2 * Copyright (C) 2004 the ffmpeg project
4 * This library 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 of the License, or (at your option) any later version.
9 * This library 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 this library; if not, write to the Free Software
16 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
21 * Standard C DSP-oriented functions cribbed from the original VP3
30 #define IdctAdjustBeforeShift 8
39 void vp3_dsp_init_c(void)
44 void vp3_idct_c(int16_t *input_data, int16_t *dequant_matrix,
45 int coeff_count, int16_t *output_data)
47 int32_t dequantized_data[64];
48 int32_t *ip = dequantized_data;
49 int16_t *op = output_data;
51 int32_t A_, B_, C_, D_, _Ad, _Bd, _Cd, _Dd, E_, F_, G_, H_;
52 int32_t _Ed, _Gd, _Add, _Bdd, _Fd, _Hd;
57 /* de-zigzag and dequantize */
58 for (i = 0; i < coeff_count; i++) {
59 j = dezigzag_index[i];
60 dequantized_data[j] = dequant_matrix[i] * input_data[i];
63 /* Inverse DCT on the rows now */
64 for (i = 0; i < 8; i++) {
65 /* Check for non-zero values */
66 if ( ip[0] | ip[1] | ip[2] | ip[3] | ip[4] | ip[5] | ip[6] | ip[7] ) {
67 t1 = (int32_t)(xC1S7 * ip[1]);
68 t2 = (int32_t)(xC7S1 * ip[7]);
73 t1 = (int32_t)(xC7S1 * ip[1]);
74 t2 = (int32_t)(xC1S7 * ip[7]);
79 t1 = (int32_t)(xC3S5 * ip[3]);
80 t2 = (int32_t)(xC5S3 * ip[5]);
85 t1 = (int32_t)(xC3S5 * ip[5]);
86 t2 = (int32_t)(xC5S3 * ip[3]);
92 t1 = (int32_t)(xC4S4 * (A_ - C_));
96 t1 = (int32_t)(xC4S4 * (B_ - D_));
104 t1 = (int32_t)(xC4S4 * (ip[0] + ip[4]));
108 t1 = (int32_t)(xC4S4 * (ip[0] - ip[4]));
112 t1 = (int32_t)(xC2S6 * ip[2]);
113 t2 = (int32_t)(xC6S2 * ip[6]);
118 t1 = (int32_t)(xC6S2 * ip[2]);
119 t2 = (int32_t)(xC2S6 * ip[6]);
134 /* Final sequence of operations over-write original inputs. */
135 ip[0] = (int16_t)((_Gd + _Cd ) >> 0);
136 ip[7] = (int16_t)((_Gd - _Cd ) >> 0);
138 ip[1] = (int16_t)((_Add + _Hd ) >> 0);
139 ip[2] = (int16_t)((_Add - _Hd ) >> 0);
141 ip[3] = (int16_t)((_Ed + _Dd ) >> 0);
142 ip[4] = (int16_t)((_Ed - _Dd ) >> 0);
144 ip[5] = (int16_t)((_Fd + _Bdd ) >> 0);
145 ip[6] = (int16_t)((_Fd - _Bdd ) >> 0);
149 ip += 8; /* next row */
152 ip = dequantized_data;
154 for ( i = 0; i < 8; i++) {
155 /* Check for non-zero values (bitwise or faster than ||) */
156 if ( ip[0 * 8] | ip[1 * 8] | ip[2 * 8] | ip[3 * 8] |
157 ip[4 * 8] | ip[5 * 8] | ip[6 * 8] | ip[7 * 8] ) {
159 t1 = (int32_t)(xC1S7 * ip[1*8]);
160 t2 = (int32_t)(xC7S1 * ip[7*8]);
165 t1 = (int32_t)(xC7S1 * ip[1*8]);
166 t2 = (int32_t)(xC1S7 * ip[7*8]);
171 t1 = (int32_t)(xC3S5 * ip[3*8]);
172 t2 = (int32_t)(xC5S3 * ip[5*8]);
177 t1 = (int32_t)(xC3S5 * ip[5*8]);
178 t2 = (int32_t)(xC5S3 * ip[3*8]);
184 t1 = (int32_t)(xC4S4 * (A_ - C_));
188 t1 = (int32_t)(xC4S4 * (B_ - D_));
196 t1 = (int32_t)(xC4S4 * (ip[0*8] + ip[4*8]));
200 t1 = (int32_t)(xC4S4 * (ip[0*8] - ip[4*8]));
204 t1 = (int32_t)(xC2S6 * ip[2*8]);
205 t2 = (int32_t)(xC6S2 * ip[6*8]);
210 t1 = (int32_t)(xC6S2 * ip[2*8]);
211 t2 = (int32_t)(xC2S6 * ip[6*8]);
226 _Gd += IdctAdjustBeforeShift;
227 _Add += IdctAdjustBeforeShift;
228 _Ed += IdctAdjustBeforeShift;
229 _Fd += IdctAdjustBeforeShift;
231 /* Final sequence of operations over-write original inputs. */
232 op[0*8] = (int16_t)((_Gd + _Cd ) >> 4);
233 op[7*8] = (int16_t)((_Gd - _Cd ) >> 4);
235 op[1*8] = (int16_t)((_Add + _Hd ) >> 4);
236 op[2*8] = (int16_t)((_Add - _Hd ) >> 4);
238 op[3*8] = (int16_t)((_Ed + _Dd ) >> 4);
239 op[4*8] = (int16_t)((_Ed - _Dd ) >> 4);
241 op[5*8] = (int16_t)((_Fd + _Bdd ) >> 4);
242 op[6*8] = (int16_t)((_Fd - _Bdd ) >> 4);
256 ip++; /* next column */