hpeldsp_template.c
Go to the documentation of this file.
1 /*
2  * Half-pel DSP functions.
3  * Copyright (c) 2000, 2001 Fabrice Bellard
4  * Copyright (c) 2002-2004 Michael Niedermayer <michaelni@gmx.at>
5  *
6  * gmc & q-pel & 32/64 bit based MC by Michael Niedermayer <michaelni@gmx.at>
7  *
8  * This file is part of FFmpeg.
9  *
10  * FFmpeg is free software; you can redistribute it and/or
11  * modify it under the terms of the GNU Lesser General Public
12  * License as published by the Free Software Foundation; either
13  * version 2.1 of the License, or (at your option) any later version.
14  *
15  * FFmpeg is distributed in the hope that it will be useful,
16  * but WITHOUT ANY WARRANTY; without even the implied warranty of
17  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
18  * Lesser General Public License for more details.
19  *
20  * You should have received a copy of the GNU Lesser General Public
21  * License along with FFmpeg; if not, write to the Free Software
22  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
23  */
24 
25 /**
26  * @file
27  * Half-pel DSP functions.
28  */
29 
30 #include "bit_depth_template.c"
31 
32 #include "hpel_template.c"
33 
34 #define PIXOP2(OPNAME, OP) \
35 static inline void FUNC(OPNAME ## _no_rnd_pixels8_l2)(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int dst_stride, \
36  int src_stride1, int src_stride2, int h){\
37  int i;\
38  for(i=0; i<h; i++){\
39  pixel4 a,b;\
40  a= AV_RN4P(&src1[i*src_stride1 ]);\
41  b= AV_RN4P(&src2[i*src_stride2 ]);\
42  OP(*((pixel4*)&dst[i*dst_stride ]), no_rnd_avg_pixel4(a, b));\
43  a= AV_RN4P(&src1[i*src_stride1+4*sizeof(pixel)]);\
44  b= AV_RN4P(&src2[i*src_stride2+4*sizeof(pixel)]);\
45  OP(*((pixel4*)&dst[i*dst_stride+4*sizeof(pixel)]), no_rnd_avg_pixel4(a, b));\
46  }\
47 }\
48 \
49 static inline void FUNCC(OPNAME ## _no_rnd_pixels8_x2)(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h){\
50  FUNC(OPNAME ## _no_rnd_pixels8_l2)(block, pixels, pixels+sizeof(pixel), line_size, line_size, line_size, h);\
51 }\
52 \
53 static inline void FUNCC(OPNAME ## _pixels8_x2)(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h){\
54  FUNC(OPNAME ## _pixels8_l2)(block, pixels, pixels+sizeof(pixel), line_size, line_size, line_size, h);\
55 }\
56 \
57 static inline void FUNCC(OPNAME ## _no_rnd_pixels8_y2)(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h){\
58  FUNC(OPNAME ## _no_rnd_pixels8_l2)(block, pixels, pixels+line_size, line_size, line_size, line_size, h);\
59 }\
60 \
61 static inline void FUNCC(OPNAME ## _pixels8_y2)(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h){\
62  FUNC(OPNAME ## _pixels8_l2)(block, pixels, pixels+line_size, line_size, line_size, line_size, h);\
63 }\
64 \
65 static inline void FUNCC(OPNAME ## _pixels4_x2)(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h){\
66  FUNC(OPNAME ## _pixels4_l2)(block, pixels, pixels+sizeof(pixel), line_size, line_size, line_size, h);\
67 }\
68 \
69 static inline void FUNCC(OPNAME ## _pixels4_y2)(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h){\
