Mercurial > hg > pmhd

annotate ffmpeg/libavcodec/mips/ac3dsp_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 * Authors: Branimir Vasic (bvasic@mips.com)
yading@10 30 * Nedeljko Babic (nbabic@mips.com)
yading@10 31 *
yading@10 32 * Various AC-3 DSP Utils optimized for MIPS
yading@10 33 *
yading@10 34 * This file is part of FFmpeg.
yading@10 35 *
yading@10 36 * FFmpeg is free software; you can redistribute it and/or
yading@10 37 * modify it under the terms of the GNU Lesser General Public
yading@10 38 * License as published by the Free Software Foundation; either
yading@10 39 * version 2.1 of the License, or (at your option) any later version.
yading@10 40 *
yading@10 41 * FFmpeg is distributed in the hope that it will be useful,
yading@10 42 * but WITHOUT ANY WARRANTY; without even the implied warranty of
yading@10 43 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
yading@10 44 * Lesser General Public License for more details.
yading@10 45 *
yading@10 46 * You should have received a copy of the GNU Lesser General Public
yading@10 47 * License along with FFmpeg; if not, write to the Free Software
yading@10 48 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
yading@10 49 */
yading@10 50
yading@10 51 /**
yading@10 52 * @file
yading@10 53 * Reference: libavcodec/ac3dsp.c
yading@10 54 */
yading@10 55
yading@10 56 #include "config.h"
yading@10 57 #include "libavcodec/ac3dsp.h"
yading@10 58 #include "libavcodec/ac3.h"
yading@10 59
yading@10 60
yading@10 61 #if HAVE_INLINE_ASM
yading@10 62 #if HAVE_MIPSDSPR1
yading@10 63 static void ac3_bit_alloc_calc_bap_mips(int16_t *mask, int16_t *psd,
yading@10 64 int start, int end,
yading@10 65 int snr_offset, int floor,
yading@10 66 const uint8_t *bap_tab, uint8_t *bap)
yading@10 67 {
yading@10 68 int band, band_end, cond;
yading@10 69 int m, address1, address2;
yading@10 70 int16_t *psd1, *psd_end;
yading@10 71 uint8_t *bap1;
yading@10 72
yading@10 73 if (snr_offset == -960) {
yading@10 74 memset(bap, 0, AC3_MAX_COEFS);
yading@10 75 return;
yading@10 76 }
yading@10 77
yading@10 78 psd1 = &psd[start];
yading@10 79 bap1 = &bap[start];
yading@10 80 band = ff_ac3_bin_to_band_tab[start];
yading@10 81
yading@10 82 do {
yading@10 83 m = (FFMAX(mask[band] - snr_offset - floor, 0) & 0x1FE0) + floor;
yading@10 84 band_end = ff_ac3_band_start_tab[++band];
yading@10 85 band_end = FFMIN(band_end, end);
yading@10 86 psd_end = psd + band_end - 1;
yading@10 87
yading@10 88 __asm__ volatile (
yading@10 89 "slt %[cond], %[psd1], %[psd_end] \n\t"
yading@10 90 "beqz %[cond], 1f \n\t"
yading@10 91 "2: \n\t"
yading@10 92 "lh %[address1], 0(%[psd1]) \n\t"
yading@10 93 "lh %[address2], 2(%[psd1]) \n\t"
yading@10 94 "addiu %[psd1], %[psd1], 4 \n\t"
yading@10 95 "subu %[address1], %[address1], %[m] \n\t"
yading@10 96 "sra %[address1], %[address1], 5 \n\t"
yading@10 97 "addiu %[address1], %[address1], -32 \n\t"
