libavcodec/sbrdsp_fixed.c - third_party/ffmpeg - Git at Google

 /*
  * AAC Spectral Band Replication decoding functions
  * Copyright (c) 2008-2009 Robert Swain ( rob opendot cl )
  * Copyright (c) 2009-2010 Alex Converse <alex.converse@gmail.com>
  *
  * 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
  *
  * Note: Rounding-to-nearest used unless otherwise stated
  *
  */

 #define USE_FIXED 1

 #include "aac.h"
 #include "config.h"
 #include "libavutil/attributes.h"
 #include "libavutil/intfloat.h"
 #include "sbrdsp.h"

 static SoftFloat sbr_sum_square_c(int (*x)[2], int n)
 {
     SoftFloat ret;
     uint64_t accu, round;
     uint64_t accu0 = 0, accu1 = 0, accu2 = 0, accu3 = 0;
     int i, nz, nz0;
     unsigned u;

     for (i = 0; i < n; i += 2) {
         // Larger values are inavlid and could cause overflows of accu.
         av_assert2(FFABS(x[i + 0][0]) >> 30 == 0);
         accu0 += (int64_t)x[i + 0][0] * x[i + 0][0];
         av_assert2(FFABS(x[i + 0][1]) >> 30 == 0);
         accu1 += (int64_t)x[i + 0][1] * x[i + 0][1];
         av_assert2(FFABS(x[i + 1][0]) >> 30 == 0);
         accu2 += (int64_t)x[i + 1][0] * x[i + 1][0];
         av_assert2(FFABS(x[i + 1][1]) >> 30 == 0);
         accu3 += (int64_t)x[i + 1][1] * x[i + 1][1];
     }

     nz0 = 15;
     while ((accu0|accu1|accu2|accu3) >> 62) {
         accu0 >>= 1;
         accu1 >>= 1;
         accu2 >>= 1;
         accu3 >>= 1;
         nz0 --;
     }
     accu = accu0 + accu1 + accu2 + accu3;

     u = accu >> 32;
     if (u) {
         nz = 33;
         while (u < 0x80000000U) {
             u <<= 1;
             nz--;
         }
     } else
         nz = 1;

     round = 1ULL << (nz-1);
     u = ((accu + round) >> nz);
     u >>= 1;
     ret = av_int2sf(u, nz0 - nz);

     return ret;
 }

 static void sbr_neg_odd_64_c(int *x)
 {
     int i;
     for (i = 1; i < 64; i += 2)
         x[i] = -x[i];
 }

 static void sbr_qmf_pre_shuffle_c(int *z)
 {
     int k;
     z[64] = z[0];
     z[65] = z[1];
     for (k = 1; k < 32; k++) {
         z[64+2*k  ] = -z[64 - k];
         z[64+2*k+1] =  z[ k + 1];
     }
 }

 static void sbr_qmf_post_shuffle_c(int W[32][2], const int *z)
 {
     int k;
     for (k = 0; k < 32; k++) {
         W[k][0] = -z[63-k];
         W[k][1] = z[k];
     }
 }

 static void sbr_qmf_deint_neg_c(int *v, const int *src)
 {
     int i;
     for (i = 0; i < 32; i++) {
         v[     i] = ( src[63 - 2*i    ] + 0x10) >> 5;
         v[63 - i] = (-src[63 - 2*i - 1] + 0x10) >> 5;
     }
 }

 static av_always_inline SoftFloat autocorr_calc(int64_t accu)
 {
         int nz, mant, expo;
         unsigned round;
         int i = (int)(accu >> 32);
         if (i == 0) {
             nz = 1;
         } else {
             nz = 0;
             while (FFABS(i) < 0x40000000) {
                 i *= 2;
                 nz++;
             }
             nz = 32-nz;
         }

         round = 1U << (nz-1);
         mant = (int)((accu + round) >> nz);
         mant = (mant + 0x40LL)>>7;
         mant *= 64;
         expo = nz + 15;
         return av_int2sf(mant, 30 - expo);
 }

 static av_always_inline void autocorrelate(const int x[40][2], SoftFloat phi[3][2][2], int lag)
 {
     int i;
     int64_t real_sum, imag_sum;
     int64_t accu_re = 0, accu_im = 0;

     if (lag) {
         for (i = 1; i < 38; i++) {
             accu_re += (uint64_t)x[i][0] * x[i+lag][0];
             accu_re += (uint64_t)x[i][1] * x[i+lag][1];
             accu_im += (uint64_t)x[i][0] * x[i+lag][1];
             accu_im -= (uint64_t)x[i][1] * x[i+lag][0];
         }

