dsputil.h
Go to the documentation of this file.
1 /*
2  * DSP utils
3  * Copyright (c) 2000, 2001, 2002 Fabrice Bellard
4  * Copyright (c) 2002-2004 Michael Niedermayer <michaelni@gmx.at>
5  *
6  * This file is part of Libav.
7  *
8  * Libav is free software; you can redistribute it and/or
9  * modify it under the terms of the GNU Lesser General Public
10  * License as published by the Free Software Foundation; either
11  * version 2.1 of the License, or (at your option) any later version.
12  *
13  * Libav is distributed in the hope that it will be useful,
14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
16  * Lesser General Public License for more details.
17  *
18  * You should have received a copy of the GNU Lesser General Public
19  * License along with Libav; if not, write to the Free Software
20  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
21  */
22 
30 #ifndef AVCODEC_DSPUTIL_H
31 #define AVCODEC_DSPUTIL_H
32 
33 #include "libavutil/intreadwrite.h"
34 #include "avcodec.h"
35 
36 
37 //#define DEBUG
38 /* dct code */
39 typedef short DCTELEM;
40 
41 void fdct_ifast (DCTELEM *data);
42 void fdct_ifast248 (DCTELEM *data);
47 
48 void j_rev_dct (DCTELEM *data);
49 void j_rev_dct4 (DCTELEM *data);
50 void j_rev_dct2 (DCTELEM *data);
51 void j_rev_dct1 (DCTELEM *data);
53 
57 
58 #define H264_IDCT(depth) \
59 void ff_h264_idct8_add_ ## depth ## _c(uint8_t *dst, DCTELEM *block, int stride);\
60 void ff_h264_idct_add_ ## depth ## _c(uint8_t *dst, DCTELEM *block, int stride);\
61 void ff_h264_idct8_dc_add_ ## depth ## _c(uint8_t *dst, DCTELEM *block, int stride);\
62 void ff_h264_idct_dc_add_ ## depth ## _c(uint8_t *dst, DCTELEM *block, int stride);\
63 void ff_h264_idct_add16_ ## depth ## _c(uint8_t *dst, const int *blockoffset, DCTELEM *block, int stride, const uint8_t nnzc[6*8]);\
64 void ff_h264_idct_add16intra_ ## depth ## _c(uint8_t *dst, const int *blockoffset, DCTELEM *block, int stride, const uint8_t nnzc[6*8]);\
65 void ff_h264_idct8_add4_ ## depth ## _c(uint8_t *dst, const int *blockoffset, DCTELEM *block, int stride, const uint8_t nnzc[6*8]);\
66 void ff_h264_idct_add8_422_ ## depth ## _c(uint8_t **dest, const int *blockoffset, DCTELEM *block, int stride, const uint8_t nnzc[6*8]);\
67 void ff_h264_idct_add8_ ## depth ## _c(uint8_t **dest, const int *blockoffset, DCTELEM *block, int stride, const uint8_t nnzc[6*8]);\
68 void ff_h264_luma_dc_dequant_idct_ ## depth ## _c(DCTELEM *output, DCTELEM *input, int qmul);\
69 void ff_h264_chroma422_dc_dequant_idct_ ## depth ## _c(DCTELEM *block, int qmul);\
70 void ff_h264_chroma_dc_dequant_idct_ ## depth ## _c(DCTELEM *block, int qmul);
71 
72 H264_IDCT( 8)
73 H264_IDCT( 9)
74 H264_IDCT(10)
75 
76 void ff_svq3_luma_dc_dequant_idct_c(DCTELEM *output, DCTELEM *input, int qp);
77 void ff_svq3_add_idct_c(uint8_t *dst, DCTELEM *block, int stride, int qp, int dc);
78 
79 /* encoding scans */
80 extern const uint8_t ff_alternate_horizontal_scan[64];
