2002-05-26 01:45:33 +03:00
|
|
|
/*
|
|
|
|
* DSP utils
|
|
|
|
* Copyright (c) 2000, 2001, 2002 Fabrice Bellard.
|
2004-01-10 18:04:55 +02:00
|
|
|
* Copyright (c) 2002-2004 Michael Niedermayer <michaelni@gmx.at>
|
2002-05-26 01:45:33 +03:00
|
|
|
*
|
2006-10-07 18:30:46 +03:00
|
|
|
* This file is part of FFmpeg.
|
|
|
|
*
|
|
|
|
* FFmpeg is free software; you can redistribute it and/or
|
2002-05-26 01:45:33 +03:00
|
|
|
* modify it under the terms of the GNU Lesser General Public
|
|
|
|
* License as published by the Free Software Foundation; either
|
2006-10-07 18:30:46 +03:00
|
|
|
* version 2.1 of the License, or (at your option) any later version.
|
2002-05-26 01:45:33 +03:00
|
|
|
*
|
2006-10-07 18:30:46 +03:00
|
|
|
* FFmpeg is distributed in the hope that it will be useful,
|
2002-05-26 01:45:33 +03:00
|
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
|
|
|
|
* Lesser General Public License for more details.
|
|
|
|
*
|
|
|
|
* You should have received a copy of the GNU Lesser General Public
|
2006-10-07 18:30:46 +03:00
|
|
|
* License along with FFmpeg; if not, write to the Free Software
|
2006-01-13 00:43:26 +02:00
|
|
|
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
|
2002-05-26 01:45:33 +03:00
|
|
|
*/
|
2003-03-06 12:08:15 +02:00
|
|
|
|
|
|
|
/**
|
|
|
|
* @file dsputil.h
|
2003-03-06 13:32:04 +02:00
|
|
|
* DSP utils.
|
2003-04-25 03:21:53 +03:00
|
|
|
* note, many functions in here may use MMX which trashes the FPU state, it is
|
|
|
|
* absolutely necessary to call emms_c() between dsp & float/double code
|
2003-03-06 12:08:15 +02:00
|
|
|
*/
|
|
|
|
|
2008-08-31 10:39:47 +03:00
|
|
|
#ifndef AVCODEC_DSPUTIL_H
|
|
|
|
#define AVCODEC_DSPUTIL_H
|
2001-07-22 17:18:56 +03:00
|
|
|
|
2002-02-27 00:14:27 +02:00
|
|
|
#include "avcodec.h"
|
2001-07-22 17:18:56 +03:00
|
|
|
|
2003-03-06 12:08:15 +02:00
|
|
|
|
2002-03-09 15:01:16 +02:00
|
|
|
//#define DEBUG
|
2001-07-22 17:18:56 +03:00
|
|
|
/* dct code */
|
|
|
|
typedef short DCTELEM;
|
2006-03-16 21:18:18 +02:00
|
|
|
typedef int DWTELEM;
|
2007-08-25 06:00:51 +03:00
|
|
|
typedef short IDWTELEM;
|
2001-07-22 17:18:56 +03:00
|
|
|
|
2002-06-04 15:58:40 +03:00
|
|
|
void fdct_ifast (DCTELEM *data);
|
2003-10-25 03:32:54 +03:00
|
|
|
void fdct_ifast248 (DCTELEM *data);
|
2002-08-30 02:55:32 +03:00
|
|
|
void ff_jpeg_fdct_islow (DCTELEM *data);
|
2003-10-24 21:28:01 +03:00
|
|
|
void ff_fdct248_islow (DCTELEM *data);
|
2001-07-22 17:18:56 +03:00
|
|
|
|
|
|
|
void j_rev_dct (DCTELEM *data);
|
2004-09-26 02:18:58 +03:00
|
|
|
void j_rev_dct4 (DCTELEM *data);
|
2004-09-26 03:18:12 +03:00
|
|
|
void j_rev_dct2 (DCTELEM *data);
|
2004-09-26 13:01:40 +03:00
|
|
|
void j_rev_dct1 (DCTELEM *data);
|
2007-11-09 23:37:48 +02:00
|
|
|
void ff_wmv2_idct_c(DCTELEM *data);
|
2001-07-22 17:18:56 +03:00
|
|
|
|
2002-09-20 01:14:53 +03:00
|
|
|
void ff_fdct_mmx(DCTELEM *block);
|
2003-10-23 16:22:20 +03:00
|
|
|
void ff_fdct_mmx2(DCTELEM *block);
|
2004-01-29 03:15:25 +02:00
|
|
|
void ff_fdct_sse2(DCTELEM *block);
|
2001-07-22 17:18:56 +03:00
|
|
|
|
2005-06-03 00:15:20 +03:00
|
|
|
void ff_h264_idct8_add_c(uint8_t *dst, DCTELEM *block, int stride);
|
2004-09-27 22:47:17 +03:00
|
|
|
void ff_h264_idct_add_c(uint8_t *dst, DCTELEM *block, int stride);
|
2006-02-10 08:55:25 +02:00
|
|
|
void ff_h264_idct8_dc_add_c(uint8_t *dst, DCTELEM *block, int stride);
|
|
|
|
void ff_h264_idct_dc_add_c(uint8_t *dst, DCTELEM *block, int stride);
|
2004-09-27 22:47:17 +03:00
|
|
|
void ff_h264_lowres_idct_add_c(uint8_t *dst, int stride, DCTELEM *block);
|
|
|
|
void ff_h264_lowres_idct_put_c(uint8_t *dst, int stride, DCTELEM *block);
