FFmpeg
fixed_dsp.c
Go to the documentation of this file.
1 /*
2  * Copyright (c) 2012
3  * MIPS Technologies, Inc., California.
4  *
5  * Redistribution and use in source and binary forms, with or without
6  * modification, are permitted provided that the following conditions
7  * are met:
8  * 1. Redistributions of source code must retain the above copyright
9  * notice, this list of conditions and the following disclaimer.
10  * 2. Redistributions in binary form must reproduce the above copyright
11  * notice, this list of conditions and the following disclaimer in the
12  * documentation and/or other materials provided with the distribution.
13  * 3. Neither the name of the MIPS Technologies, Inc., nor the names of its
14  * contributors may be used to endorse or promote products derived from
15  * this software without specific prior written permission.
16  *
17  * THIS SOFTWARE IS PROVIDED BY THE MIPS TECHNOLOGIES, INC. ``AS IS'' AND
18  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20  * ARE DISCLAIMED. IN NO EVENT SHALL THE MIPS TECHNOLOGIES, INC. BE LIABLE
21  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27  * SUCH DAMAGE.
28  *
29  * Author: Nedeljko Babic (nedeljko.babic imgtec com)
30  *
31  * This file is part of FFmpeg.
32  *
33  * FFmpeg is free software; you can redistribute it and/or
34  * modify it under the terms of the GNU Lesser General Public
35  * License as published by the Free Software Foundation; either
36  * version 2.1 of the License, or (at your option) any later version.
37  *
38  * FFmpeg is distributed in the hope that it will be useful,
39  * but WITHOUT ANY WARRANTY; without even the implied warranty of
40  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
41  * Lesser General Public License for more details.
42  *
43  * You should have received a copy of the GNU Lesser General Public
44  * License along with FFmpeg; if not, write to the Free Software
45  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
46  */
47 
48 #include "common.h"
49 #include "fixed_dsp.h"
50 
51 static void vector_fmul_add_c(int *dst, const int *src0, const int *src1, const int *src2, int len){
52  int i;
53  int64_t accu;
54 
55  for (i=0; i<len; i++) {
56  accu = (int64_t)src0[i] * src1[i];
57  dst[i] = src2[i] + (int)((accu + 0x40000000) >> 31);
58  }
59 }
60 
61 static void vector_fmul_reverse_c(int *dst, const int *src0, const int *src1, int len)
62 {
63  int i;
64  int64_t accu;
65 
66  src1 += len-1;
67  for (i=0; i<len; i++) {
68  accu = (int64_t)src0[i] * src1[-i];
69  dst[i] = (int)((accu+0x40000000) >> 31);
70  }
71 }
72 
73 static void vector_fmul_window_scaled_c(int16_t *dst, const int32_t *src0,
74  const int32_t *src1, const int32_t *win,
75  int len, uint8_t bits)
76 {
77  int32_t s0, s1, wi, wj, i,j, round;
78 
79  dst += len;
80  win += len;
81  src0+= len;
82  round = bits? 1 << (bits-1) : 0;
83 
84  for (i=-len, j=len-1; i<0; i++, j--) {
85  s0 = src0[i];
86  s1 = src1[j];
87  wi = win[i];
88  wj = win[j];
89  dst[i] = av_clip_int16(((((int64_t)s0*wj - (int64_t)s1*wi + 0x40000000) >> 31) + round) >> bits);
90  dst[j] = av_clip_int16(((((int64_t)s0*wi + (int64_t)s1*wj + 0x40000000) >> 31) + round) >> bits);
91  }
92 }
93 
94 static void vector_fmul_window_c(int32_t *dst, const int32_t *src0,
95  const int32_t *src1, const int32_t *win,
96  int len)
97 {
98  int32_t s0, s1, wi, wj, i, j;
99 
100  dst += len;
101  win += len;
102  src0+= len;
103 
104  for (i=-len, j=len-1; i<0; i++, j--) {
105  s0 = src0[i];
106  s1 = src1[j];
107  wi = win[i];
108  wj = win[j];
109  dst[i] = ((int64_t)s0*wj - (int64_t)s1*wi + 0x40000000) >> 31;
110  dst[j] = ((int64_t)s0*wi + (int64_t)s1*wj + 0x40000000) >> 31;
111  }
112 }
113 
114 static void vector_fmul_c(int *dst, const int *src0, const int *src1, int len)
115 {
116  int i;
117  int64_t accu;
118 
119  for (i = 0; i < len; i++){
120  accu = (int64_t)src0[i] * src1[i];
121  dst[i] = (int)((accu+0x40000000) >> 31);
122  }
123 }
124 
125 static int scalarproduct_fixed_c(const int *v1, const int *v2, int len)
126 {
127  /** p is initialized with 0x40000000 so that the proper rounding will occur
128  * at the end */
129  int64_t p = 0x40000000;
130  int i;
131 
132  for (i = 0; i < len; i++)
133  p += (int64_t)v1[i] * v2[i];
134 
135  return (int)(p >> 31);
136 }
137 
138 static void butterflies_fixed_c(int *av_restrict v1s, int *av_restrict v2, int len)
139 {
140  int i;
141  unsigned int *v1 = v1s;
142 
143  for (i = 0; i < len; i++){
144  int t = v1[i] - v2[i];
145  v1[i] += v2[i];
146  v2[i] = t;
147  }
148 }
149 
151 {
153 
154  if (!fdsp)
155  return NULL;
156 
159  fdsp->vector_fmul = vector_fmul_c;
164 
165 #if ARCH_RISCV
167 #elif ARCH_X86
168  ff_fixed_dsp_init_x86(fdsp);
169 #endif
170 
171  return fdsp;
172 }
src1
const pixel * src1
Definition: h264pred_template.c:421
vector_fmul_c
static void vector_fmul_c(int *dst, const int *src0, const int *src1, int len)
Definition: fixed_dsp.c:114
AVFixedDSPContext
Definition: fixed_dsp.h:56
av_malloc
#define av_malloc(s)
Definition: tableprint_vlc.h:30
win
static float win(SuperEqualizerContext *s, float n, int N)
Definition: af_superequalizer.c:119
AVFixedDSPContext::vector_fmul_add
void(* vector_fmul_add)(int *dst, const int *src0, const int *src1, const int *src2, int len)
Calculate the entry wise product of two vectors of integers, add a third vector of integers and store...