         real_sum = accu_re;
         imag_sum = accu_im;

         accu_re += (uint64_t)x[ 0][0] * x[lag][0];
         accu_re += (uint64_t)x[ 0][1] * x[lag][1];
         accu_im += (uint64_t)x[ 0][0] * x[lag][1];
         accu_im -= (uint64_t)x[ 0][1] * x[lag][0];

         phi[2-lag][1][0] = autocorr_calc(accu_re);
         phi[2-lag][1][1] = autocorr_calc(accu_im);

         if (lag == 1) {
             accu_re = real_sum;
             accu_im = imag_sum;
             accu_re += (uint64_t)x[38][0] * x[39][0];
             accu_re += (uint64_t)x[38][1] * x[39][1];
             accu_im += (uint64_t)x[38][0] * x[39][1];
             accu_im -= (uint64_t)x[38][1] * x[39][0];

             phi[0][0][0] = autocorr_calc(accu_re);
             phi[0][0][1] = autocorr_calc(accu_im);
         }
     } else {
         for (i = 1; i < 38; i++) {
             accu_re += (uint64_t)x[i][0] * x[i][0];
             accu_re += (uint64_t)x[i][1] * x[i][1];
         }
         real_sum = accu_re;
         accu_re += (uint64_t)x[ 0][0] * x[ 0][0];
         accu_re += (uint64_t)x[ 0][1] * x[ 0][1];

         phi[2][1][0] = autocorr_calc(accu_re);

         accu_re = real_sum;
         accu_re += (uint64_t)x[38][0] * x[38][0];
         accu_re += (uint64_t)x[38][1] * x[38][1];

         phi[1][0][0] = autocorr_calc(accu_re);
     }
 }

 static void sbr_autocorrelate_c(const int x[40][2], SoftFloat phi[3][2][2])
 {
     autocorrelate(x, phi, 0);
     autocorrelate(x, phi, 1);
     autocorrelate(x, phi, 2);
 }

 static void sbr_hf_gen_c(int (*X_high)[2], const int (*X_low)[2],
                        const int alpha0[2], const int alpha1[2],
                        int bw, int start, int end)
 {
     int alpha[4];
     int i;
     int64_t accu;

     accu = (int64_t)alpha0[0] * bw;
     alpha[2] = (int)((accu + 0x40000000) >> 31);
     accu = (int64_t)alpha0[1] * bw;
     alpha[3] = (int)((accu + 0x40000000) >> 31);
     accu = (int64_t)bw * bw;
     bw = (int)((accu + 0x40000000) >> 31);
     accu = (int64_t)alpha1[0] * bw;
     alpha[0] = (int)((accu + 0x40000000) >> 31);
     accu = (int64_t)alpha1[1] * bw;
     alpha[1] = (int)((accu + 0x40000000) >> 31);

     for (i = start; i < end; i++) {
         accu  = (int64_t)X_low[i][0] * 0x20000000;
         accu += (int64_t)X_low[i - 2][0] * alpha[0];
         accu -= (int64_t)X_low[i - 2][1] * alpha[1];
         accu += (int64_t)X_low[i - 1][0] * alpha[2];
         accu -= (int64_t)X_low[i - 1][1] * alpha[3];
         X_high[i][0] = (int)((accu + 0x10000000) >> 29);

         accu  = (int64_t)X_low[i][1] * 0x20000000;
         accu += (int64_t)X_low[i - 2][1] * alpha[0];
         accu += (int64_t)X_low[i - 2][0] * alpha[1];
         accu += (int64_t)X_low[i - 1][1] * alpha[2];
         accu += (int64_t)X_low[i - 1][0] * alpha[3];
         X_high[i][1] = (int)((accu + 0x10000000) >> 29);
     }
 }

 static void sbr_hf_g_filt_c(int (*Y)[2], const int (*X_high)[40][2],
                           const SoftFloat *g_filt, int m_max, intptr_t ixh)
 {
     int m;
     int64_t accu;

     for (m = 0; m < m_max; m++) {
         if (22 - g_filt[m].exp < 61) {
             int64_t r = 1LL << (22-g_filt[m].exp);
             accu = (int64_t)X_high[m][ixh][0] * ((g_filt[m].mant + 0x40)>>7);
             Y[m][0] = (int)((accu + r) >> (23-g_filt[m].exp));