|
2008-12-18 04:36:48 +02:00
|
|
|
void ff_h264_idct_add16_c(uint8_t *dst, const int *blockoffset, DCTELEM *block, int stride, const uint8_t nnzc[6*8]);
|
|
|
|
void ff_h264_idct_add16intra_c(uint8_t *dst, const int *blockoffset, DCTELEM *block, int stride, const uint8_t nnzc[6*8]);
|
|
|
|
void ff_h264_idct8_add4_c(uint8_t *dst, const int *blockoffset, DCTELEM *block, int stride, const uint8_t nnzc[6*8]);
|
|
|
|
void ff_h264_idct_add8_c(uint8_t **dest, const int *blockoffset, DCTELEM *block, int stride, const uint8_t nnzc[6*8]);
|
2004-09-27 22:47:17 +03:00
|
|
|
|
2006-08-10 22:06:25 +03:00
|
|
|
void ff_vector_fmul_add_add_c(float *dst, const float *src0, const float *src1,
|
|
|
|
const float *src2, int src3, int blocksize, int step);
|
2008-07-13 17:56:01 +03:00
|
|
|
void ff_vector_fmul_window_c(float *dst, const float *src0, const float *src1,
|
|
|
|
const float *win, float add_bias, int len);
|
2008-07-08 00:04:29 +03:00
|
|
|
void ff_float_to_int16_c(int16_t *dst, const float *src, long len);
|
2008-07-29 17:17:13 +03:00
|
|
|
void ff_float_to_int16_interleave_c(int16_t *dst, const float **src, long len, int channels);
|
2006-08-10 22:06:25 +03:00
|
|
|
|
2001-08-08 01:43:19 +03:00
|
|
|
/* encoding scans */
|
2003-02-11 18:35:48 +02:00
|
|
|
extern const uint8_t ff_alternate_horizontal_scan[64];
|
|
|
|
extern const uint8_t ff_alternate_vertical_scan[64];
|
|
|
|
extern const uint8_t ff_zigzag_direct[64];
|
2003-10-24 21:28:01 +03:00
|
|
|
extern const uint8_t ff_zigzag248_direct[64];
|
2002-01-10 02:51:45 +02:00
|
|
|
|
2001-07-22 17:18:56 +03:00
|
|
|
/* pixel operations */
|
2004-06-21 04:01:44 +03:00
|
|
|
#define MAX_NEG_CROP 1024
|
2001-07-22 17:18:56 +03:00
|
|
|
|
|
|
|
/* temporary */
|
2006-11-12 23:48:04 +02:00
|
|
|
extern uint32_t ff_squareTbl[512];
|
2006-11-12 22:08:09 +02:00
|
|
|
extern uint8_t ff_cropTbl[256 + 2 * MAX_NEG_CROP];
|
2001-07-22 17:18:56 +03:00
|
|
|
|
2004-03-08 04:13:45 +02:00
|
|
|
/* VP3 DSP functions */
|
2005-05-17 14:12:04 +03:00
|
|
|
void ff_vp3_idct_c(DCTELEM *block/* align 16*/);
|
|
|
|
void ff_vp3_idct_put_c(uint8_t *dest/*align 8*/, int line_size, DCTELEM *block/*align 16*/);
|
|
|
|
void ff_vp3_idct_add_c(uint8_t *dest/*align 8*/, int line_size, DCTELEM *block/*align 16*/);
|
2001-07-22 17:18:56 +03:00
|
|
|
|
2008-10-04 13:26:17 +03:00
|
|
|
void ff_vp3_v_loop_filter_c(uint8_t *src, int stride, int *bounding_values);
|
|
|
|
void ff_vp3_h_loop_filter_c(uint8_t *src, int stride, int *bounding_values);
|
|
|
|
|
2006-03-31 23:43:37 +03:00
|
|
|
/* 1/2^n downscaling functions from imgconvert.c */
|
|
|
|
void ff_img_copy_plane(uint8_t *dst, int dst_wrap, const uint8_t *src, int src_wrap, int width, int height);
|
|
|
|
void ff_shrink22(uint8_t *dst, int dst_wrap, const uint8_t *src, int src_wrap, int width, int height);
|
|
|
|
void ff_shrink44(uint8_t *dst, int dst_wrap, const uint8_t *src, int src_wrap, int width, int height);
|
|
|
|
void ff_shrink88(uint8_t *dst, int dst_wrap, const uint8_t *src, int src_wrap, int width, int height);
|
2006-04-04 12:23:45 +03:00
|
|
|
|
|
|
|
void ff_gmc_c(uint8_t *dst, uint8_t *src, int stride, int h, int ox, int oy,
|
|
|
|
int dxx, int dxy, int dyx, int dyy, int shift, int r, int width, int height);
|
2006-03-31 23:43:37 +03:00
|
|
|
|
2002-09-14 22:00:09 +03:00
|
|
|
/* minimum alignment rules ;)
|
2008-03-10 20:42:09 +02:00
|
|
|
If you notice errors in the align stuff, need more alignment for some ASM code
|
|
|
|
for some CPU or need to use a function with less aligned data then send a mail
|
|
|
|
to the ffmpeg-devel mailing list, ...