yading@10 98 "shll_s.w %[address1], %[address1], 26 \n\t"
yading@10 99 "subu %[address2], %[address2], %[m] \n\t"
yading@10 100 "sra %[address2], %[address2], 5 \n\t"
yading@10 101 "sra %[address1], %[address1], 26 \n\t"
yading@10 102 "addiu %[address1], %[address1], 32 \n\t"
yading@10 103 "lbux %[address1], %[address1](%[bap_tab]) \n\t"
yading@10 104 "addiu %[address2], %[address2], -32 \n\t"
yading@10 105 "shll_s.w %[address2], %[address2], 26 \n\t"
yading@10 106 "sb %[address1], 0(%[bap1]) \n\t"
yading@10 107 "slt %[cond], %[psd1], %[psd_end] \n\t"
yading@10 108 "sra %[address2], %[address2], 26 \n\t"
yading@10 109 "addiu %[address2], %[address2], 32 \n\t"
yading@10 110 "lbux %[address2], %[address2](%[bap_tab]) \n\t"
yading@10 111 "sb %[address2], 1(%[bap1]) \n\t"
yading@10 112 "addiu %[bap1], %[bap1], 2 \n\t"
yading@10 113 "bnez %[cond], 2b \n\t"
yading@10 114 "addiu %[psd_end], %[psd_end], 2 \n\t"
yading@10 115 "slt %[cond], %[psd1], %[psd_end] \n\t"
yading@10 116 "beqz %[cond], 3f \n\t"
yading@10 117 "1: \n\t"
yading@10 118 "lh %[address1], 0(%[psd1]) \n\t"
yading@10 119 "addiu %[psd1], %[psd1], 2 \n\t"
yading@10 120 "subu %[address1], %[address1], %[m] \n\t"
yading@10 121 "sra %[address1], %[address1], 5 \n\t"
yading@10 122 "addiu %[address1], %[address1], -32 \n\t"
yading@10 123 "shll_s.w %[address1], %[address1], 26 \n\t"
yading@10 124 "sra %[address1], %[address1], 26 \n\t"
yading@10 125 "addiu %[address1], %[address1], 32 \n\t"
yading@10 126 "lbux %[address1], %[address1](%[bap_tab]) \n\t"
yading@10 127 "sb %[address1], 0(%[bap1]) \n\t"
yading@10 128 "addiu %[bap1], %[bap1], 1 \n\t"
yading@10 129 "3: \n\t"
yading@10 130
yading@10 131 : [address1]"=&r"(address1), [address2]"=&r"(address2),
yading@10 132 [cond]"=&r"(cond), [bap1]"+r"(bap1),
yading@10 133 [psd1]"+r"(psd1), [psd_end]"+r"(psd_end)
yading@10 134 : [m]"r"(m), [bap_tab]"r"(bap_tab)
yading@10 135 : "memory"
yading@10 136 );
yading@10 137 } while (end > band_end);
yading@10 138 }
yading@10 139
yading@10 140 static void ac3_update_bap_counts_mips(uint16_t mant_cnt[16], uint8_t *bap,
yading@10 141 int len)
yading@10 142 {
yading@10 143 int temp0, temp1, temp2, temp3, temp4, temp5, temp6, temp7;
yading@10 144
yading@10 145 __asm__ volatile (
yading@10 146 "andi %[temp3], %[len], 3 \n\t"
yading@10 147 "addu %[temp2], %[bap], %[len] \n\t"
yading@10 148 "addu %[temp4], %[bap], %[temp3] \n\t"
yading@10 149 "beq %[temp2], %[temp4], 4f \n\t"
yading@10 150 "1: \n\t"
yading@10 151 "lbu %[temp0], -1(%[temp2]) \n\t"
yading@10 152 "lbu %[temp5], -2(%[temp2]) \n\t"
yading@10 153 "lbu %[temp6], -3(%[temp2]) \n\t"
yading@10 154 "sll %[temp0], %[temp0], 1 \n\t"
yading@10 155 "addu %[temp0], %[mant_cnt], %[temp0] \n\t"
yading@10 156 "sll %[temp5], %[temp5], 1 \n\t"
yading@10 157 "addu %[temp5], %[mant_cnt], %[temp5] \n\t"
yading@10 158 "lhu %[temp1], 0(%[temp0]) \n\t"
yading@10 159 "sll %[temp6], %[temp6], 1 \n\t"