             accu = (int64_t)X_high[m][ixh][1] * ((g_filt[m].mant + 0x40)>>7);
             Y[m][1] = (int)((accu + r) >> (23-g_filt[m].exp));
         }
     }
 }

 static av_always_inline int sbr_hf_apply_noise(int (*Y)[2],
                                                 const SoftFloat *s_m,
                                                 const SoftFloat *q_filt,
                                                 int noise,
                                                 int phi_sign0,
                                                 int phi_sign1,
                                                 int m_max)
 {
     int m;

     for (m = 0; m < m_max; m++) {
         unsigned y0 = Y[m][0];
         unsigned y1 = Y[m][1];
         noise = (noise + 1) & 0x1ff;
         if (s_m[m].mant) {
             int shift, round;

             shift = 22 - s_m[m].exp;
             if (shift < 1) {
                 av_log(NULL, AV_LOG_ERROR, "Overflow in sbr_hf_apply_noise, shift=%d\n", shift);
                 return AVERROR(ERANGE);
             } else if (shift < 30) {
                 round = 1 << (shift-1);
                 y0 += (s_m[m].mant * phi_sign0 + round) >> shift;
                 y1 += (s_m[m].mant * phi_sign1 + round) >> shift;
             }
         } else {
             int shift, round, tmp;
             int64_t accu;

             shift = 22 - q_filt[m].exp;
             if (shift < 1) {
                 av_log(NULL, AV_LOG_ERROR, "Overflow in sbr_hf_apply_noise, shift=%d\n", shift);
                 return AVERROR(ERANGE);
             } else if (shift < 30) {
                 round = 1 << (shift-1);

                 accu = (int64_t)q_filt[m].mant * ff_sbr_noise_table_fixed[noise][0];
                 tmp = (int)((accu + 0x40000000) >> 31);
                 y0 += (tmp + round) >> shift;

                 accu = (int64_t)q_filt[m].mant * ff_sbr_noise_table_fixed[noise][1];
                 tmp = (int)((accu + 0x40000000) >> 31);
                 y1 += (tmp + round) >> shift;
             }
         }
         Y[m][0] = y0;
         Y[m][1] = y1;
         phi_sign1 = -phi_sign1;
     }
     return 0;
 }

 #include "sbrdsp_template.c"
	/*
	* AAC Spectral Band Replication decoding functions
	* Copyright (c) 2008-2009 Robert Swain ( rob opendot cl )
	* Copyright (c) 2009-2010 Alex Converse <alex.converse@gmail.com>
	*
	* 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
	*
	* Note: Rounding-to-nearest used unless otherwise stated
	*
	*/

	#define USE_FIXED 1

	#include "aac.h"
	#include "config.h"
	#include "libavutil/attributes.h"
	#include "libavutil/intfloat.h"
	#include "sbrdsp.h"

	static SoftFloat sbr_sum_square_c(int (*x)[2], int n)
	{
	SoftFloat ret;
	uint64_t accu, round;
	uint64_t accu0 = 0, accu1 = 0, accu2 = 0, accu3 = 0;
	int i, nz, nz0;
	unsigned u;

	for (i = 0; i < n; i += 2) {
	// Larger values are inavlid and could cause overflows of accu.
	av_assert2(FFABS(x[i + 0][0]) >> 30 == 0);
	accu0 += (int64_t)x[i + 0][0] * x[i + 0][0];
	av_assert2(FFABS(x[i + 0][1]) >> 30 == 0);
	accu1 += (int64_t)x[i + 0][1] * x[i + 0][1];
	av_assert2(FFABS(x[i + 1][0]) >> 30 == 0);
	accu2 += (int64_t)x[i + 1][0] * x[i + 1][0];
	av_assert2(FFABS(x[i + 1][1]) >> 30 == 0);
	accu3 += (int64_t)x[i + 1][1] * x[i + 1][1];
	}

	nz0 = 15;
	while ((accu0\|accu1\|accu2\|accu3) >> 62) {
	accu0 >>= 1;
	accu1 >>= 1;
	accu2 >>= 1;
	accu3 >>= 1;
	nz0 --;
	}
	accu = accu0 + accu1 + accu2 + accu3;

	u = accu >> 32;
	if (u) {
	nz = 33;
	while (u < 0x80000000U) {
	u <<= 1;
	nz--;
	}
	} else
	nz = 1;

