FFmpeg
llviddsp.c
Go to the documentation of this file.
1 /*
2  * Copyright (c) 2016 Alexandra Hájková
3  *
4  * This file is part of FFmpeg.
5  *
6  * FFmpeg is free software; you can redistribute it and/or modify
7  * it under the terms of the GNU General Public License as published by
8  * the Free Software Foundation; either version 2 of the License, or
9  * (at your option) any later version.
10  *
11  * FFmpeg is distributed in the hope that it will be useful,
12  * but WITHOUT ANY WARRANTY; without even the implied warranty of
13  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
14  * GNU General Public License for more details.
15  *
16  * You should have received a copy of the GNU General Public License along
17  * with FFmpeg; if not, write to the Free Software Foundation, Inc.,
18  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
19  */
20 
21 #include <string.h>
22 
23 #include "libavutil/common.h"
24 #include "libavutil/intreadwrite.h"
25 #include "libavutil/mem.h"
26 
28 
29 #include "checkasm.h"
30 
31 #define randomize_buffers(buf, size) \
32  do { \
33  int j; \
34  uint8_t *tmp_buf = (uint8_t *)buf;\
35  for (j = 0; j < size; j++) \
36  tmp_buf[j] = rnd() & 0xFF; \
37  } while (0)
38 
39 #define init_buffer(a0, a1, type, width)\
40  if (!a0 || !a1)\
41  fail();\
42  randomize_buffers(a0, width * sizeof(type));\
43  memcpy(a1, a0, width*sizeof(type));\
44 
46 {
47  uint8_t *dst0 = av_mallocz(width);
48  uint8_t *dst1 = av_mallocz(width);
49  uint8_t *src0 = av_mallocz_array(width, sizeof(uint8_t));
50  uint8_t *src1 = av_mallocz_array(width, sizeof(uint8_t));
51  declare_func_emms(AV_CPU_FLAG_MMX, void, uint8_t *dst, uint8_t *src, ptrdiff_t w);
52 
53  init_buffer(src0, src1, uint8_t, width);
54 
55  if (!dst0 || !dst1)
56  fail();
57 
58 
59  if (check_func(c.add_bytes, "add_bytes")) {
60  call_ref(dst0, src0, width);
61  call_new(dst1, src1, width);
62  if (memcmp(dst0, dst1, width))
63  fail();
64  bench_new(dst1, src1, width);
65  }
66 
67  av_free(src0);
68  av_free(src1);
69  av_free(dst0);
70  av_free(dst1);
71 }
72 
74  int A0, A1, B0, B1;
75  uint8_t *dst0 = av_mallocz(width);
76  uint8_t *dst1 = av_mallocz(width);
77  uint8_t *src0 = av_mallocz_array(width, sizeof(uint8_t));
78  uint8_t *src1 = av_mallocz_array(width, sizeof(uint8_t));
79  uint8_t *diff0 = av_mallocz_array(width, sizeof(uint8_t));
80  uint8_t *diff1 = av_mallocz_array(width, sizeof(uint8_t));
81  declare_func_emms(AV_CPU_FLAG_MMX, void, uint8_t *dst, const uint8_t *src1,
82  const uint8_t *diff, ptrdiff_t w,
83  int *left, int *left_top);
84 
85  init_buffer(src0, src1, uint8_t, width);
86  init_buffer(diff0, diff1, uint8_t, width);
87 
88  A0 = rnd() & 0xFF;
89  B0 = rnd() & 0xFF;
90  A1 = A0;
91  B1 = B0;
92 
93 
94  if (check_func(c.add_median_pred, "add_median_pred")) {
95  call_ref(dst0, src0, diff0, width, &A0, &B0);
96  call_new(dst1, src1, diff1, width, &A1, &B1);
97  if (memcmp(dst0, dst1, width) || (A0 != A1) || (B0 != B1))
98  fail();
99  bench_new(dst1, src1, diff1, width, &A1, &B1);
100  }
101 
102  av_free(src0);
103  av_free(src1);
104  av_free(diff0);
105  av_free(diff1);
106  av_free(dst0);
107  av_free(dst1);
108 }
109 
110 static void check_add_left_pred(LLVidDSPContext c, int width, int acc, const char * report)
111 {
112  int res0, res1;
113  uint8_t *dst0 = av_mallocz(width);
114  uint8_t *dst1 = av_mallocz(width);
115  uint8_t *src0 = av_mallocz_array(width, sizeof(uint8_t));
