Mercurial > hg > pmhd

annotate ffmpeg/libavcodec/mips/amrwbdec_mips.c @ 13:844d341cf643 tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Back up before ISMIR
author Yading Song <yading.song@eecs.qmul.ac.uk>
date Thu, 31 Oct 2013 13:17:06 +0000
parents 6840f77b83aa
children
rev   line source
yading@10 1 /*
yading@10 2 * Copyright (c) 2012
yading@10 3 * MIPS Technologies, Inc., California.
yading@10 4 *
yading@10 5 * Redistribution and use in source and binary forms, with or without
yading@10 6 * modification, are permitted provided that the following conditions
yading@10 7 * are met:
yading@10 8 * 1. Redistributions of source code must retain the above copyright
yading@10 9 * notice, this list of conditions and the following disclaimer.
yading@10 10 * 2. Redistributions in binary form must reproduce the above copyright
yading@10 11 * notice, this list of conditions and the following disclaimer in the
yading@10 12 * documentation and/or other materials provided with the distribution.
yading@10 13 * 3. Neither the name of the MIPS Technologies, Inc., nor the names of its
yading@10 14 * contributors may be used to endorse or promote products derived from
yading@10 15 * this software without specific prior written permission.
yading@10 16 *
yading@10 17 * THIS SOFTWARE IS PROVIDED BY THE MIPS TECHNOLOGIES, INC. ``AS IS'' AND
yading@10 18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
yading@10 19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
yading@10 20 * ARE DISCLAIMED. IN NO EVENT SHALL THE MIPS TECHNOLOGIES, INC. BE LIABLE
yading@10 21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
yading@10 22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
yading@10 23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
yading@10 24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
yading@10 25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
yading@10 26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
yading@10 27 * SUCH DAMAGE.
yading@10 28 *
yading@10 29 * Author: Nedeljko Babic (nbabic@mips.com)
yading@10 30 *
yading@10 31 * This file is part of FFmpeg.
yading@10 32 *
yading@10 33 * FFmpeg is free software; you can redistribute it and/or
yading@10 34 * modify it under the terms of the GNU Lesser General Public
yading@10 35 * License as published by the Free Software Foundation; either
yading@10 36 * version 2.1 of the License, or (at your option) any later version.
yading@10 37 *
yading@10 38 * FFmpeg is distributed in the hope that it will be useful,
yading@10 39 * but WITHOUT ANY WARRANTY; without even the implied warranty of
yading@10 40 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
yading@10 41 * Lesser General Public License for more details.
yading@10 42 *
yading@10 43 * You should have received a copy of the GNU Lesser General Public
yading@10 44 * License along with FFmpeg; if not, write to the Free Software
yading@10 45 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
yading@10 46 */
yading@10 47
yading@10 48 /**
yading@10 49 * @file
yading@10 50 * Reference: libavcodec/amrwbdec.c
