3 * Copyright (c) 2011 Justin Ruggles
5 * This file is part of FFmpeg.
7 * FFmpeg is free software; you can redistribute it and/or
8 * modify it under the terms of the GNU Lesser General Public
9 * License as published by the Free Software Foundation; either
10 * version 2.1 of the License, or (at your option) any later version.
12 * FFmpeg is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15 * Lesser General Public License for more details.
17 * You should have received a copy of the GNU Lesser General Public
18 * License along with FFmpeg; if not, write to the Free Software
19 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
22 #include "libavutil/avassert.h"
28 static void ac3_exponent_min_c(uint8_t *exp, int num_reuse_blocks, int nb_coefs)
32 if (!num_reuse_blocks)
35 for (i = 0; i < nb_coefs; i++) {
36 uint8_t min_exp = *exp;
37 uint8_t *exp1 = exp + 256;
38 for (blk = 0; blk < num_reuse_blocks; blk++) {
39 uint8_t next_exp = *exp1;
40 if (next_exp < min_exp)
48 static int ac3_max_msb_abs_int16_c(const int16_t *src, int len)
51 for (i = 0; i < len; i++)
56 static void ac3_lshift_int16_c(int16_t *src, unsigned int len,
59 uint32_t *src32 = (uint32_t *)src;
60 const uint32_t mask = ~(((1 << shift) - 1) << 16);
63 for (i = 0; i < len; i += 8) {
64 src32[i ] = (src32[i ] << shift) & mask;
65 src32[i+1] = (src32[i+1] << shift) & mask;
66 src32[i+2] = (src32[i+2] << shift) & mask;
67 src32[i+3] = (src32[i+3] << shift) & mask;
68 src32[i+4] = (src32[i+4] << shift) & mask;
69 src32[i+5] = (src32[i+5] << shift) & mask;
70 src32[i+6] = (src32[i+6] << shift) & mask;
71 src32[i+7] = (src32[i+7] << shift) & mask;
75 static void ac3_rshift_int32_c(int32_t *src, unsigned int len,
91 static void float_to_fixed24_c(int32_t *dst, const float *src, unsigned int len)
93 const float scale = 1 << 24;
95 *dst++ = lrintf(*src++ * scale);
96 *dst++ = lrintf(*src++ * scale);
97 *dst++ = lrintf(*src++ * scale);
98 *dst++ = lrintf(*src++ * scale);
99 *dst++ = lrintf(*src++ * scale);
100 *dst++ = lrintf(*src++ * scale);
101 *dst++ = lrintf(*src++ * scale);
102 *dst++ = lrintf(*src++ * scale);
107 static void ac3_bit_alloc_calc_bap_c(int16_t *mask, int16_t *psd,
109 int snr_offset, int floor,
110 const uint8_t *bap_tab, uint8_t *bap)
114 /* special case, if snr offset is -960, set all bap's to zero */
115 if (snr_offset == -960) {
116 memset(bap, 0, AC3_MAX_COEFS);
121 band = ff_ac3_bin_to_band_tab[start];
123 int m = (FFMAX(mask[band] - snr_offset - floor, 0) & 0x1FE0) + floor;
124 int band_end = FFMIN(ff_ac3_band_start_tab[band+1], end);
125 for (; bin < band_end; bin++) {
126 int address = av_clip((psd[bin] - m) >> 5, 0, 63);
127 bap[bin] = bap_tab[address];
129 } while (end > ff_ac3_band_start_tab[band++]);
132 static void ac3_update_bap_counts_c(uint16_t mant_cnt[16], uint8_t *bap,
136 mant_cnt[bap[len]]++;
139 DECLARE_ALIGNED(16, const uint16_t, ff_ac3_bap_bits)[16] = {
140 0, 0, 0, 3, 0, 4, 5, 6, 7, 8, 9, 10, 11, 12, 14, 16
143 static int ac3_compute_mantissa_size_c(uint16_t mant_cnt[6][16])
148 for (blk = 0; blk < AC3_MAX_BLOCKS; blk++) {
149 // bap=1 : 3 mantissas in 5 bits
150 bits += (mant_cnt[blk][1] / 3) * 5;
151 // bap=2 : 3 mantissas in 7 bits
152 // bap=4 : 2 mantissas in 7 bits
153 bits += ((mant_cnt[blk][2] / 3) + (mant_cnt[blk][4] >> 1)) * 7;
154 // bap=3 : 1 mantissa in 3 bits
155 bits += mant_cnt[blk][3] * 3;
156 // bap=5 to 15 : get bits per mantissa from table
157 for (bap = 5; bap < 16; bap++)
158 bits += mant_cnt[blk][bap] * ff_ac3_bap_bits[bap];
163 static void ac3_extract_exponents_c(uint8_t *exp, int32_t *coef, int nb_coefs)
167 for (i = 0; i < nb_coefs; i++) {
168 int v = abs(coef[i]);
169 exp[i] = v ? 23 - av_log2(v) : 24;
173 static void ac3_sum_square_butterfly_int32_c(int64_t sum[4],
174 const int32_t *coef0,
175 const int32_t *coef1,
180 sum[0] = sum[1] = sum[2] = sum[3] = 0;
182 for (i = 0; i < len; i++) {
187 MAC64(sum[0], lt, lt);
188 MAC64(sum[1], rt, rt);
189 MAC64(sum[2], md, md);
190 MAC64(sum[3], sd, sd);
194 static void ac3_sum_square_butterfly_float_c(float sum[4],
201 sum[0] = sum[1] = sum[2] = sum[3] = 0;
203 for (i = 0; i < len; i++) {
215 av_cold void ff_ac3dsp_init(AC3DSPContext *c, int bit_exact)
217 c->ac3_exponent_min = ac3_exponent_min_c;
218 c->ac3_max_msb_abs_int16 = ac3_max_msb_abs_int16_c;
219 c->ac3_lshift_int16 = ac3_lshift_int16_c;
220 c->ac3_rshift_int32 = ac3_rshift_int32_c;
221 c->float_to_fixed24 = float_to_fixed24_c;
222 c->bit_alloc_calc_bap = ac3_bit_alloc_calc_bap_c;
223 c->update_bap_counts = ac3_update_bap_counts_c;
224 c->compute_mantissa_size = ac3_compute_mantissa_size_c;
225 c->extract_exponents = ac3_extract_exponents_c;
226 c->sum_square_butterfly_int32 = ac3_sum_square_butterfly_int32_c;
227 c->sum_square_butterfly_float = ac3_sum_square_butterfly_float_c;
230 ff_ac3dsp_init_arm(c, bit_exact);
232 ff_ac3dsp_init_x86(c, bit_exact);