[FFmpeg-devel] [PATCH 2/5] libavcodec: Implementation of AC3 fixedpoint decoder

Nedeljko Babic Nedeljko.Babic at imgtec.com
Tue Mar 18 11:38:33 CET 2014


>> diff --git a/Changelog b/Changelog
>> index 7d50591..a4c3566b 100644
>> --- a/Changelog
>> +++ b/Changelog
>> @@ -7,6 +7,7 @@ version <next>
>>  
>>  version 2.2:
>>  
>> +- added AC3 fixed-point decoding
>
>This should be under <next>.
>

You are correct and this will be fixed.

>
>> diff --git a/libavcodec/ac3dec_fixed.c b/libavcodec/ac3dec_fixed.c
>> new file mode 100644
>> index 0000000..aeb93e1
>> --- /dev/null
>> +++ b/libavcodec/ac3dec_fixed.c
>> @@ -0,0 +1,176 @@
>> +/*
>> + * Copyright (c) 2012
>> + *      MIPS Technologies, Inc., California.
>> + *
>> + * Redistribution and use in source and binary forms, with or without
>> + * modification, are permitted provided that the following conditions
>> + * are met:
>> + * 1. Redistributions of source code must retain the above copyright
>> + *    notice, this list of conditions and the following disclaimer.
>> + * 2. Redistributions in binary form must reproduce the above copyright
>> + *    notice, this list of conditions and the following disclaimer in the
>> + *    documentation and/or other materials provided with the distribution.
>> + * 3. Neither the name of the MIPS Technologies, Inc., nor the names of its
>> + *    contributors may be used to endorse or promote products derived from
>> + *    this software without specific prior written permission.
>> + *
>> + * THIS SOFTWARE IS PROVIDED BY THE MIPS TECHNOLOGIES, INC. ``AS IS'' AND
>> + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
>> + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
>> + * ARE DISCLAIMED.  IN NO EVENT SHALL THE MIPS TECHNOLOGIES, INC. BE LIABLE
>> + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
>> + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
>> + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
>> + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
>> + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
>> + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
>> + * SUCH DAMAGE.
>> + *
>> + * Author:  Stanislav Ocovaj (socovaj at mips.com)
>> + *
>> + * AC3 fixed-point decoder for MIPS platforms
>> + *
>> + * This file is part of FFmpeg.
>> + *
>> + * FFmpeg is free software; you can redistribute it and/or
>> + * modify it under the terms of the GNU Lesser General Public
>> + * License as published by the Free Software Foundation; either
>> + * version 2.1 of the License, or (at your option) any later version.
>> + *
>> + * FFmpeg is distributed in the hope that it will be useful,
>> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
>> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
>> + * Lesser General Public License for more details.
>> + *
>> + * You should have received a copy of the GNU Lesser General Public
>> + * License along with FFmpeg; if not, write to the Free Software
>> + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
>> + */
>> +
>> +#define FFT_FLOAT 0
>> +#define CONFIG_AC3_FIXED 1
>> +#define FFT_FIXED_32 1
>> +#include "ac3dec.h"
>> +
>> +
>> +/**
>> + * Table for center mix levels
>> + * reference: Section 5.4.2.4 cmixlev
>> + */
>> +static const uint8_t center_levels[4] = { 4, 5, 6, 5 };
>> +
>> +/**
>> + * Table for surround mix levels
>> + * reference: Section 5.4.2.5 surmixlev
>> + */
>> +static const uint8_t surround_levels[4] = { 4, 6, 7, 6 };
>> +
>> +int end_freq_inv_tab[8] =
>> +{
>> +    50529027, 44278013, 39403370, 32292987, 27356480, 23729101, 20951060, 18755316
>> +};
>> +
>> +static void scale_coefs (
>> +    int32_t *dst,
>> +    const int32_t *src,
>> +    int dynrng,
>> +    int len)
>> +{
>> +    int i, shift, round;
>> +    int16_t mul;
>> +    int temp, temp1, temp2, temp3, temp4, temp5, temp6, temp7;
>> +
>> +    mul = (dynrng & 0x1f) + 0x20;
>> +    shift = 4 - ((dynrng << 24) >> 29);
>> +    round = 1 << (shift-1);
>> +    for (i=0; i<len; i+=8) {
>> +
>> +        temp = src[i] * mul;
>> +        temp1 = src[i+1] * mul;
>> +        temp = temp + round;
>> +        temp2 = src[i+2] * mul;
>> +
>> +        temp1 = temp1 + round;
>> +        dst[i] = temp >> shift;
>> +        temp3 = src[i+3] * mul;
>> +        temp2 = temp2 + round;
>> +
>> +        dst[i+1] = temp1 >> shift;
>> +        temp4 = src[i + 4] * mul;
>> +        temp3 = temp3 + round;
>> +        dst[i+2] = temp2 >> shift;
>> +
>> +        temp5 = src[i+5] * mul;
>> +        temp4 = temp4 + round;
>> +        dst[i+3] = temp3 >> shift;
>> +        temp6 = src[i+6] * mul;
>> +
>> +        dst[i+4] = temp4 >> shift;
>> +        temp5 = temp5 + round;
>> +        temp7 = src[i+7] * mul;
>> +        temp6 = temp6 + round;
>> +
>> +        dst[i+5] = temp5 >> shift;
>> +        temp7 = temp7 + round;
>> +        dst[i+6] = temp6 >> shift;
>> +        dst[i+7] = temp7 >> shift;
>> +
>> +    }
>> +}
>> +
>
>I'm fairly certain this function can be optimized with simd. Maybe add it to AC3DSPContext?