81 extern const uint8_t ff_alternate_vertical_scan[64];
82 extern const uint8_t ff_zigzag_direct[64];
83 extern const uint8_t ff_zigzag248_direct[64];
84 
85 /* pixel operations */
86 #define MAX_NEG_CROP 1024
87 
88 /* temporary */
89 extern uint32_t ff_squareTbl[512];
90 extern uint8_t ff_cropTbl[256 + 2 * MAX_NEG_CROP];
91 
92 #define PUTAVG_PIXELS(depth)\
93 void ff_put_pixels8x8_ ## depth ## _c(uint8_t *dst, uint8_t *src, int stride);\
94 void ff_avg_pixels8x8_ ## depth ## _c(uint8_t *dst, uint8_t *src, int stride);\
95 void ff_put_pixels16x16_ ## depth ## _c(uint8_t *dst, uint8_t *src, int stride);\
96 void ff_avg_pixels16x16_ ## depth ## _c(uint8_t *dst, uint8_t *src, int stride);
97 
98 PUTAVG_PIXELS( 8)
99 PUTAVG_PIXELS( 9)
100 PUTAVG_PIXELS(10)
101 
102 #define ff_put_pixels8x8_c ff_put_pixels8x8_8_c
103 #define ff_avg_pixels8x8_c ff_avg_pixels8x8_8_c
104 #define ff_put_pixels16x16_c ff_put_pixels16x16_8_c
105 #define ff_avg_pixels16x16_c ff_avg_pixels16x16_8_c
106 
107 /* VP3 DSP functions */
108 void ff_vp3_idct_c(DCTELEM *block/* align 16*/);
109 void ff_vp3_idct_put_c(uint8_t *dest/*align 8*/, int line_size, DCTELEM *block/*align 16*/);
110 void ff_vp3_idct_add_c(uint8_t *dest/*align 8*/, int line_size, DCTELEM *block/*align 16*/);
111 void ff_vp3_idct_dc_add_c(uint8_t *dest/*align 8*/, int line_size, const DCTELEM *block/*align 16*/);
112 
113 void ff_vp3_v_loop_filter_c(uint8_t *src, int stride, int *bounding_values);
114 void ff_vp3_h_loop_filter_c(uint8_t *src, int stride, int *bounding_values);
115 
116 /* EA functions */
117 void ff_ea_idct_put_c(uint8_t *dest, int linesize, DCTELEM *block);
118 
119 /* RV40 functions */
120 void ff_put_rv40_qpel16_mc33_c(uint8_t *dst, uint8_t *src, int stride);
121 void ff_avg_rv40_qpel16_mc33_c(uint8_t *dst, uint8_t *src, int stride);
122 void ff_put_rv40_qpel8_mc33_c(uint8_t *dst, uint8_t *src, int stride);
123 void ff_avg_rv40_qpel8_mc33_c(uint8_t *dst, uint8_t *src, int stride);
124 
125 /* 1/2^n downscaling functions from imgconvert.c */
126 void ff_shrink22(uint8_t *dst, int dst_wrap, const uint8_t *src, int src_wrap, int width, int height);
127 void ff_shrink44(uint8_t *dst, int dst_wrap, const uint8_t *src, int src_wrap, int width, int height);
128 void ff_shrink88(uint8_t *dst, int dst_wrap, const uint8_t *src, int src_wrap, int width, int height);
129 
130 void ff_gmc_c(uint8_t *dst, uint8_t *src, int stride, int h, int ox, int oy,
131  int dxx, int dxy, int dyx, int dyy, int shift, int r, int width, int height);
132 
133 /* minimum alignment rules ;)
134 If you notice errors in the align stuff, need more alignment for some ASM code
135 for some CPU or need to use a function with less aligned data then send a mail
136 to the libav-devel mailing list, ...
137 
138 !warning These alignments might not match reality, (missing attribute((align))
139 stuff somewhere possible).
140 I (Michael) did not check them, these are just the alignments which I think
141 could be reached easily ...