yading@10 51 */
yading@10 52 #include "libavutil/avutil.h"
yading@10 53 #include "libavcodec/amrwbdata.h"
yading@10 54 #include "amrwbdec_mips.h"
yading@10 55
yading@10 56 #if HAVE_INLINE_ASM
yading@10 57 void hb_fir_filter_mips(float *out, const float fir_coef[HB_FIR_SIZE + 1],
yading@10 58 float mem[HB_FIR_SIZE], const float *in)
yading@10 59 {
yading@10 60 int i;
yading@10 61 float data[AMRWB_SFR_SIZE_16k + HB_FIR_SIZE]; // past and current samples
yading@10 62
yading@10 63 memcpy(data, mem, HB_FIR_SIZE * sizeof(float));
yading@10 64 memcpy(data + HB_FIR_SIZE, in, AMRWB_SFR_SIZE_16k * sizeof(float));
yading@10 65
yading@10 66 for (i = 0; i < AMRWB_SFR_SIZE_16k; i++) {
yading@10 67 float output;
yading@10 68 float * p_data = (data+i);
yading@10 69
yading@10 70 /**
yading@10 71 * inner loop is entirely unrolled and instructions are scheduled
yading@10 72 * to minimize pipeline stall
yading@10 73 */
yading@10 74 __asm__ volatile(
yading@10 75 "mtc1 $zero, %[output] \n\t"
yading@10 76 "lwc1 $f0, 0(%[p_data]) \n\t"
yading@10 77 "lwc1 $f1, 0(%[fir_coef]) \n\t"
yading@10 78 "lwc1 $f2, 4(%[p_data]) \n\t"
yading@10 79 "madd.s %[output], %[output], $f0, $f1 \n\t"
yading@10 80 "lwc1 $f3, 4(%[fir_coef]) \n\t"
yading@10 81 "lwc1 $f4, 8(%[p_data]) \n\t"
yading@10 82 "madd.s %[output], %[output], $f2, $f3 \n\t"
yading@10 83 "lwc1 $f5, 8(%[fir_coef]) \n\t"
yading@10 84
yading@10 85 "lwc1 $f0, 12(%[p_data]) \n\t"
yading@10 86 "lwc1 $f1, 12(%[fir_coef]) \n\t"
yading@10 87 "madd.s %[output], %[output], $f4, $f5 \n\t"
yading@10 88 "lwc1 $f2, 16(%[p_data]) \n\t"
yading@10 89 "madd.s %[output], %[output], $f0, $f1 \n\t"
yading@10 90 "lwc1 $f3, 16(%[fir_coef]) \n\t"
yading@10 91 "lwc1 $f4, 20(%[p_data]) \n\t"
yading@10 92 "lwc1 $f5, 20(%[fir_coef]) \n\t"
yading@10 93 "madd.s %[output], %[output], $f2, $f3 \n\t"
yading@10 94
yading@10 95 "lwc1 $f0, 24(%[p_data]) \n\t"
yading@10 96 "lwc1 $f1, 24(%[fir_coef]) \n\t"
yading@10 97 "lwc1 $f2, 28(%[p_data]) \n\t"
yading@10 98 "madd.s %[output], %[output], $f4, $f5 \n\t"
yading@10 99 "lwc1 $f3, 28(%[fir_coef]) \n\t"
yading@10 100 "madd.s %[output], %[output], $f0, $f1 \n\t"
yading@10 101 "lwc1 $f4, 32(%[p_data]) \n\t"
yading@10 102 "madd.s %[output], %[output], $f2, $f3 \n\t"
yading@10 103 "lwc1 $f5, 32(%[fir_coef]) \n\t"
yading@10 104 "madd.s %[output], %[output], $f4, $f5 \n\t"
yading@10 105
yading@10 106 "lwc1 $f0, 36(%[p_data]) \n\t"
yading@10 107 "lwc1 $f1, 36(%[fir_coef]) \n\t"
yading@10 108 "lwc1 $f2, 40(%[p_data]) \n\t"
yading@10 109 "lwc1 $f3, 40(%[fir_coef]) \n\t"
yading@10 110 "madd.s %[output], %[output], $f0, $f1 \n\t"
yading@10 111 "lwc1 $f4, 44(%[p_data]) \n\t"
yading@10 112 "lwc1 $f5, 44(%[fir_coef]) \n\t"
yading@10 113 "madd.s %[output], %[output], $f2, $f3 \n\t"
yading@10 114
yading@10 115 "lwc1 $f0, 48(%[p_data]) \n\t"
yading@10 116 "lwc1 $f1, 48(%[fir_coef]) \n\t"
yading@10 117 "lwc1 $f2, 52(%[p_data]) \n\t"
yading@10 118 "madd.s %[output], %[output], $f4, $f5 \n\t"
yading@10 119 "lwc1 $f3, 52(%[fir_coef]) \n\t"
yading@10 120 "lwc1 $f4, 56(%[p_data]) \n\t"