	round = 1ULL << (nz-1);
	u = ((accu + round) >> nz);
	u >>= 1;
	ret = av_int2sf(u, nz0 - nz);

	return ret;
	}

	static void sbr_neg_odd_64_c(int *x)
	{
	int i;
	for (i = 1; i < 64; i += 2)
	x[i] = -x[i];
	}

	static void sbr_qmf_pre_shuffle_c(int *z)
	{
	int k;
	z[64] = z[0];
	z[65] = z[1];
	for (k = 1; k < 32; k++) {
	z[64+2*k ] = -z[64 - k];
	z[64+2*k+1] = z[ k + 1];
	}
	}

	static void sbr_qmf_post_shuffle_c(int W[32][2], const int *z)
	{
	int k;
	for (k = 0; k < 32; k++) {
	W[k][0] = -z[63-k];
	W[k][1] = z[k];
	}
	}

	static void sbr_qmf_deint_neg_c(int v, const int src)
	{
	int i;
	for (i = 0; i < 32; i++) {
	v[ i] = ( src[63 - 2*i ] + 0x10) >> 5;
	v[63 - i] = (-src[63 - 2*i - 1] + 0x10) >> 5;
	}
	}

	static av_always_inline SoftFloat autocorr_calc(int64_t accu)
	{
	int nz, mant, expo;
	unsigned round;
	int i = (int)(accu >> 32);
	if (i == 0) {
	nz = 1;
	} else {
	nz = 0;
	while (FFABS(i) < 0x40000000) {
	i *= 2;
	nz++;
	}
	nz = 32-nz;
	}

	round = 1U << (nz-1);
	mant = (int)((accu + round) >> nz);
	mant = (mant + 0x40LL)>>7;
	mant *= 64;
	expo = nz + 15;
	return av_int2sf(mant, 30 - expo);
	}

	static av_always_inline void autocorrelate(const int x[40][2], SoftFloat phi[3][2][2], int lag)
	{
	int i;
	int64_t real_sum, imag_sum;
	int64_t accu_re = 0, accu_im = 0;

	if (lag) {
	for (i = 1; i < 38; i++) {
	accu_re += (uint64_t)x[i][0] * x[i+lag][0];
	accu_re += (uint64_t)x[i][1] * x[i+lag][1];
	accu_im += (uint64_t)x[i][0] * x[i+lag][1];
	accu_im -= (uint64_t)x[i][1] * x[i+lag][0];
	}

	real_sum = accu_re;
	imag_sum = accu_im;

	accu_re += (uint64_t)x[ 0][0] * x[lag][0];
	accu_re += (uint64_t)x[ 0][1] * x[lag][1];
	accu_im += (uint64_t)x[ 0][0] * x[lag][1];
	accu_im -= (uint64_t)x[ 0][1] * x[lag][0];

	phi[2-lag][1][0] = autocorr_calc(accu_re);
	phi[2-lag][1][1] = autocorr_calc(accu_im);

	if (lag == 1) {
	accu_re = real_sum;
	accu_im = imag_sum;
	accu_re += (uint64_t)x[38][0] * x[39][0];
	accu_re += (uint64_t)x[38][1] * x[39][1];
	accu_im += (uint64_t)x[38][0] * x[39][1];
	accu_im -= (uint64_t)x[38][1] * x[39][0];

	phi[0][0][0] = autocorr_calc(accu_re);
	phi[0][0][1] = autocorr_calc(accu_im);
	}
	} else {
	for (i = 1; i < 38; i++) {
	accu_re += (uint64_t)x[i][0] * x[i][0];
	accu_re += (uint64_t)x[i][1] * x[i][1];
	}
	real_sum = accu_re;
	accu_re += (uint64_t)x[ 0][0] * x[ 0][0];
	accu_re += (uint64_t)x[ 0][1] * x[ 0][1];

	phi[2][1][0] = autocorr_calc(accu_re);

	accu_re = real_sum;
	accu_re += (uint64_t)x[38][0] * x[38][0];
	accu_re += (uint64_t)x[38][1] * x[38][1];

	phi[1][0][0] = autocorr_calc(accu_re);
	}
	}

	static void sbr_autocorrelate_c(const int x[40][2], SoftFloat phi[3][2][2])
	{
	autocorrelate(x, phi, 0);
	autocorrelate(x, phi, 1);
	autocorrelate(x, phi, 2);
	}