142 
143 !future video codecs might need functions with less strict alignment
144 */
145 
146 /*
147 void get_pixels_c(DCTELEM *block, const uint8_t *pixels, int line_size);
148 void diff_pixels_c(DCTELEM *block, const uint8_t *s1, const uint8_t *s2, int stride);
149 void put_pixels_clamped_c(const DCTELEM *block, uint8_t *pixels, int line_size);
150 void add_pixels_clamped_c(const DCTELEM *block, uint8_t *pixels, int line_size);
151 void clear_blocks_c(DCTELEM *blocks);
152 */
153 
154 /* add and put pixel (decoding) */
155 // blocksizes for op_pixels_func are 8x4,8x8 16x8 16x16
156 //h for op_pixels_func is limited to {width/2, width} but never larger than 16 and never smaller than 4
157 typedef void (*op_pixels_func)(uint8_t *block/*align width (8 or 16)*/, const uint8_t *pixels/*align 1*/, int line_size, int h);
158 typedef void (*tpel_mc_func)(uint8_t *block/*align width (8 or 16)*/, const uint8_t *pixels/*align 1*/, int line_size, int w, int h);
159 typedef void (*qpel_mc_func)(uint8_t *dst/*align width (8 or 16)*/, uint8_t *src/*align 1*/, int stride);
160 typedef void (*h264_chroma_mc_func)(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int srcStride, int h, int x, int y);
161 
162 typedef void (*op_fill_func)(uint8_t *block/*align width (8 or 16)*/, uint8_t value, int line_size, int h);
163 
164 #define DEF_OLD_QPEL(name)\
165 void ff_put_ ## name (uint8_t *dst/*align width (8 or 16)*/, uint8_t *src/*align 1*/, int stride);\
166 void ff_put_no_rnd_ ## name (uint8_t *dst/*align width (8 or 16)*/, uint8_t *src/*align 1*/, int stride);\
167 void ff_avg_ ## name (uint8_t *dst/*align width (8 or 16)*/, uint8_t *src/*align 1*/, int stride);
168 
169 DEF_OLD_QPEL(qpel16_mc11_old_c)
170 DEF_OLD_QPEL(qpel16_mc31_old_c)
171 DEF_OLD_QPEL(qpel16_mc12_old_c)
172 DEF_OLD_QPEL(qpel16_mc32_old_c)
173 DEF_OLD_QPEL(qpel16_mc13_old_c)
174 DEF_OLD_QPEL(qpel16_mc33_old_c)
175 DEF_OLD_QPEL(qpel8_mc11_old_c)
176 DEF_OLD_QPEL(qpel8_mc31_old_c)
177 DEF_OLD_QPEL(qpel8_mc12_old_c)
178 DEF_OLD_QPEL(qpel8_mc32_old_c)
179 DEF_OLD_QPEL(qpel8_mc13_old_c)
180 DEF_OLD_QPEL(qpel8_mc33_old_c)
181 
182 #define CALL_2X_PIXELS(a, b, n)\
183 static void a(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
184  b(block , pixels , line_size, h);\
185  b(block+n, pixels+n, line_size, h);\
186 }
187 
188 /* motion estimation */
189 // h is limited to {width/2, width, 2*width} but never larger than 16 and never smaller than 2
190 // although currently h<4 is not used as functions with width <8 are neither used nor implemented
191 typedef int (*me_cmp_func)(void /*MpegEncContext*/ *s, uint8_t *blk1/*align width (8 or 16)*/, uint8_t *blk2/*align 1*/, int line_size, int h)/* __attribute__ ((const))*/;
192 
196 typedef struct ScanTable{
197  const uint8_t *scantable;
198  uint8_t permutated[64];
199  uint8_t raster_end[64];
200 #if ARCH_PPC
201 
202  DECLARE_ALIGNED(16, uint8_t, inverse)[64];
203 #endif
204 } ScanTable;
205 
206 void ff_init_scantable(uint8_t *, ScanTable *st, const uint8_t *src_scantable);
207 void ff_init_scantable_permutation(uint8_t *idct_permutation,
208  int idct_permutation_type);
209 
210 #define EMULATED_EDGE(depth) \
211 void ff_emulated_edge_mc_ ## depth (uint8_t *buf, const uint8_t *src, int linesize,\