70  FUNC(OPNAME ## _pixels4_l2)(block, pixels, pixels+line_size, line_size, line_size, line_size, h);\
71 }\
72 \
73 static inline void FUNCC(OPNAME ## _pixels2_x2)(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h){\
74  FUNC(OPNAME ## _pixels2_l2)(block, pixels, pixels+sizeof(pixel), line_size, line_size, line_size, h);\
75 }\
76 \
77 static inline void FUNCC(OPNAME ## _pixels2_y2)(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h){\
78  FUNC(OPNAME ## _pixels2_l2)(block, pixels, pixels+line_size, line_size, line_size, line_size, h);\
79 }\
80 \
81 static inline void FUNCC(OPNAME ## _pixels2_xy2)(uint8_t *_block, const uint8_t *_pixels, ptrdiff_t line_size, int h)\
82 {\
83  int i, a0, b0, a1, b1;\
84  pixel *block = (pixel*)_block;\
85  const pixel *pixels = (const pixel*)_pixels;\
86  line_size >>= sizeof(pixel)-1;\
87  a0= pixels[0];\
88  b0= pixels[1] + 2;\
89  a0 += b0;\
90  b0 += pixels[2];\
91 \
92  pixels+=line_size;\
93  for(i=0; i<h; i+=2){\
94  a1= pixels[0];\
95  b1= pixels[1];\
96  a1 += b1;\
97  b1 += pixels[2];\
98 \
99  block[0]= (a1+a0)>>2; /* FIXME non put */\
100  block[1]= (b1+b0)>>2;\
101 \
102  pixels+=line_size;\
103  block +=line_size;\
104 \
105  a0= pixels[0];\
106  b0= pixels[1] + 2;\
107  a0 += b0;\
108  b0 += pixels[2];\
109 \
110  block[0]= (a1+a0)>>2;\
111  block[1]= (b1+b0)>>2;\
112  pixels+=line_size;\
113  block +=line_size;\
114  }\
115 }\
116 \
117 static inline void FUNCC(OPNAME ## _pixels4_xy2)(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)\
118 {\
119  /* FIXME HIGH BIT DEPTH */\
120  int i;\
121  const uint32_t a= AV_RN32(pixels );\
122  const uint32_t b= AV_RN32(pixels+1);\
123  uint32_t l0= (a&0x03030303UL)\
124  + (b&0x03030303UL)\
125  + 0x02020202UL;\
126  uint32_t h0= ((a&0xFCFCFCFCUL)>>2)\
127  + ((b&0xFCFCFCFCUL)>>2);\
128  uint32_t l1,h1;\
129 \
130  pixels+=line_size;\
131  for(i=0; i<h; i+=2){\
132  uint32_t a= AV_RN32(pixels );\
133  uint32_t b= AV_RN32(pixels+1);\
134  l1= (a&0x03030303UL)\
135  + (b&0x03030303UL);\
136  h1= ((a&0xFCFCFCFCUL)>>2)\
137  + ((b&0xFCFCFCFCUL)>>2);\
138  OP(*((uint32_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\
139  pixels+=line_size;\
140  block +=line_size;\
141  a= AV_RN32(pixels );\
142  b= AV_RN32(pixels+1);\
143  l0= (a&0x03030303UL)\
144  + (b&0x03030303UL)\
145  + 0x02020202UL;\
146  h0= ((a&0xFCFCFCFCUL)>>2)\
147  + ((b&0xFCFCFCFCUL)>>2);\
148  OP(*((uint32_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\
149  pixels+=line_size;\
150  block +=line_size;\
151  }\
152 }\
153 \
154 static inline void FUNCC(OPNAME ## _pixels8_xy2)(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)\
155 {\
156  /* FIXME HIGH BIT DEPTH */\
157  int j;\
158  for(j=0; j<2; j++){\
159  int i;\
160  const uint32_t a= AV_RN32(pixels );\
161  const uint32_t b= AV_RN32(pixels+1);\
162  uint32_t l0= (a&0x03030303UL)\
163  + (b&0x03030303UL)\
164  + 0x02020202UL;\
165  uint32_t h0= ((a&0xFCFCFCFCUL)>>2)\
166  + ((b&0xFCFCFCFCUL)>>2);\
167  uint32_t l1,h1;\
168 \
169  pixels+=line_size;\
170  for(i=0; i<h; i+=2){\
171  uint32_t a= AV_RN32(pixels );\