	static void sbr_hf_gen_c(int (X_high)[2], const int (X_low)[2],
	const int alpha0[2], const int alpha1[2],
	int bw, int start, int end)
	{
	int alpha[4];
	int i;
	int64_t accu;

	accu = (int64_t)alpha0[0] * bw;
	alpha[2] = (int)((accu + 0x40000000) >> 31);
	accu = (int64_t)alpha0[1] * bw;
	alpha[3] = (int)((accu + 0x40000000) >> 31);
	accu = (int64_t)bw * bw;
	bw = (int)((accu + 0x40000000) >> 31);
	accu = (int64_t)alpha1[0] * bw;
	alpha[0] = (int)((accu + 0x40000000) >> 31);
	accu = (int64_t)alpha1[1] * bw;
	alpha[1] = (int)((accu + 0x40000000) >> 31);

	for (i = start; i < end; i++) {
	accu = (int64_t)X_low[i][0] * 0x20000000;
	accu += (int64_t)X_low[i - 2][0] * alpha[0];
	accu -= (int64_t)X_low[i - 2][1] * alpha[1];
	accu += (int64_t)X_low[i - 1][0] * alpha[2];
	accu -= (int64_t)X_low[i - 1][1] * alpha[3];
	X_high[i][0] = (int)((accu + 0x10000000) >> 29);

	accu = (int64_t)X_low[i][1] * 0x20000000;
	accu += (int64_t)X_low[i - 2][1] * alpha[0];
	accu += (int64_t)X_low[i - 2][0] * alpha[1];
	accu += (int64_t)X_low[i - 1][1] * alpha[2];
	accu += (int64_t)X_low[i - 1][0] * alpha[3];
	X_high[i][1] = (int)((accu + 0x10000000) >> 29);
	}
	}

	static void sbr_hf_g_filt_c(int (Y)[2], const int (X_high)[40][2],
	const SoftFloat *g_filt, int m_max, intptr_t ixh)
	{
	int m;
	int64_t accu;

	for (m = 0; m < m_max; m++) {
	if (22 - g_filt[m].exp < 61) {
	int64_t r = 1LL << (22-g_filt[m].exp);
	accu = (int64_t)X_high[m][ixh][0] * ((g_filt[m].mant + 0x40)>>7);
	Y[m][0] = (int)((accu + r) >> (23-g_filt[m].exp));

	accu = (int64_t)X_high[m][ixh][1] * ((g_filt[m].mant + 0x40)>>7);
	Y[m][1] = (int)((accu + r) >> (23-g_filt[m].exp));
	}
	}
	}

	static av_always_inline int sbr_hf_apply_noise(int (*Y)[2],
	const SoftFloat *s_m,
	const SoftFloat *q_filt,
	int noise,
	int phi_sign0,
	int phi_sign1,
	int m_max)
	{
	int m;

	for (m = 0; m < m_max; m++) {
	unsigned y0 = Y[m][0];
	unsigned y1 = Y[m][1];
	noise = (noise + 1) & 0x1ff;
	if (s_m[m].mant) {
	int shift, round;

	shift = 22 - s_m[m].exp;
	if (shift < 1) {
	av_log(NULL, AV_LOG_ERROR, "Overflow in sbr_hf_apply_noise, shift=%d\n", shift);
	return AVERROR(ERANGE);
	} else if (shift < 30) {
	round = 1 << (shift-1);
	y0 += (s_m[m].mant * phi_sign0 + round) >> shift;
	y1 += (s_m[m].mant * phi_sign1 + round) >> shift;
	}
	} else {
	int shift, round, tmp;
	int64_t accu;

	shift = 22 - q_filt[m].exp;
	if (shift < 1) {
	av_log(NULL, AV_LOG_ERROR, "Overflow in sbr_hf_apply_noise, shift=%d\n", shift);
	return AVERROR(ERANGE);
	} else if (shift < 30) {
	round = 1 << (shift-1);

	accu = (int64_t)q_filt[m].mant * ff_sbr_noise_table_fixed[noise][0];
	tmp = (int)((accu + 0x40000000) >> 31);
	y0 += (tmp + round) >> shift;

	accu = (int64_t)q_filt[m].mant * ff_sbr_noise_table_fixed[noise][1];
	tmp = (int)((accu + 0x40000000) >> 31);
	y1 += (tmp + round) >> shift;
	}
	}
	Y[m][0] = y0;
	Y[m][1] = y1;
	phi_sign1 = -phi_sign1;
	}
	return 0;
	}

	#include "sbrdsp_template.c"