212  int block_w, int block_h,\
213  int src_x, int src_y, int w, int h);
214 
215 EMULATED_EDGE(8)
216 EMULATED_EDGE(9)
217 EMULATED_EDGE(10)
218 
219 void ff_add_pixels_clamped_c(const DCTELEM *block, uint8_t *dest, int linesize);
220 void ff_put_pixels_clamped_c(const DCTELEM *block, uint8_t *dest, int linesize);
221 void ff_put_signed_pixels_clamped_c(const DCTELEM *block, uint8_t *dest, int linesize);
222 
226 typedef struct DSPContext {
230  int dct_bits;
231 
232  /* pixel ops : interface with DCT */
233  void (*get_pixels)(DCTELEM *block/*align 16*/, const uint8_t *pixels/*align 8*/, int line_size);
234  void (*diff_pixels)(DCTELEM *block/*align 16*/, const uint8_t *s1/*align 8*/, const uint8_t *s2/*align 8*/, int stride);
235  void (*put_pixels_clamped)(const DCTELEM *block/*align 16*/, uint8_t *pixels/*align 8*/, int line_size);
236  void (*put_signed_pixels_clamped)(const DCTELEM *block/*align 16*/, uint8_t *pixels/*align 8*/, int line_size);
237  void (*add_pixels_clamped)(const DCTELEM *block/*align 16*/, uint8_t *pixels/*align 8*/, int line_size);
238  void (*add_pixels8)(uint8_t *pixels, DCTELEM *block, int line_size);
239  void (*add_pixels4)(uint8_t *pixels, DCTELEM *block, int line_size);
240  int (*sum_abs_dctelem)(DCTELEM *block/*align 16*/);
253  void (*emulated_edge_mc)(uint8_t *buf, const uint8_t *src, int linesize,
254  int block_w, int block_h,
255  int src_x, int src_y, int w, int h);
259  void (*gmc1)(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int srcStride, int h, int x16, int y16, int rounder);
263  void (*gmc )(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int stride, int h, int ox, int oy,
264  int dxx, int dxy, int dyx, int dyy, int shift, int r, int width, int height);
265  void (*clear_block)(DCTELEM *block/*align 16*/);
266  void (*clear_blocks)(DCTELEM *blocks/*align 16*/);
267  int (*pix_sum)(uint8_t * pix, int line_size);
268  int (*pix_norm1)(uint8_t * pix, int line_size);
269 // 16x16 8x8 4x4 2x2 16x8 8x4 4x2 8x16 4x8 2x4
270 
271  me_cmp_func sad[6]; /* identical to pix_absAxA except additional void * */
273  me_cmp_func hadamard8_diff[6];
274  me_cmp_func dct_sad[6];
275  me_cmp_func quant_psnr[6];
276  me_cmp_func bit[6];
277  me_cmp_func rd[6];
278  me_cmp_func vsad[6];
279  me_cmp_func vsse[6];
280  me_cmp_func nsse[6];
281  me_cmp_func w53[6];
282  me_cmp_func w97[6];
283  me_cmp_func dct_max[6];
284  me_cmp_func dct264_sad[6];
285 
286  me_cmp_func me_pre_cmp[6];
287  me_cmp_func me_cmp[6];
288  me_cmp_func me_sub_cmp[6];
289  me_cmp_func mb_cmp[6];
290  me_cmp_func ildct_cmp[6]; //only width 16 used
291  me_cmp_func frame_skip_cmp[6]; //only width 8 used
292 
293  int (*ssd_int8_vs_int16)(const int8_t *pix1, const int16_t *pix2,
294  int size);
295 
306  op_pixels_func put_pixels_tab[4][4];
307 
318  op_pixels_func avg_pixels_tab[4][4];
319 
330  op_pixels_func put_no_rnd_pixels_tab[4][4];
331 
342  op_pixels_func avg_no_rnd_pixels_tab[4][4];
343 
344  void (*put_no_rnd_pixels_l2[2])(uint8_t *block/*align width (8 or 16)*/, const uint8_t *a/*align 1*/, const uint8_t *b/*align 1*/, int line_size, int h);
345 
356  tpel_mc_func put_tpel_pixels_tab[11]; //FIXME individual func ptr per width?