172  uint32_t b= AV_RN32(pixels+1);\
173  l1= (a&0x03030303UL)\
174  + (b&0x03030303UL);\
175  h1= ((a&0xFCFCFCFCUL)>>2)\
176  + ((b&0xFCFCFCFCUL)>>2);\
177  OP(*((uint32_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\
178  pixels+=line_size;\
179  block +=line_size;\
180  a= AV_RN32(pixels );\
181  b= AV_RN32(pixels+1);\
182  l0= (a&0x03030303UL)\
183  + (b&0x03030303UL)\
184  + 0x02020202UL;\
185  h0= ((a&0xFCFCFCFCUL)>>2)\
186  + ((b&0xFCFCFCFCUL)>>2);\
187  OP(*((uint32_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\
188  pixels+=line_size;\
189  block +=line_size;\
190  }\
191  pixels+=4-line_size*(h+1);\
192  block +=4-line_size*h;\
193  }\
194 }\
195 \
196 static inline void FUNCC(OPNAME ## _no_rnd_pixels8_xy2)(uint8_t *block, const uint8_t *pixels, ptrdiff_t line_size, int h)\
197 {\
198  /* FIXME HIGH BIT DEPTH */\
199  int j;\
200  for(j=0; j<2; j++){\
201  int i;\
202  const uint32_t a= AV_RN32(pixels );\
203  const uint32_t b= AV_RN32(pixels+1);\
204  uint32_t l0= (a&0x03030303UL)\
205  + (b&0x03030303UL)\
206  + 0x01010101UL;\
207  uint32_t h0= ((a&0xFCFCFCFCUL)>>2)\
208  + ((b&0xFCFCFCFCUL)>>2);\
209  uint32_t l1,h1;\
210 \
211  pixels+=line_size;\
212  for(i=0; i<h; i+=2){\
213  uint32_t a= AV_RN32(pixels );\
214  uint32_t b= AV_RN32(pixels+1);\
215  l1= (a&0x03030303UL)\
216  + (b&0x03030303UL);\
217  h1= ((a&0xFCFCFCFCUL)>>2)\
218  + ((b&0xFCFCFCFCUL)>>2);\
219  OP(*((uint32_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\
220  pixels+=line_size;\
221  block +=line_size;\
222  a= AV_RN32(pixels );\
223  b= AV_RN32(pixels+1);\
224  l0= (a&0x03030303UL)\
225  + (b&0x03030303UL)\
226  + 0x01010101UL;\
227  h0= ((a&0xFCFCFCFCUL)>>2)\
228  + ((b&0xFCFCFCFCUL)>>2);\
229  OP(*((uint32_t*)block), h0+h1+(((l0+l1)>>2)&0x0F0F0F0FUL));\
230  pixels+=line_size;\
231  block +=line_size;\
232  }\
233  pixels+=4-line_size*(h+1);\
234  block +=4-line_size*h;\
235  }\
236 }\
237 \
238 CALL_2X_PIXELS(FUNCC(OPNAME ## _pixels16_x2) , FUNCC(OPNAME ## _pixels8_x2) , 8*sizeof(pixel))\
239 CALL_2X_PIXELS(FUNCC(OPNAME ## _pixels16_y2) , FUNCC(OPNAME ## _pixels8_y2) , 8*sizeof(pixel))\
240 CALL_2X_PIXELS(FUNCC(OPNAME ## _pixels16_xy2), FUNCC(OPNAME ## _pixels8_xy2), 8*sizeof(pixel))\
241 av_unused CALL_2X_PIXELS(FUNCC(OPNAME ## _no_rnd_pixels16) , FUNCC(OPNAME ## _pixels8) , 8*sizeof(pixel))\
242 CALL_2X_PIXELS(FUNCC(OPNAME ## _no_rnd_pixels16_x2) , FUNCC(OPNAME ## _no_rnd_pixels8_x2) , 8*sizeof(pixel))\
243 CALL_2X_PIXELS(FUNCC(OPNAME ## _no_rnd_pixels16_y2) , FUNCC(OPNAME ## _no_rnd_pixels8_y2) , 8*sizeof(pixel))\
244 CALL_2X_PIXELS(FUNCC(OPNAME ## _no_rnd_pixels16_xy2), FUNCC(OPNAME ## _no_rnd_pixels8_xy2), 8*sizeof(pixel))\
245 
246 #define op_avg(a, b) a = rnd_avg_pixel4(a, b)
247 #define op_put(a, b) a = b
248 #if BIT_DEPTH == 8
249 #define put_no_rnd_pixels8_8_c put_pixels8_8_c
250 PIXOP2(avg, op_avg)
251 PIXOP2(put, op_put)
252 #endif
253 #undef op_avg
254 #undef op_put
#define PIXOP2(OPNAME, OP)
#define op_avg(a, b)
#define put(d, s)
Definition: dsputil_align.c:51
#define op_put(a, b)
#define avg(d, s)
Definition: dsputil_align.c:52