2 * This file is part of FFmpeg.
4 * FFmpeg is free software; you can redistribute it and/or
5 * modify it under the terms of the GNU Lesser General Public
6 * License as published by the Free Software Foundation; either
7 * version 2.1 of the License, or (at your option) any later version.
9 * FFmpeg is distributed in the hope that it will be useful,
10 * but WITHOUT ANY WARRANTY; without even the implied warranty of
11 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
12 * Lesser General Public License for more details.
14 * You should have received a copy of the GNU Lesser General Public
15 * License along with FFmpeg; if not, write to the Free Software
16 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
21 int ff_tx_type_is_mdct(enum AVTXType type)
24 case AV_TX_FLOAT_MDCT:
25 case AV_TX_DOUBLE_MDCT:
26 case AV_TX_INT32_MDCT:
33 /* Calculates the modular multiplicative inverse */
34 static av_always_inline int mulinv(int n, int m)
37 for (int x = 1; x < m; x++)
38 if (((n * x) % m) == 1)
40 av_assert0(0); /* Never reached */
43 /* Guaranteed to work for any n, m where gcd(n, m) == 1 */
44 int ff_tx_gen_compound_mapping(AVTXContext *s)
46 int *in_map, *out_map;
49 const int inv = s->inv;
51 const int m_inv = mulinv(m, n);
52 const int n_inv = mulinv(n, m);
53 const int mdct = ff_tx_type_is_mdct(s->type);
55 if (!(s->pfatab = av_malloc(2*len*sizeof(*s->pfatab))))
56 return AVERROR(ENOMEM);
59 out_map = s->pfatab + n*m;
61 /* Ruritanian map for input, CRT map for output, can be swapped */
62 for (int j = 0; j < m; j++) {
63 for (int i = 0; i < n; i++) {
64 /* Shifted by 1 to simplify MDCTs */
65 in_map[j*n + i] = ((i*m + j*n) % len) << mdct;
66 out_map[(i*m*m_inv + j*n*n_inv) % len] = i*m + j;
70 /* Change transform direction by reversing all ACs */
72 for (int i = 0; i < m; i++) {
73 int *in = &in_map[i*n + 1]; /* Skip the DC */
74 for (int j = 0; j < ((n - 1) >> 1); j++)
75 FFSWAP(int, in[j], in[n - j - 2]);
79 /* Our 15-point transform is also a compound one, so embed its input map */
81 for (int k = 0; k < m; k++) {
83 memcpy(tmp, &in_map[k*15], 15*sizeof(*tmp));
84 for (int i = 0; i < 5; i++) {
85 for (int j = 0; j < 3; j++)
86 in_map[k*15 + i*3 + j] = tmp[(i*3 + j*5) % 15];
94 static inline int split_radix_permutation(int i, int m, int inverse)
100 return (split_radix_permutation(i, m, inverse) << 1);
102 return (split_radix_permutation(i, m, inverse) << 2) + 1 - 2*(!(i & m) ^ inverse);
105 int ff_tx_gen_ptwo_revtab(AVTXContext *s, int invert_lookup)
107 const int m = s->m, inv = s->inv;
109 if (!(s->revtab = av_malloc(s->m*sizeof(*s->revtab))))
110 return AVERROR(ENOMEM);
111 if (!(s->revtab_c = av_malloc(m*sizeof(*s->revtab_c))))
112 return AVERROR(ENOMEM);
115 for (int i = 0; i < m; i++) {
116 int k = -split_radix_permutation(i, m, inv) & (m - 1);
118 s->revtab[i] = s->revtab_c[i] = k;
120 s->revtab[i] = s->revtab_c[k] = i;
126 int ff_tx_gen_ptwo_inplace_revtab_idx(AVTXContext *s, int *revtab)
128 int nb_inplace_idx = 0;
130 if (!(s->inplace_idx = av_malloc(s->m*sizeof(*s->inplace_idx))))
131 return AVERROR(ENOMEM);
133 /* The first coefficient is always already in-place */
134 for (int src = 1; src < s->m; src++) {
135 int dst = revtab[src];
141 /* This just checks if a closed loop has been encountered before,
142 * and if so, skips it, since to fully permute a loop we must only
145 for (int j = 0; j < nb_inplace_idx; j++) {
146 if (dst == s->inplace_idx[j]) {
152 } while (dst != src && !found);
155 s->inplace_idx[nb_inplace_idx++] = src;
158 s->inplace_idx[nb_inplace_idx++] = 0;
163 static void parity_revtab_generator(int *revtab, int n, int inv, int offset,
164 int is_dual, int dual_high, int len,
165 int basis, int dual_stride)
170 int k1, k2, *even, *odd, stride;
172 is_dual = is_dual && dual_stride;
173 dual_high = is_dual & dual_high;
174 stride = is_dual ? FFMIN(dual_stride, len) : 0;
176 even = &revtab[offset + dual_high*(stride - 2*len)];
177 odd = &even[len + (is_dual && !dual_high)*len + dual_high*len];
179 for (int i = 0; i < len; i++) {
180 k1 = -split_radix_permutation(offset + i*2 + 0, n, inv) & (n - 1);
181 k2 = -split_radix_permutation(offset + i*2 + 1, n, inv) & (n - 1);
184 if (stride && !((i + 1) % stride)) {
193 parity_revtab_generator(revtab, n, inv, offset,
194 0, 0, len >> 0, basis, dual_stride);
195 parity_revtab_generator(revtab, n, inv, offset + (len >> 0),
196 1, 0, len >> 1, basis, dual_stride);
197 parity_revtab_generator(revtab, n, inv, offset + (len >> 0) + (len >> 1),
198 1, 1, len >> 1, basis, dual_stride);
201 void ff_tx_gen_split_radix_parity_revtab(int *revtab, int len, int inv,
202 int basis, int dual_stride)
207 av_assert0(!dual_stride || !(dual_stride & (dual_stride - 1)));
208 av_assert0(dual_stride <= basis);
209 parity_revtab_generator(revtab, len, inv, 0, 0, 0, len, basis, dual_stride);
212 av_cold void av_tx_uninit(AVTXContext **ctx)
217 av_free((*ctx)->pfatab);
218 av_free((*ctx)->exptab);
219 av_free((*ctx)->revtab);
220 av_free((*ctx)->revtab_c);
221 av_free((*ctx)->inplace_idx);
222 av_free((*ctx)->tmp);
227 av_cold int av_tx_init(AVTXContext **ctx, av_tx_fn *tx, enum AVTXType type,
228 int inv, int len, const void *scale, uint64_t flags)
231 AVTXContext *s = av_mallocz(sizeof(*s));
233 return AVERROR(ENOMEM);
236 case AV_TX_FLOAT_FFT:
237 case AV_TX_FLOAT_MDCT:
238 if ((err = ff_tx_init_mdct_fft_float(s, tx, type, inv, len, scale, flags)))
241 ff_tx_init_float_x86(s, tx);
243 case AV_TX_DOUBLE_FFT:
244 case AV_TX_DOUBLE_MDCT:
245 if ((err = ff_tx_init_mdct_fft_double(s, tx, type, inv, len, scale, flags)))
248 case AV_TX_INT32_FFT:
249 case AV_TX_INT32_MDCT:
250 if ((err = ff_tx_init_mdct_fft_int32(s, tx, type, inv, len, scale, flags)))
254 err = AVERROR(EINVAL);