357  tpel_mc_func avg_tpel_pixels_tab[11]; //FIXME individual func ptr per width?
358 
359  qpel_mc_func put_qpel_pixels_tab[2][16];
360  qpel_mc_func avg_qpel_pixels_tab[2][16];
361  qpel_mc_func put_no_rnd_qpel_pixels_tab[2][16];
362  qpel_mc_func avg_no_rnd_qpel_pixels_tab[2][16];
363  qpel_mc_func put_mspel_pixels_tab[8];
364 
368  h264_chroma_mc_func put_h264_chroma_pixels_tab[3];
369  h264_chroma_mc_func avg_h264_chroma_pixels_tab[3];
370 
371  qpel_mc_func put_h264_qpel_pixels_tab[4][16];
372  qpel_mc_func avg_h264_qpel_pixels_tab[4][16];
373 
374  qpel_mc_func put_2tap_qpel_pixels_tab[4][16];
375  qpel_mc_func avg_2tap_qpel_pixels_tab[4][16];
376 
377  me_cmp_func pix_abs[2][4];
378 
379  /* huffyuv specific */
380  void (*add_bytes)(uint8_t *dst/*align 16*/, uint8_t *src/*align 16*/, int w);
381  void (*add_bytes_l2)(uint8_t *dst/*align 16*/, uint8_t *src1/*align 16*/, uint8_t *src2/*align 16*/, int w);
382  void (*diff_bytes)(uint8_t *dst/*align 16*/, uint8_t *src1/*align 16*/, uint8_t *src2/*align 1*/,int w);
387  void (*sub_hfyu_median_prediction)(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int w, int *left, int *left_top);
388  void (*add_hfyu_median_prediction)(uint8_t *dst, const uint8_t *top, const uint8_t *diff, int w, int *left, int *left_top);
389  int (*add_hfyu_left_prediction)(uint8_t *dst, const uint8_t *src, int w, int left);
390  void (*add_hfyu_left_prediction_bgr32)(uint8_t *dst, const uint8_t *src, int w, int *red, int *green, int *blue, int *alpha);
391  /* this might write to dst[w] */
392  void (*add_png_paeth_prediction)(uint8_t *dst, uint8_t *src, uint8_t *top, int w, int bpp);
393  void (*bswap_buf)(uint32_t *dst, const uint32_t *src, int w);
394  void (*bswap16_buf)(uint16_t *dst, const uint16_t *src, int len);
395 
396  void (*h263_v_loop_filter)(uint8_t *src, int stride, int qscale);
397  void (*h263_h_loop_filter)(uint8_t *src, int stride, int qscale);
398 
399  void (*h261_loop_filter)(uint8_t *src, int stride);
400 
401  void (*x8_v_loop_filter)(uint8_t *src, int stride, int qscale);
402  void (*x8_h_loop_filter)(uint8_t *src, int stride, int qscale);
403 
404  void (*vp3_idct_dc_add)(uint8_t *dest/*align 8*/, int line_size, const DCTELEM *block/*align 16*/);
405  void (*vp3_v_loop_filter)(uint8_t *src, int stride, int *bounding_values);
406  void (*vp3_h_loop_filter)(uint8_t *src, int stride, int *bounding_values);
407 
408  /* assume len is a multiple of 4, and arrays are 16-byte aligned */
409  void (*vorbis_inverse_coupling)(float *mag, float *ang, int blocksize);
410  void (*ac3_downmix)(float (*samples)[256], float (*matrix)[2], int out_ch, int in_ch, int len);
411  /* assume len is a multiple of 8, and arrays are 16-byte aligned */
412  void (*vector_fmul)(float *dst, const float *src0, const float *src1, int len);
413  void (*vector_fmul_reverse)(float *dst, const float *src0, const float *src1, int len);