yading@10 121 "madd.s %[output], %[output], $f0, $f1 \n\t"
yading@10 122 "lwc1 $f5, 56(%[fir_coef]) \n\t"
yading@10 123 "madd.s %[output], %[output], $f2, $f3 \n\t"
yading@10 124
yading@10 125 "lwc1 $f0, 60(%[p_data]) \n\t"
yading@10 126 "lwc1 $f1, 60(%[fir_coef]) \n\t"
yading@10 127 "lwc1 $f2, 64(%[p_data]) \n\t"
yading@10 128 "madd.s %[output], %[output], $f4, $f5 \n\t"
yading@10 129 "lwc1 $f3, 64(%[fir_coef]) \n\t"
yading@10 130 "madd.s %[output], %[output], $f0, $f1 \n\t"
yading@10 131 "lwc1 $f4, 68(%[p_data]) \n\t"
yading@10 132 "madd.s %[output], %[output], $f2, $f3 \n\t"
yading@10 133 "lwc1 $f5, 68(%[fir_coef]) \n\t"
yading@10 134 "madd.s %[output], %[output], $f4, $f5 \n\t"
yading@10 135
yading@10 136 "lwc1 $f0, 72(%[p_data]) \n\t"
yading@10 137 "lwc1 $f1, 72(%[fir_coef]) \n\t"
yading@10 138 "lwc1 $f2, 76(%[p_data]) \n\t"
yading@10 139 "lwc1 $f3, 76(%[fir_coef]) \n\t"
yading@10 140 "madd.s %[output], %[output], $f0, $f1 \n\t"
yading@10 141 "lwc1 $f4, 80(%[p_data]) \n\t"
yading@10 142 "lwc1 $f5, 80(%[fir_coef]) \n\t"
yading@10 143 "madd.s %[output], %[output], $f2, $f3 \n\t"
yading@10 144
yading@10 145 "lwc1 $f0, 84(%[p_data]) \n\t"
yading@10 146 "lwc1 $f1, 84(%[fir_coef]) \n\t"
yading@10 147 "lwc1 $f2, 88(%[p_data]) \n\t"
yading@10 148 "madd.s %[output], %[output], $f4, $f5 \n\t"
yading@10 149 "lwc1 $f3, 88(%[fir_coef]) \n\t"
yading@10 150 "lwc1 $f4, 92(%[p_data]) \n\t"
yading@10 151 "madd.s %[output], %[output], $f0, $f1 \n\t"
yading@10 152 "lwc1 $f5, 92(%[fir_coef]) \n\t"
yading@10 153 "madd.s %[output], %[output], $f2, $f3 \n\t"
yading@10 154
yading@10 155 "lwc1 $f0, 96(%[p_data]) \n\t"
yading@10 156 "lwc1 $f1, 96(%[fir_coef]) \n\t"
yading@10 157 "lwc1 $f2, 100(%[p_data]) \n\t"
yading@10 158 "madd.s %[output], %[output], $f4, $f5 \n\t"
yading@10 159 "lwc1 $f3, 100(%[fir_coef]) \n\t"
yading@10 160 "lwc1 $f4, 104(%[p_data]) \n\t"
yading@10 161 "madd.s %[output], %[output], $f0, $f1 \n\t"
yading@10 162 "lwc1 $f5, 104(%[fir_coef]) \n\t"
yading@10 163 "madd.s %[output], %[output], $f2, $f3 \n\t"
yading@10 164
yading@10 165 "lwc1 $f0, 108(%[p_data]) \n\t"
yading@10 166 "lwc1 $f1, 108(%[fir_coef]) \n\t"
yading@10 167 "madd.s %[output], %[output], $f4, $f5 \n\t"
yading@10 168 "lwc1 $f2, 112(%[p_data]) \n\t"
yading@10 169 "lwc1 $f3, 112(%[fir_coef]) \n\t"
yading@10 170 "madd.s %[output], %[output], $f0, $f1 \n\t"
yading@10 171 "lwc1 $f4, 116(%[p_data]) \n\t"
yading@10 172 "lwc1 $f5, 116(%[fir_coef]) \n\t"
yading@10 173 "lwc1 $f0, 120(%[p_data]) \n\t"
yading@10 174 "madd.s %[output], %[output], $f2, $f3 \n\t"
yading@10 175 "lwc1 $f1, 120(%[fir_coef]) \n\t"
yading@10 176 "madd.s %[output], %[output], $f4, $f5 \n\t"
yading@10 177 "madd.s %[output], %[output], $f0, $f1 \n\t"
yading@10 178
yading@10 179 : [output]"=&f"(output)
yading@10 180 : [fir_coef]"r"(fir_coef), [p_data]"r"(p_data)
yading@10 181 : "$f0", "$f1", "$f2", "$f3", "$f4", "$f5", "memory"
yading@10 182 );
yading@10 183 out[i] = output;
yading@10 184 }
yading@10 185 memcpy(mem, data + AMRWB_SFR_SIZE_16k, HB_FIR_SIZE * sizeof(float));
yading@10 186 }
yading@10 187 #endif /* HAVE_INLINE_ASM */