3 * Copyright (c) 2004-2012 Michael Niedermayer <michaelni@gmx.at>
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
25 * @author Michael Niedermayer <michaelni@gmx.at>
28 #if defined(TEMPLATE_RESAMPLE_DBL) \
29 || defined(TEMPLATE_RESAMPLE_DBL_SSE2)
31 # define FILTER_SHIFT 0
34 # define FELEM2 double
35 # define FELEML double
36 # define OUT(d, v) d = v
38 # if defined(TEMPLATE_RESAMPLE_DBL)
39 # define RENAME(N) N ## _double
40 # elif defined(TEMPLATE_RESAMPLE_DBL_SSE2)
41 # define COMMON_CORE COMMON_CORE_DBL_SSE2
42 # define LINEAR_CORE LINEAR_CORE_DBL_SSE2
43 # define RENAME(N) N ## _double_sse2
46 #elif defined(TEMPLATE_RESAMPLE_FLT) \
47 || defined(TEMPLATE_RESAMPLE_FLT_SSE) \
48 || defined(TEMPLATE_RESAMPLE_FLT_AVX)
50 # define FILTER_SHIFT 0
55 # define OUT(d, v) d = v
57 # if defined(TEMPLATE_RESAMPLE_FLT)
58 # define RENAME(N) N ## _float
59 # elif defined(TEMPLATE_RESAMPLE_FLT_SSE)
60 # define COMMON_CORE COMMON_CORE_FLT_SSE
61 # define LINEAR_CORE LINEAR_CORE_FLT_SSE
62 # define RENAME(N) N ## _float_sse
63 # elif defined(TEMPLATE_RESAMPLE_FLT_AVX)
64 # define COMMON_CORE COMMON_CORE_FLT_AVX
65 # define LINEAR_CORE LINEAR_CORE_FLT_AVX
66 # define RENAME(N) N ## _float_avx
69 #elif defined(TEMPLATE_RESAMPLE_S32)
70 # define RENAME(N) N ## _int32
71 # define FILTER_SHIFT 30
72 # define DELEM int32_t
73 # define FELEM int32_t
74 # define FELEM2 int64_t
75 # define FELEML int64_t
76 # define FELEM_MAX INT32_MAX
77 # define FELEM_MIN INT32_MIN
78 # define OUT(d, v) v = (v + (1<<(FILTER_SHIFT-1)))>>FILTER_SHIFT;\
79 d = (uint64_t)(v + 0x80000000) > 0xFFFFFFFF ? (v>>63) ^ 0x7FFFFFFF : v
81 #elif defined(TEMPLATE_RESAMPLE_S16) \
82 || defined(TEMPLATE_RESAMPLE_S16_MMX2) \
83 || defined(TEMPLATE_RESAMPLE_S16_SSE2)
85 # define FILTER_SHIFT 15
86 # define DELEM int16_t
87 # define FELEM int16_t
88 # define FELEM2 int32_t
89 # define FELEML int64_t
90 # define FELEM_MAX INT16_MAX
91 # define FELEM_MIN INT16_MIN
92 # define OUT(d, v) v = (v + (1<<(FILTER_SHIFT-1)))>>FILTER_SHIFT;\
93 d = (unsigned)(v + 32768) > 65535 ? (v>>31) ^ 32767 : v
95 # if defined(TEMPLATE_RESAMPLE_S16)
96 # define RENAME(N) N ## _int16
97 # elif defined(TEMPLATE_RESAMPLE_S16_MMX2)
98 # define COMMON_CORE COMMON_CORE_INT16_MMX2
99 # define LINEAR_CORE LINEAR_CORE_INT16_MMX2
100 # define RENAME(N) N ## _int16_mmx2
101 # elif defined(TEMPLATE_RESAMPLE_S16_SSE2)
102 # define COMMON_CORE COMMON_CORE_INT16_SSE2
103 # define LINEAR_CORE LINEAR_CORE_INT16_SSE2
104 # define RENAME(N) N ## _int16_sse2
109 int RENAME(swri_resample)(ResampleContext *c, DELEM *dst, const DELEM *src, int *consumed, int src_size, int dst_size, int update_ctx){
113 int dst_incr_frac= c->dst_incr % c->src_incr;
114 int dst_incr= c->dst_incr / c->src_incr;
115 int compensation_distance= c->compensation_distance;
117 av_assert1(c->filter_shift == FILTER_SHIFT);
118 av_assert1(c->felem_size == sizeof(FELEM));
120 if(compensation_distance == 0 && c->filter_length == 1 && c->phase_shift==0){
121 int64_t index2= (1LL<<32)*c->frac/c->src_incr + (1LL<<32)*index;
122 int64_t incr= (1LL<<32) * c->dst_incr / c->src_incr;
123 int new_size = (src_size * (int64_t)c->src_incr - frac + c->dst_incr - 1) / c->dst_incr;