414  /* assume len is a multiple of 8, and src arrays are 16-byte aligned */
415  void (*vector_fmul_add)(float *dst, const float *src0, const float *src1, const float *src2, int len);
416  /* assume len is a multiple of 4, and arrays are 16-byte aligned */
417  void (*vector_fmul_window)(float *dst, const float *src0, const float *src1, const float *win, int len);
418  /* assume len is a multiple of 8, and arrays are 16-byte aligned */
419  void (*vector_clipf)(float *dst /* align 16 */, const float *src /* align 16 */, float min, float max, int len /* align 16 */);
428  void (*vector_fmul_scalar)(float *dst, const float *src, float mul,
429  int len);
439  void (*vector_fmac_scalar)(float *dst, const float *src, float mul,
440  int len);
447  float (*scalarproduct_float)(const float *v1, const float *v2, int len);
454  void (*butterflies_float)(float *restrict v1, float *restrict v2, int len);
455 
470  void (*butterflies_float_interleave)(float *dst, const float *src0,
471  const float *src1, int len);
472 
473  /* (I)DCT */
474  void (*fdct)(DCTELEM *block/* align 16*/);
475  void (*fdct248)(DCTELEM *block/* align 16*/);
476 
477  /* IDCT really*/
478  void (*idct)(DCTELEM *block/* align 16*/);
479 
485  void (*idct_put)(uint8_t *dest/*align 8*/, int line_size, DCTELEM *block/*align 16*/);
486 
491  void (*idct_add)(uint8_t *dest/*align 8*/, int line_size, DCTELEM *block/*align 16*/);
492 
505  uint8_t idct_permutation[64];
507 #define FF_NO_IDCT_PERM 1
508 #define FF_LIBMPEG2_IDCT_PERM 2
509 #define FF_SIMPLE_IDCT_PERM 3
510 #define FF_TRANSPOSE_IDCT_PERM 4
511 #define FF_PARTTRANS_IDCT_PERM 5
512 #define FF_SSE2_IDCT_PERM 6
513 
514  int (*try_8x8basis)(int16_t rem[64], int16_t weight[64], int16_t basis[64], int scale);
515  void (*add_8x8basis)(int16_t rem[64], int16_t basis[64], int scale);
516 #define BASIS_SHIFT 16
517 #define RECON_SHIFT 6
518 
519  void (*draw_edges)(uint8_t *buf, int wrap, int width, int height, int w, int h, int sides);
520 #define EDGE_WIDTH 16
521 #define EDGE_TOP 1
522 #define EDGE_BOTTOM 2
523 
524  void (*prefetch)(void *mem, int stride, int h);
525 
526  void (*shrink[4])(uint8_t *dst, int dst_wrap, const uint8_t *src, int src_wrap, int width, int height);
527 
528  /* mlp/truehd functions */
529  void (*mlp_filter_channel)(int32_t *state, const int32_t *coeff,
530  int firorder, int iirorder,
531  unsigned int filter_shift, int32_t mask, int blocksize,
532  int32_t *sample_buffer);
533 
534  /* intrax8 functions */
535  void (*x8_spatial_compensation[12])(uint8_t *src , uint8_t *dst, int linesize);
536  void (*x8_setup_spatial_compensation)(uint8_t *src, uint8_t *dst, int linesize,
537  int * range, int * sum, int edges);
538 
544  int32_t (*scalarproduct_int16)(const int16_t *v1, const int16_t *v2/*align 16*/, int len, int shift);
545  /* ape functions */