yading@10 160 "addu %[temp6], %[mant_cnt], %[temp6] \n\t"
yading@10 161 "addiu %[temp1], %[temp1], 1 \n\t"
yading@10 162 "sh %[temp1], 0(%[temp0]) \n\t"
yading@10 163 "lhu %[temp1], 0(%[temp5]) \n\t"
yading@10 164 "lbu %[temp7], -4(%[temp2]) \n\t"
yading@10 165 "addiu %[temp2], %[temp2], -4 \n\t"
yading@10 166 "addiu %[temp1], %[temp1], 1 \n\t"
yading@10 167 "sh %[temp1], 0(%[temp5]) \n\t"
yading@10 168 "lhu %[temp1], 0(%[temp6]) \n\t"
yading@10 169 "sll %[temp7], %[temp7], 1 \n\t"
yading@10 170 "addu %[temp7], %[mant_cnt], %[temp7] \n\t"
yading@10 171 "addiu %[temp1], %[temp1],1 \n\t"
yading@10 172 "sh %[temp1], 0(%[temp6]) \n\t"
yading@10 173 "lhu %[temp1], 0(%[temp7]) \n\t"
yading@10 174 "addiu %[temp1], %[temp1], 1 \n\t"
yading@10 175 "sh %[temp1], 0(%[temp7]) \n\t"
yading@10 176 "bne %[temp2], %[temp4], 1b \n\t"
yading@10 177 "4: \n\t"
yading@10 178 "beqz %[temp3], 2f \n\t"
yading@10 179 "3: \n\t"
yading@10 180 "addiu %[temp3], %[temp3], -1 \n\t"
yading@10 181 "lbu %[temp0], -1(%[temp2]) \n\t"
yading@10 182 "addiu %[temp2], %[temp2], -1 \n\t"
yading@10 183 "sll %[temp0], %[temp0], 1 \n\t"
yading@10 184 "addu %[temp0], %[mant_cnt], %[temp0] \n\t"
yading@10 185 "lhu %[temp1], 0(%[temp0]) \n\t"
yading@10 186 "addiu %[temp1], %[temp1], 1 \n\t"
yading@10 187 "sh %[temp1], 0(%[temp0]) \n\t"
yading@10 188 "bgtz %[temp3], 3b \n\t"
yading@10 189 "2: \n\t"
yading@10 190
yading@10 191 : [temp0] "=&r" (temp0), [temp1] "=&r" (temp1),
yading@10 192 [temp2] "=&r" (temp2), [temp3] "=&r" (temp3),
yading@10 193 [temp4] "=&r" (temp4), [temp5] "=&r" (temp5),
yading@10 194 [temp6] "=&r" (temp6), [temp7] "=&r" (temp7)
yading@10 195 : [len] "r" (len), [bap] "r" (bap),
yading@10 196 [mant_cnt] "r" (mant_cnt)
yading@10 197 : "memory"
yading@10 198 );
yading@10 199 }
yading@10 200 #endif
yading@10 201
yading@10 202 #if HAVE_MIPSFPU && HAVE_MIPS32R2
yading@10 203 static void float_to_fixed24_mips(int32_t *dst, const float *src, unsigned int len)
yading@10 204 {
yading@10 205 const float scale = 1 << 24;
yading@10 206 float src0, src1, src2, src3, src4, src5, src6, src7;
yading@10 207 int temp0, temp1, temp2, temp3, temp4, temp5, temp6, temp7;
yading@10 208
yading@10 209 do {
yading@10 210 __asm__ volatile (
yading@10 211 "lwc1 %[src0], 0(%[src]) \n\t"
yading@10 212 "lwc1 %[src1], 4(%[src]) \n\t"
yading@10 213 "lwc1 %[src2], 8(%[src]) \n\t"
yading@10 214 "lwc1 %[src3], 12(%[src]) \n\t"
yading@10 215 "lwc1 %[src4], 16(%[src]) \n\t"
yading@10 216 "lwc1 %[src5], 20(%[src]) \n\t"
yading@10 217 "lwc1 %[src6], 24(%[src]) \n\t"
yading@10 218 "lwc1 %[src7], 28(%[src]) \n\t"
yading@10 219 "mul.s %[src0], %[src0], %[scale] \n\t"
yading@10 220 "mul.s %[src1], %[src1], %[scale] \n\t"
yading@10 221 "mul.s %[src2], %[src2], %[scale] \n\t"
yading@10 222 "mul.s %[src3], %[src3], %[scale] \n\t"
yading@10 223 "mul.s %[src4], %[src4], %[scale] \n\t"
yading@10 224 "mul.s %[src5], %[src5], %[scale] \n\t"