116  uint8_t *src1 = av_mallocz_array(width, sizeof(uint8_t));
117  declare_func_emms(AV_CPU_FLAG_MMX, int, uint8_t *dst, uint8_t *src, ptrdiff_t w, int acc);
118 
119  init_buffer(src0, src1, uint8_t, width);
120 
121  if (!dst0 || !dst1)
122  fail();
123 
124  if (check_func(c.add_left_pred, "%s", report)) {
125  res0 = call_ref(dst0, src0, width, acc);
126  res1 = call_new(dst1, src1, width, acc);
127  if ((res0 & 0xFF) != (res1 & 0xFF)||\
128  memcmp(dst0, dst1, width))
129  fail();
130  bench_new(dst1, src1, width, acc);
131  }
132 
133  av_free(src0);
134  av_free(src1);
135  av_free(dst0);
136  av_free(dst1);
137 }
138 
139 static void check_add_left_pred_16(LLVidDSPContext c, unsigned mask, int width, unsigned acc, const char * report)
140 {
141  int res0, res1;
142  uint16_t *dst0 = av_mallocz_array(width, sizeof(uint16_t));
143  uint16_t *dst1 = av_mallocz_array(width, sizeof(uint16_t));
144  uint16_t *src0 = av_mallocz_array(width, sizeof(uint16_t));
145  uint16_t *src1 = av_mallocz_array(width, sizeof(uint16_t));
146  declare_func_emms(AV_CPU_FLAG_MMX, int, uint16_t *dst, uint16_t *src, unsigned mask, ptrdiff_t w, unsigned acc);
147 
148  init_buffer(src0, src1, uint16_t, width);
149 
150  if (!dst0 || !dst1)
151  fail();
152 
153  if (check_func(c.add_left_pred_int16, "%s", report)) {
154  res0 = call_ref(dst0, src0, mask, width, acc);
155  res1 = call_new(dst1, src1, mask, width, acc);
156  if ((res0 &0xFFFF) != (res1 &0xFFFF)||\
157  memcmp(dst0, dst1, width))
158  fail();
159  bench_new(dst1, src1, mask, width, acc);
160  }
161 
162  av_free(src0);
163  av_free(src1);
164  av_free(dst0);
165  av_free(dst1);
166 }
167 
169  int src_size, stride;
170  uint8_t *src0, *src1;
171  declare_func_emms(AV_CPU_FLAG_MMX, void, uint8_t *src, const ptrdiff_t stride,
172  const ptrdiff_t width);
173 
174  stride = w + 32;
175  src_size = (stride + 32) * 2; /* dsp need previous line, and ignore the start of the line */
176  src0 = av_mallocz(src_size);
177  src1 = av_mallocz(src_size);
178 
179  init_buffer(src0, src1, uint8_t, src_size);
180 
181  if (check_func(c.add_gradient_pred, "add_gradient_pred")) {
182  call_ref(src0 + stride + 32, stride, w);
183  call_new(src1 + stride + 32, stride, w);
184  if (memcmp(src0, src1, stride)||/* previous line doesn't change */
185  memcmp(src0+stride, src1 + stride, w + 32)) {
186  fail();
187  }
188  bench_new(src1 + stride + 32, stride, w);
189  }
190 
191  av_free(src0);
192  av_free(src1);
193 }
194 
196 {
198  int width = 16 * av_clip(rnd(), 16, 128);
199  int accRnd = rnd() & 0xFF;
200 
201  ff_llviddsp_init(&c);
202 
203  check_add_bytes(c, width);
204  report("add_bytes");
205 
206  check_add_median_pred(c, width);
207  report("add_median_pred");
208 
209  check_add_left_pred(c, width, 0, "add_left_pred_zero");
210  report("add_left_pred_zero");
211 
212  check_add_left_pred(c, width, accRnd, "add_left_pred_rnd_acc");
213  report("add_left_pred_rnd_acc");
214 
215  check_add_left_pred_16(c, 255, width, accRnd, "add_left_pred_int16");
216  report("add_left_pred_int16");
217 
218  check_add_gradient_pred(c, width);
219  report("add_gradient_pred");
220 }
int(* add_left_pred)(uint8_t *dst, const uint8_t *src, ptrdiff_t w, int left)
#define A1
Definition: binkdsp.c:31
Memory handling functions.
int acc
Definition: yuv2rgb.c:554
void * av_mallocz(size_t size)
Allocate a memory block with alignment suitable for all memory accesses (including vectors if availab...