551  int32_t (*scalarproduct_and_madd_int16)(int16_t *v1/*align 16*/, const int16_t *v2, const int16_t *v3, int len, int mul);
552 
564  void (*apply_window_int16)(int16_t *output, const int16_t *input,
565  const int16_t *window, unsigned int len);
566 
580  void (*vector_clip_int32)(int32_t *dst, const int32_t *src, int32_t min,
581  int32_t max, unsigned int len);
582 
583  op_fill_func fill_block_tab[2];
584 } DSPContext;
585 
586 void dsputil_static_init(void);
587 void dsputil_init(DSPContext* p, AVCodecContext *avctx);
588 
589 int ff_check_alignment(void);
590 
595 void ff_block_permute(DCTELEM *block, uint8_t *permutation, const uint8_t *scantable, int last);
596 
597 void ff_set_cmp(DSPContext* c, me_cmp_func *cmp, int type);
598 
599 #define BYTE_VEC32(c) ((c)*0x01010101UL)
600 #define BYTE_VEC64(c) ((c)*0x0001000100010001UL)
601 
602 static inline uint32_t rnd_avg32(uint32_t a, uint32_t b)
603 {
604  return (a | b) - (((a ^ b) & ~BYTE_VEC32(0x01)) >> 1);
605 }
606 
607 static inline uint32_t no_rnd_avg32(uint32_t a, uint32_t b)
608 {
609  return (a & b) + (((a ^ b) & ~BYTE_VEC32(0x01)) >> 1);
610 }
611 
612 static inline uint64_t rnd_avg64(uint64_t a, uint64_t b)
613 {
614  return (a | b) - (((a ^ b) & ~BYTE_VEC64(0x01)) >> 1);
615 }
616 
617 static inline uint64_t no_rnd_avg64(uint64_t a, uint64_t b)
618 {
619  return (a & b) + (((a ^ b) & ~BYTE_VEC64(0x01)) >> 1);
620 }
621 
622 static inline int get_penalty_factor(int lambda, int lambda2, int type){
623  switch(type&0xFF){
624  default:
625  case FF_CMP_SAD:
626  return lambda>>FF_LAMBDA_SHIFT;
627  case FF_CMP_DCT:
628  return (3*lambda)>>(FF_LAMBDA_SHIFT+1);
629  case FF_CMP_W53:
630  return (4*lambda)>>(FF_LAMBDA_SHIFT);
631  case FF_CMP_W97:
632  return (2*lambda)>>(FF_LAMBDA_SHIFT);
633  case FF_CMP_SATD:
634  case FF_CMP_DCT264:
635  return (2*lambda)>>FF_LAMBDA_SHIFT;
636  case FF_CMP_RD:
637  case FF_CMP_PSNR:
638  case FF_CMP_SSE:
639  case FF_CMP_NSSE:
640  return lambda2>>FF_LAMBDA_SHIFT;
641  case FF_CMP_BIT:
642  return 1;
643  }
644 }
645 
655 
658 void ff_mlp_init(DSPContext* c, AVCodecContext *avctx);
659 void ff_mlp_init_x86(DSPContext* c, AVCodecContext *avctx);
660 
661 #if ARCH_ARM
662 
663 #if HAVE_NEON
664 # define STRIDE_ALIGN 16
665 #endif
666 
667 #elif ARCH_PPC
668 
669 #define STRIDE_ALIGN 16
670 
671 #elif HAVE_MMI
672 
673 #define STRIDE_ALIGN 16
674 
675 #endif
676 
677 #ifndef STRIDE_ALIGN
678 # define STRIDE_ALIGN 8
679 #endif
680 
681 #define LOCAL_ALIGNED_A(a, t, v, s, o, ...) \
682  uint8_t la_##v[sizeof(t s o) + (a)]; \
683  t (*v) o = (void *)FFALIGN((uintptr_t)la_##v, a)
684 
685 #define LOCAL_ALIGNED_D(a, t, v, s, o, ...) DECLARE_ALIGNED(a, t, v) s o
686 
687 #define LOCAL_ALIGNED(a, t, v, ...) LOCAL_ALIGNED_A(a, t, v, __VA_ARGS__,,)
688 
689 #if HAVE_LOCAL_ALIGNED_8
690 # define LOCAL_ALIGNED_8(t, v, ...) LOCAL_ALIGNED_D(8, t, v, __VA_ARGS__,,)
691 #else