yading@10 225 "mul.s %[src6], %[src6], %[scale] \n\t"
yading@10 226 "mul.s %[src7], %[src7], %[scale] \n\t"
yading@10 227 "cvt.w.s %[src0], %[src0] \n\t"
yading@10 228 "cvt.w.s %[src1], %[src1] \n\t"
yading@10 229 "cvt.w.s %[src2], %[src2] \n\t"
yading@10 230 "cvt.w.s %[src3], %[src3] \n\t"
yading@10 231 "cvt.w.s %[src4], %[src4] \n\t"
yading@10 232 "cvt.w.s %[src5], %[src5] \n\t"
yading@10 233 "cvt.w.s %[src6], %[src6] \n\t"
yading@10 234 "cvt.w.s %[src7], %[src7] \n\t"
yading@10 235 "mfc1 %[temp0], %[src0] \n\t"
yading@10 236 "mfc1 %[temp1], %[src1] \n\t"
yading@10 237 "mfc1 %[temp2], %[src2] \n\t"
yading@10 238 "mfc1 %[temp3], %[src3] \n\t"
yading@10 239 "mfc1 %[temp4], %[src4] \n\t"
yading@10 240 "mfc1 %[temp5], %[src5] \n\t"
yading@10 241 "mfc1 %[temp6], %[src6] \n\t"
yading@10 242 "mfc1 %[temp7], %[src7] \n\t"
yading@10 243 "sw %[temp0], 0(%[dst]) \n\t"
yading@10 244 "sw %[temp1], 4(%[dst]) \n\t"
yading@10 245 "sw %[temp2], 8(%[dst]) \n\t"
yading@10 246 "sw %[temp3], 12(%[dst]) \n\t"
yading@10 247 "sw %[temp4], 16(%[dst]) \n\t"
yading@10 248 "sw %[temp5], 20(%[dst]) \n\t"
yading@10 249 "sw %[temp6], 24(%[dst]) \n\t"
yading@10 250 "sw %[temp7], 28(%[dst]) \n\t"
yading@10 251
yading@10 252 : [dst] "+r" (dst), [src] "+r" (src),
yading@10 253 [src0] "=&f" (src0), [src1] "=&f" (src1),
yading@10 254 [src2] "=&f" (src2), [src3] "=&f" (src3),
yading@10 255 [src4] "=&f" (src4), [src5] "=&f" (src5),
yading@10 256 [src6] "=&f" (src6), [src7] "=&f" (src7),
yading@10 257 [temp0] "=r" (temp0), [temp1] "=r" (temp1),
yading@10 258 [temp2] "=r" (temp2), [temp3] "=r" (temp3),
yading@10 259 [temp4] "=r" (temp4), [temp5] "=r" (temp5),
yading@10 260 [temp6] "=r" (temp6), [temp7] "=r" (temp7)
yading@10 261 : [scale] "f" (scale)
yading@10 262 : "memory"
yading@10 263 );
yading@10 264 src = src + 8;
yading@10 265 dst = dst + 8;
yading@10 266 len -= 8;
yading@10 267 } while (len > 0);
yading@10 268 }
yading@10 269
yading@10 270 static void ac3_downmix_mips(float **samples, float (*matrix)[2],
yading@10 271 int out_ch, int in_ch, int len)
yading@10 272 {
yading@10 273 int i, j, i1, i2, i3;
yading@10 274 float v0, v1, v2, v3;
yading@10 275 float v4, v5, v6, v7;
yading@10 276 float samples0, samples1, samples2, samples3, matrix_j, matrix_j2;
yading@10 277 float *samples_p,*matrix_p, **samples_x, **samples_end, **samples_sw;
yading@10 278
yading@10 279 __asm__ volatile(
yading@10 280 ".set push \n\t"
yading@10 281 ".set noreorder \n\t"
yading@10 282
yading@10 283 "li %[i1], 2 \n\t"
yading@10 284 "sll %[len], 2 \n\t"
yading@10 285 "move %[i], $zero \n\t"
yading@10 286 "sll %[j], %[in_ch], 2 \n\t"
yading@10 287
yading@10 288 "bne %[out_ch], %[i1], 3f \n\t" // if (out_ch == 2)
yading@10 289 " li %[i2], 1 \n\t"
yading@10 290
yading@10 291 "2: \n\t" // start of the for loop (for (i = 0; i < len; i+=4))
yading@10 292 "move %[matrix_p], %[matrix] \n\t"
yading@10 293 "move %[samples_x], %[samples] \n\t"