Definition: mem.c:236
#define src
Definition: vp8dsp.c:254
#define report
Definition: checkasm.h:124
#define B1
Definition: faandct.c:41
void checkasm_check_llviddsp(void)
Definition: llviddsp.c:195
uint8_t
Undefined Behavior In the C some operations are like signed integer dereferencing freed accessing outside allocated Undefined Behavior must not occur in a C it is not safe even if the output of undefined operations is unused The unsafety may seem nit picking but Optimizing compilers have in fact optimized code on the assumption that no undefined Behavior occurs Optimizing code based on wrong assumptions can and has in some cases lead to effects beyond the output of computations The signed integer overflow problem in speed critical code Code which is highly optimized and works with signed integers sometimes has the problem that often the output of the computation does not c
Definition: undefined.txt:32
void(* add_median_pred)(uint8_t *dst, const uint8_t *top, const uint8_t *diff, ptrdiff_t w, int *left, int *left_top)
void(* add_gradient_pred)(uint8_t *src, const ptrdiff_t stride, const ptrdiff_t width)
static const uint16_t mask[17]
Definition: lzw.c:38
#define fail()
Definition: checkasm.h:121
#define width
uint8_t w
Definition: llviddspenc.c:38
#define declare_func_emms(cpu_flags, ret,...)
Definition: checkasm.h:118
#define call_ref(...)
Definition: checkasm.h:127
int(* add_left_pred_int16)(uint16_t *dst, const uint16_t *src, unsigned mask, ptrdiff_t w, unsigned left)
#define src1
Definition: h264pred.c:139
static void check_add_bytes(LLVidDSPContext c, int width)
Definition: llviddsp.c:45
#define AV_CPU_FLAG_MMX
standard MMX
Definition: cpu.h:31
static void check_add_median_pred(LLVidDSPContext c, int width)
Definition: llviddsp.c:73
Tag MUST be and< 10hcoeff half pel interpolation filter coefficients, hcoeff[0] are the 2 middle coefficients[1] are the next outer ones and so on, resulting in a filter like:...eff[2], hcoeff[1], hcoeff[0], hcoeff[0], hcoeff[1], hcoeff[2]...the sign of the coefficients is not explicitly stored but alternates after each coeff and coeff[0] is positive, so...,+,-,+,-,+,+,-,+,-,+,...hcoeff[0] is not explicitly stored but found by subtracting the sum of all stored coefficients with signs from 32 hcoeff[0]=32-hcoeff[1]-hcoeff[2]-...a good choice for hcoeff and htaps is htaps=6 hcoeff={40,-10, 2}an alternative which requires more computations at both encoder and decoder side and may or may not be better is htaps=8 hcoeff={42,-14, 6,-2}ref_frames minimum of the number of available reference frames and max_ref_frames for example the first frame after a key frame always has ref_frames=1spatial_decomposition_type wavelet type 0 is a 9/7 symmetric compact integer wavelet 1 is a 5/3 symmetric compact integer wavelet others are reserved stored as delta from last, last is reset to 0 if always_reset||keyframeqlog quality(logarithmic quantizer scale) stored as delta from last, last is reset to 0 if always_reset||keyframemv_scale stored as delta from last, last is reset to 0 if always_reset||keyframe FIXME check that everything works fine if this changes between framesqbias dequantization bias stored as delta from last, last is reset to 0 if always_reset||keyframeblock_max_depth maximum depth of the block tree stored as delta from last, last is reset to 0 if always_reset||keyframequant_table quantization tableHighlevel bitstream structure:==============================--------------------------------------------|Header|--------------------------------------------|------------------------------------|||Block0||||split?||||yes no||||.........intra?||||:Block01:yes no||||:Block02:.................||||:Block03::y DC::ref index:||||:Block04::cb DC::motion x:||||.........:cr DC::motion y:||||.................|||------------------------------------||------------------------------------|||Block1|||...