692 # define LOCAL_ALIGNED_8(t, v, ...) LOCAL_ALIGNED(8, t, v, __VA_ARGS__)
693 #endif
694 
695 #if HAVE_LOCAL_ALIGNED_16
696 # define LOCAL_ALIGNED_16(t, v, ...) LOCAL_ALIGNED_D(16, t, v, __VA_ARGS__,,)
697 #else
698 # define LOCAL_ALIGNED_16(t, v, ...) LOCAL_ALIGNED(16, t, v, __VA_ARGS__)
699 #endif
700 
701 #define WRAPPER8_16(name8, name16)\
702 static int name16(void /*MpegEncContext*/ *s, uint8_t *dst, uint8_t *src, int stride, int h){\
703  return name8(s, dst , src , stride, h)\
704  +name8(s, dst+8 , src+8 , stride, h);\
705 }
706 
707 #define WRAPPER8_16_SQ(name8, name16)\
708 static int name16(void /*MpegEncContext*/ *s, uint8_t *dst, uint8_t *src, int stride, int h){\
709  int score=0;\
710  score +=name8(s, dst , src , stride, 8);\
711  score +=name8(s, dst+8 , src+8 , stride, 8);\
712  if(h==16){\
713  dst += 8*stride;\
714  src += 8*stride;\
715  score +=name8(s, dst , src , stride, 8);\
716  score +=name8(s, dst+8 , src+8 , stride, 8);\
717  }\
718  return score;\
719 }
720 
721 
722 static inline void copy_block2(uint8_t *dst, const uint8_t *src, int dstStride, int srcStride, int h)
723 {
724  int i;
725  for(i=0; i<h; i++)
726  {
727  AV_WN16(dst , AV_RN16(src ));
728  dst+=dstStride;
729  src+=srcStride;
730  }
731 }
732 
733 static inline void copy_block4(uint8_t *dst, const uint8_t *src, int dstStride, int srcStride, int h)
734 {
735  int i;
736  for(i=0; i<h; i++)
737  {
738  AV_WN32(dst , AV_RN32(src ));
739  dst+=dstStride;
740  src+=srcStride;
741  }
742 }
743 
744 static inline void copy_block8(uint8_t *dst, const uint8_t *src, int dstStride, int srcStride, int h)
745 {
746  int i;
747  for(i=0; i<h; i++)
748  {
749  AV_WN32(dst , AV_RN32(src ));
750  AV_WN32(dst+4 , AV_RN32(src+4 ));
751  dst+=dstStride;
752  src+=srcStride;
753  }
754 }
755 
756 static inline void copy_block9(uint8_t *dst, const uint8_t *src, int dstStride, int srcStride, int h)
757 {
758  int i;
759  for(i=0; i<h; i++)
760  {
761  AV_WN32(dst , AV_RN32(src ));
762  AV_WN32(dst+4 , AV_RN32(src+4 ));
763  dst[8]= src[8];
764  dst+=dstStride;
765  src+=srcStride;
766  }
767 }
768 
769 static inline void copy_block16(uint8_t *dst, const uint8_t *src, int dstStride, int srcStride, int h)
770 {
771  int i;
772  for(i=0; i<h; i++)
773  {
774  AV_WN32(dst , AV_RN32(src ));
775  AV_WN32(dst+4 , AV_RN32(src+4 ));
776  AV_WN32(dst+8 , AV_RN32(src+8 ));
777  AV_WN32(dst+12, AV_RN32(src+12));
778  dst+=dstStride;
779  src+=srcStride;
780  }
781 }
782 
783 static inline void copy_block17(uint8_t *dst, const uint8_t *src, int dstStride, int srcStride, int h)
784 {
785  int i;
786  for(i=0; i<h; i++)
787  {
788  AV_WN32(dst , AV_RN32(src ));
789  AV_WN32(dst+4 , AV_RN32(src+4 ));
790  AV_WN32(dst+8 , AV_RN32(src+8 ));
791  AV_WN32(dst+12, AV_RN32(src+12));
792  dst[16]= src[16];
793  dst+=dstStride;
794  src+=srcStride;
795  }
796 }
797 
798 #endif /* AVCODEC_DSPUTIL_H */