yading@10 294 "mtc1 $zero, %[v0] \n\t"
yading@10 295 "mtc1 $zero, %[v1] \n\t"
yading@10 296 "mtc1 $zero, %[v2] \n\t"
yading@10 297 "mtc1 $zero, %[v3] \n\t"
yading@10 298 "mtc1 $zero, %[v4] \n\t"
yading@10 299 "mtc1 $zero, %[v5] \n\t"
yading@10 300 "mtc1 $zero, %[v6] \n\t"
yading@10 301 "mtc1 $zero, %[v7] \n\t"
yading@10 302 "addiu %[i1], %[i], 4 \n\t"
yading@10 303 "addiu %[i2], %[i], 8 \n\t"
yading@10 304 "lw %[samples_p], 0(%[samples_x]) \n\t"
yading@10 305 "addiu %[i3], %[i], 12 \n\t"
yading@10 306 "addu %[samples_end], %[samples_x], %[j] \n\t"
yading@10 307 "move %[samples_sw], %[samples_p] \n\t"
yading@10 308
yading@10 309 "1: \n\t" // start of the inner for loop (for (j = 0; j < in_ch; j++))
yading@10 310 "lwc1 %[matrix_j], 0(%[matrix_p]) \n\t"
yading@10 311 "lwc1 %[matrix_j2], 4(%[matrix_p]) \n\t"
yading@10 312 "lwxc1 %[samples0], %[i](%[samples_p]) \n\t"
yading@10 313 "lwxc1 %[samples1], %[i1](%[samples_p]) \n\t"
yading@10 314 "lwxc1 %[samples2], %[i2](%[samples_p]) \n\t"
yading@10 315 "lwxc1 %[samples3], %[i3](%[samples_p]) \n\t"
yading@10 316 "addiu %[matrix_p], 8 \n\t"
yading@10 317 "addiu %[samples_x], 4 \n\t"
yading@10 318 "madd.s %[v0], %[v0], %[samples0], %[matrix_j] \n\t"
yading@10 319 "madd.s %[v1], %[v1], %[samples1], %[matrix_j] \n\t"
yading@10 320 "madd.s %[v2], %[v2], %[samples2], %[matrix_j] \n\t"
yading@10 321 "madd.s %[v3], %[v3], %[samples3], %[matrix_j] \n\t"
yading@10 322 "madd.s %[v4], %[v4], %[samples0], %[matrix_j2]\n\t"
yading@10 323 "madd.s %[v5], %[v5], %[samples1], %[matrix_j2]\n\t"
yading@10 324 "madd.s %[v6], %[v6], %[samples2], %[matrix_j2]\n\t"
yading@10 325 "madd.s %[v7], %[v7], %[samples3], %[matrix_j2]\n\t"
yading@10 326 "bne %[samples_x], %[samples_end], 1b \n\t"
yading@10 327 " lw %[samples_p], 0(%[samples_x]) \n\t"
yading@10 328
yading@10 329 "lw %[samples_p], 4(%[samples]) \n\t"
yading@10 330 "swxc1 %[v0], %[i](%[samples_sw]) \n\t"
yading@10 331 "swxc1 %[v1], %[i1](%[samples_sw]) \n\t"
yading@10 332 "swxc1 %[v2], %[i2](%[samples_sw]) \n\t"
yading@10 333 "swxc1 %[v3], %[i3](%[samples_sw]) \n\t"
yading@10 334 "swxc1 %[v4], %[i](%[samples_p]) \n\t"
yading@10 335 "addiu %[i], 16 \n\t"
yading@10 336 "swxc1 %[v5], %[i1](%[samples_p]) \n\t"
yading@10 337 "swxc1 %[v6], %[i2](%[samples_p]) \n\t"
yading@10 338 "bne %[i], %[len], 2b \n\t"
yading@10 339 " swxc1 %[v7], %[i3](%[samples_p]) \n\t"
yading@10 340
yading@10 341 "3: \n\t"
yading@10 342 "bne %[out_ch], %[i2], 6f \n\t" // if (out_ch == 1)
yading@10 343 " nop \n\t"
yading@10 344
yading@10 345 "5: \n\t" // start of the outer for loop (for (i = 0; i < len; i+=4))
yading@10 346 "move %[matrix_p], %[matrix] \n\t"
yading@10 347 "move %[samples_x], %[samples] \n\t"
yading@10 348 "mtc1 $zero, %[v0] \n\t"
yading@10 349 "mtc1 $zero, %[v1] \n\t"
yading@10 350 "mtc1 $zero, %[v2] \n\t"
yading@10 351 "mtc1 $zero, %[v3] \n\t"
yading@10 352 "addiu %[i1], %[i], 4 \n\t"
