ANTIALIAS_FLOAT_H */
libavcodec/mips/compute_antialias_float.h