|--------------------------------------------|------------------------------------|||Y subbands||Cb subbands||Cr subbands||||------||------||------|||||LL0||HL0||||LL0||HL0||||LL0||HL0|||||------||------||------||||------||------||------|||||LH0||HH0||||LH0||HH0||||LH0||HH0|||||------||------||------||||------||------||------|||||HL1||LH1||||HL1||LH1||||HL1||LH1|||||------||------||------||||------||------||------|||||HH1||HL2||||HH1||HL2||||HH1||HL2|||||...||...||...|||------------------------------------|--------------------------------------------Decoding process:=================------------|||Subbands|------------||||------------|Intra DC||||LL0 subband prediction------------|\Dequantization-------------------\||Reference frames|\IDWT|--------------|Motion\|||Frame 0||Frame 1||Compensation.OBMC v-------|--------------|--------------.\------> Frame n output Frame Frame<----------------------------------/|...|-------------------Range Coder:============Binary Range Coder:-------------------The implemented range coder is an adapted version based upon"Range encoding: an algorithm for removing redundancy from a digitised message."by G.N.N.Martin.The symbols encoded by the Snow range coder are bits(0|1).The associated probabilities are not fix but change depending on the symbol mix seen so far.bit seen|new state---------+-----------------------------------------------0|256-state_transition_table[256-old_state];1|state_transition_table[old_state];state_transition_table={0, 0, 0, 0, 0, 0, 0, 0, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 114, 115, 116, 117, 118, 119, 120, 121, 122, 123, 124, 125, 126, 127, 128, 129, 130, 131, 132, 133, 133, 134, 135, 136, 137, 138, 139, 140, 141, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 165, 166, 167, 168, 169, 170, 171, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 190, 191, 192, 194, 194, 195, 196, 197, 198, 199, 200, 201, 202, 202, 204, 205, 206, 207, 208, 209, 209, 210, 211, 212, 213, 215, 215, 216, 217, 218, 219, 220, 220, 222, 223, 224, 225, 226, 227, 227, 229, 229, 230, 231, 232, 234, 234, 235, 236, 237, 238, 239, 240, 241, 242, 243, 244, 245, 246, 247, 248, 248, 0, 0, 0, 0, 0, 0, 0};FIXME Range Coding of integers:-------------------------FIXME Neighboring Blocks:===================left and top are set to the respective blocks unless they are outside of the image in which case they are set to the Null block top-left is set to the top left block unless it is outside of the image in which case it is set to the left block if this block has no larger parent block or it is at the left side of its parent block and the top right block is not outside of the image then the top right block is used for top-right else the top-left block is used Null block y, cb, cr are 128 level, ref, mx and my are 0 Motion Vector Prediction:=========================1.the motion vectors of all the neighboring blocks are scaled to compensate for the difference of reference frames scaled_mv=(mv *(256 *(current_reference+1)/(mv.reference+1))+128)> the median of the scaled left
Definition: snow.txt:206
void ff_llviddsp_init(LLVidDSPContext *c)
#define check_func(func,...)
Definition: checkasm.h:112
#define src0
Definition: h264pred.c:138
common internal and external API header
#define rnd()
Definition: checkasm.h:105
static void check_add_left_pred_16(LLVidDSPContext c, unsigned mask, int width, unsigned acc, const char *report)
Definition: llviddsp.c:139
static av_always_inline int diff(const uint32_t a, const uint32_t b)
#define av_free(p)
#define bench_new(...)
Definition: checkasm.h:254
#define call_new(...)
Definition: checkasm.h:194
#define init_buffer(a0, a1, type, width)
Definition: llviddsp.c:39
#define stride
void(* add_bytes)(uint8_t *dst, uint8_t *src, ptrdiff_t w)
static void check_add_left_pred(LLVidDSPContext c, int width, int acc, const char *report)
Definition: llviddsp.c:110
#define B0
Definition: faandct.c:40
static void check_add_gradient_pred(LLVidDSPContext c, int w)
Definition: llviddsp.c:168
void * av_mallocz_array(size_t nmemb, size_t size)
Definition: mem.c:191