125 dst_size= FFMIN(dst_size, new_size);
127 for(dst_index=0; dst_index < dst_size; dst_index++){
128 dst[dst_index] = src[index2>>32];
131 index += dst_index * dst_incr;
132 index += (frac + dst_index * (int64_t)dst_incr_frac) / c->src_incr;
133 frac = (frac + dst_index * (int64_t)dst_incr_frac) % c->src_incr;
134 av_assert2(index >= 0);
137 } else if (compensation_distance == 0 && index >= 0) {
138 int64_t end_index = (1 + src_size - c->filter_length) << c->phase_shift;
139 int64_t delta_frac = (end_index - index) * c->src_incr - c->frac;
140 int delta_n = (delta_frac + c->dst_incr - 1) / c->dst_incr;
141 int n = FFMIN(dst_size, delta_n);
145 sample_index = index >> c->phase_shift;
146 index &= c->phase_mask;
147 for (dst_index = 0; dst_index < n; dst_index++) {
148 FELEM *filter = ((FELEM *) c->filter_bank) + c->filter_alloc * index;
154 for (i = 0; i < c->filter_length; i++) {
155 val += src[sample_index + i] * (FELEM2)filter[i];
157 OUT(dst[dst_index], val);
160 frac += dst_incr_frac;
162 if (frac >= c->src_incr) {
166 sample_index += index >> c->phase_shift;
167 index &= c->phase_mask;
170 sample_index = index >> c->phase_shift;
171 index &= c->phase_mask;
172 for (dst_index = 0; dst_index < n; dst_index++) {
173 FELEM *filter = ((FELEM *) c->filter_bank) + c->filter_alloc * index;
174 FELEM2 val=0, v2 = 0;
179 for (i = 0; i < c->filter_length; i++) {
180 val += src[sample_index + i] * (FELEM2)filter[i];
181 v2 += src[sample_index + i] * (FELEM2)filter[i + c->filter_alloc];
184 val += (v2 - val) * (FELEML) frac / c->src_incr;
185 OUT(dst[dst_index], val);
187 frac += dst_incr_frac;
189 if (frac >= c->src_incr) {
193 sample_index += index >> c->phase_shift;
194 index &= c->phase_mask;
198 *consumed = sample_index;
200 int sample_index = 0;
201 for(dst_index=0; dst_index < dst_size; dst_index++){
205 sample_index += index >> c->phase_shift;
206 index &= c->phase_mask;
207 filter = ((FELEM*)c->filter_bank) + c->filter_alloc*index;
209 if(sample_index + c->filter_length > src_size || -sample_index >= src_size){
211 }else if(sample_index < 0){
212 for(i=0; i<c->filter_length; i++)
213 val += src[FFABS(sample_index + i)] * (FELEM2)filter[i];
214 OUT(dst[dst_index], val);
220 for(i=0; i<c->filter_length; i++){
221 val += src[sample_index + i] * (FELEM2)filter[i];
222 v2 += src[sample_index + i] * (FELEM2)filter[i + c->filter_alloc];
225 val+=(v2-val)*(FELEML)frac / c->src_incr;
226 OUT(dst[dst_index], val);
231 for(i=0; i<c->filter_length; i++){
232 val += src[sample_index + i] * (FELEM2)filter[i];
234 OUT(dst[dst_index], val);
238 frac += dst_incr_frac;
240 if(frac >= c->src_incr){
245 if(dst_index + 1 == compensation_distance){
246 compensation_distance= 0;
247 dst_incr_frac= c->ideal_dst_incr % c->src_incr;
248 dst_incr= c->ideal_dst_incr / c->src_incr;
251 *consumed= FFMAX(sample_index, 0);
252 index += FFMIN(sample_index, 0) << c->phase_shift;
254 if(compensation_distance){
255 compensation_distance -= dst_index;
256 av_assert1(compensation_distance > 0);
263 c->dst_incr= dst_incr_frac + c->src_incr*dst_incr;
264 c->compensation_distance= compensation_distance;