This function is written in a way that compiler (or programmer ) can optimize it quite easy.
However, moving it to AC3DSPContext would require adding pointer to it and accessing it by
using that pointer and it would bring overhead to execution.

>
>> +/**
>> + * Downmix samples from original signal to stereo or mono (this is for 16-bit samples
>> + * and fixed point decoder - original (for 32-bit samples) is in ac3dsp.c).
>> + */
>> +static void ac3_downmix_c_fixed16(int16_t **samples, int16_t (*matrix)[2],
>> +                                  int out_ch, int in_ch, int len)
>> +{
>> +    int i, j;
>> +    int v0, v1;
>> +    if (out_ch == 2) {
>> +        for (i = 0; i < len; i++) {
>> +            v0 = v1 = 0;
>> +            for (j = 0; j < in_ch; j++) {
>> +                v0 += samples[j][i] * matrix[j][0];
>> +                v1 += samples[j][i] * matrix[j][1];
>> +            }
>> +            samples[0][i] = (v0+2048)>>12;
>> +            samples[1][i] = (v1+2048)>>12;
>> +        }
>> +    } else if (out_ch == 1) {
>> +        for (i = 0; i < len; i++) {
>> +            v0 = 0;
>> +            for (j = 0; j < in_ch; j++)
>> +                v0 += samples[j][i] * matrix[j][0];
>> +            samples[0][i] = (v0+2048)>>12;
>> +        }
>> +    }
>> +}
>> +
>
>I'd say the same for this one, but i assume there was a reason you didn't add it alongside 
>the 32-bits one.
>

Similar as above and this one is used in only one place.

>
>> diff --git a/libavcodec/allcodecs.c b/libavcodec/allcodecs.c
>> index 715837e..d93bb42 100644
>> --- a/libavcodec/allcodecs.c
>> +++ b/libavcodec/allcodecs.c
>> @@ -322,6 +322,7 @@ void avcodec_register_all(void)
>>      REGISTER_DECODER(AAC_LATM,          aac_latm);
>>      REGISTER_ENCDEC (AC3,               ac3);
>>      REGISTER_ENCODER(AC3_FIXED,         ac3_fixed);
>> +    REGISTER_DECODER(AC3FIXED,          ac3fixed);
>
>You could use the same name as the encoder, and REGISTER_ENCDEC()
>

You are right and this will be changed.

- Nedeljko


More information about the ffmpeg-devel mailing list