Definition: fixed_dsp.h:132
ff_fixed_dsp_init_riscv
void ff_fixed_dsp_init_riscv(AVFixedDSPContext *fdsp)
Definition: fixed_dsp_init.c:30
avpriv_alloc_fixed_dsp
AVFixedDSPContext * avpriv_alloc_fixed_dsp(int bit_exact)
Allocate and initialize a fixed DSP context.
Definition: fixed_dsp.c:150
s1
#define s1
Definition: regdef.h:38
bits
uint8_t bits
Definition: vp3data.h:128
AVFixedDSPContext::scalarproduct_fixed
int(* scalarproduct_fixed)(const int *v1, const int *v2, int len)
Calculate the scalar product of two vectors of integers.
Definition: fixed_dsp.h:144
fixed_dsp.h
av_clip_int16
#define av_clip_int16
Definition: common.h:110
NULL
#define NULL
Definition: coverity.c:32
AVFixedDSPContext::vector_fmul_reverse
void(* vector_fmul_reverse)(int *dst, const int *src0, const int *src1, int len)
Definition: fixed_dsp.h:116
vector_fmul_reverse_c
static void vector_fmul_reverse_c(int *dst, const int *src0, const int *src1, int len)
Definition: fixed_dsp.c:61
vector_fmul_window_scaled_c
static void vector_fmul_window_scaled_c(int16_t *dst, const int32_t *src0, const int32_t *src1, const int32_t *win, int len, uint8_t bits)
Definition: fixed_dsp.c:73
AVFixedDSPContext::vector_fmul
void(* vector_fmul)(int *dst, const int *src0, const int *src1, int len)
Fixed-point multiplication that calculates the entry wise product of two vectors of integers and stor...
Definition: fixed_dsp.h:113
AVFixedDSPContext::butterflies_fixed
void(* butterflies_fixed)(int *av_restrict v1, int *av_restrict v2, int len)
Calculate the sum and difference of two vectors of integers.
Definition: fixed_dsp.h:153
i
#define i(width, name, range_min, range_max)
Definition: cbs_h2645.c:269
round
static av_always_inline av_const double round(double x)
Definition: libm.h:444
src2
const pixel * src2
Definition: h264pred_template.c:422
common.h
len
int len
Definition: vorbis_enc_data.h:426
butterflies_fixed_c
static void butterflies_fixed_c(int *av_restrict v1s, int *av_restrict v2, int len)
Definition: fixed_dsp.c:138
vector_fmul_window_c
static void vector_fmul_window_c(int32_t *dst, const int32_t *src0, const int32_t *src1, const int32_t *win, int len)
Definition: fixed_dsp.c:94
ff_fixed_dsp_init_x86
void ff_fixed_dsp_init_x86(AVFixedDSPContext *fdsp)
Definition: fixed_dsp_init.c:28
src0
const pixel *const src0
Definition: h264pred_template.c:420
s0
#define s0
Definition: regdef.h:37
int32_t
int32_t
Definition: audioconvert.c:56
AVFixedDSPContext::vector_fmul_window_scaled
void(* vector_fmul_window_scaled)(int16_t *dst, const int32_t *src0, const int32_t *src1, const int32_t *win, int len, uint8_t bits)
Overlap/add with window function.
Definition: fixed_dsp.h:80
scalarproduct_fixed_c
static int scalarproduct_fixed_c(const int *v1, const int *v2, int len)
Definition: fixed_dsp.c:125
int
int
Definition: ffmpeg_filter.c:156
AVFixedDSPContext::vector_fmul_window
void(* vector_fmul_window)(int32_t *dst, const int32_t *src0, const int32_t *src1, const int32_t *win, int len)
Overlap/add with window function.
Definition: fixed_dsp.h:98
vector_fmul_add_c
static void vector_fmul_add_c(int *dst, const int *src0, const int *src1, const int *src2, int len)
Definition: fixed_dsp.c:51