yading@10 353 "addiu %[i2], %[i], 8 \n\t"
yading@10 354 "lw %[samples_p], 0(%[samples_x]) \n\t"
yading@10 355 "addiu %[i3], %[i], 12 \n\t"
yading@10 356 "addu %[samples_end], %[samples_x], %[j] \n\t"
yading@10 357 "move %[samples_sw], %[samples_p] \n\t"
yading@10 358
yading@10 359 "4: \n\t" // start of the inner for loop (for (j = 0; j < in_ch; j++))
yading@10 360 "lwc1 %[matrix_j], 0(%[matrix_p]) \n\t"
yading@10 361 "lwxc1 %[samples0], %[i](%[samples_p]) \n\t"
yading@10 362 "lwxc1 %[samples1], %[i1](%[samples_p]) \n\t"
yading@10 363 "lwxc1 %[samples2], %[i2](%[samples_p]) \n\t"
yading@10 364 "lwxc1 %[samples3], %[i3](%[samples_p]) \n\t"
yading@10 365 "addiu %[matrix_p], 8 \n\t"
yading@10 366 "addiu %[samples_x], 4 \n\t"
yading@10 367 "madd.s %[v0], %[v0], %[samples0], %[matrix_j] \n\t"
yading@10 368 "madd.s %[v1], %[v1], %[samples1], %[matrix_j] \n\t"
yading@10 369 "madd.s %[v2], %[v2], %[samples2], %[matrix_j] \n\t"
yading@10 370 "madd.s %[v3], %[v3], %[samples3], %[matrix_j] \n\t"
yading@10 371 "bne %[samples_x], %[samples_end], 4b \n\t"
yading@10 372 " lw %[samples_p], 0(%[samples_x]) \n\t"
yading@10 373
yading@10 374 "swxc1 %[v0], %[i](%[samples_sw]) \n\t"
yading@10 375 "addiu %[i], 16 \n\t"
yading@10 376 "swxc1 %[v1], %[i1](%[samples_sw]) \n\t"
yading@10 377 "swxc1 %[v2], %[i2](%[samples_sw]) \n\t"
yading@10 378 "bne %[i], %[len], 5b \n\t"
yading@10 379 " swxc1 %[v3], %[i3](%[samples_sw]) \n\t"
yading@10 380 "6: \n\t"
yading@10 381
yading@10 382 ".set pop"
yading@10 383 :[samples_p]"=&r"(samples_p), [matrix_j]"=&f"(matrix_j), [matrix_j2]"=&f"(matrix_j2),
yading@10 384 [samples0]"=&f"(samples0), [samples1]"=&f"(samples1),
yading@10 385 [samples2]"=&f"(samples2), [samples3]"=&f"(samples3),
yading@10 386 [v0]"=&f"(v0), [v1]"=&f"(v1), [v2]"=&f"(v2), [v3]"=&f"(v3),
yading@10 387 [v4]"=&f"(v4), [v5]"=&f"(v5), [v6]"=&f"(v6), [v7]"=&f"(v7),
yading@10 388 [samples_x]"=&r"(samples_x), [matrix_p]"=&r"(matrix_p),
yading@10 389 [samples_end]"=&r"(samples_end), [samples_sw]"=&r"(samples_sw),
yading@10 390 [i1]"=&r"(i1), [i2]"=&r"(i2), [i3]"=&r"(i3), [i]"=&r"(i),
yading@10 391 [j]"=&r"(j), [len]"+r"(len)
yading@10 392 :[samples]"r"(samples), [matrix]"r"(matrix),
yading@10 393 [in_ch]"r"(in_ch), [out_ch]"r"(out_ch)
yading@10 394 :"memory"
yading@10 395 );
yading@10 396 }
yading@10 397 #endif
yading@10 398 #endif /* HAVE_INLINE_ASM */
yading@10 399
yading@10 400 void ff_ac3dsp_init_mips(AC3DSPContext *c, int bit_exact) {
yading@10 401 #if HAVE_INLINE_ASM
yading@10 402 #if HAVE_MIPSDSPR1
yading@10 403 c->bit_alloc_calc_bap = ac3_bit_alloc_calc_bap_mips;
yading@10 404 c->update_bap_counts = ac3_update_bap_counts_mips;
yading@10 405 #endif
yading@10 406 #if HAVE_MIPSFPU && HAVE_MIPS32R2
yading@10 407 c->float_to_fixed24 = float_to_fixed24_mips;
yading@10 408 c->downmix = ac3_downmix_mips;
yading@10 409 #endif
yading@10 410 #endif
yading@10 411
yading@10 412 }