|
|
|
|
|
|
|
|
!warning These alignments might not match reality, (missing attribute((align))
|
|
|
|
stuff somewhere possible).
|
2008-03-13 01:58:46 +02:00
|
|
|
I (Michael) did not check them, these are just the alignments which I think
|
2008-03-10 20:42:09 +02:00
|
|
|
could be reached easily ...
|
2001-07-22 17:18:56 +03:00
|
|
|
|
2002-09-14 22:00:09 +03:00
|
|
|
!future video codecs might need functions with less strict alignment
|
|
|
|
*/
|
|
|
|
|
2002-11-11 11:40:17 +02:00
|
|
|
/*
|
2003-02-11 18:35:48 +02:00
|
|
|
void get_pixels_c(DCTELEM *block, const uint8_t *pixels, int line_size);
|
|
|
|
void diff_pixels_c(DCTELEM *block, const uint8_t *s1, const uint8_t *s2, int stride);
|
|
|
|
void put_pixels_clamped_c(const DCTELEM *block, uint8_t *pixels, int line_size);
|
|
|
|
void add_pixels_clamped_c(const DCTELEM *block, uint8_t *pixels, int line_size);
|
2002-03-28 15:41:04 +02:00
|
|
|
void clear_blocks_c(DCTELEM *blocks);
|
2002-11-11 11:40:17 +02:00
|
|
|
*/
|
2001-07-22 17:18:56 +03:00
|
|
|
|
|
|
|
/* add and put pixel (decoding) */
|
2002-09-14 22:00:09 +03:00
|
|
|
// blocksizes for op_pixels_func are 8x4,8x8 16x8 16x16
|
2003-12-30 21:19:45 +02:00
|
|
|
//h for op_pixels_func is limited to {width/2, width} but never larger than 16 and never smaller then 4
|
2003-02-11 18:35:48 +02:00
|
|
|
typedef void (*op_pixels_func)(uint8_t *block/*align width (8 or 16)*/, const uint8_t *pixels/*align 1*/, int line_size, int h);
|
2003-05-16 02:30:03 +03:00
|
|
|
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);
|
2003-02-11 18:35:48 +02:00
|
|
|
typedef void (*qpel_mc_func)(uint8_t *dst/*align width (8 or 16)*/, uint8_t *src/*align 1*/, int stride);
|
2003-04-04 17:42:28 +03:00
|
|
|
typedef void (*h264_chroma_mc_func)(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int srcStride, int h, int x, int y);
|
2005-01-07 07:56:07 +02:00
|
|
|
typedef void (*h264_weight_func)(uint8_t *block, int stride, int log2_denom, int weight, int offset);
|
2006-01-09 05:38:37 +02:00
|
|
|
typedef void (*h264_biweight_func)(uint8_t *dst, uint8_t *src, int stride, int log2_denom, int weightd, int weights, int offset);
|
2002-09-11 15:39:53 +03:00
|
|
|
|
2003-01-09 22:42:37 +02:00
|
|
|
#define DEF_OLD_QPEL(name)\
|
2003-02-11 18:35:48 +02:00
|
|
|
void ff_put_ ## name (uint8_t *dst/*align width (8 or 16)*/, uint8_t *src/*align 1*/, int stride);\
|
|
|
|
void ff_put_no_rnd_ ## name (uint8_t *dst/*align width (8 or 16)*/, uint8_t *src/*align 1*/, int stride);\
|
|
|
|
void ff_avg_ ## name (uint8_t *dst/*align width (8 or 16)*/, uint8_t *src/*align 1*/, int stride);
|
2003-01-09 22:42:37 +02:00
|
|
|
|
|
|
|
DEF_OLD_QPEL(qpel16_mc11_old_c)
|
|
|
|
DEF_OLD_QPEL(qpel16_mc31_old_c)
|
|
|
|
DEF_OLD_QPEL(qpel16_mc12_old_c)
|
|
|
|
DEF_OLD_QPEL(qpel16_mc32_old_c)
|
|
|
|
DEF_OLD_QPEL(qpel16_mc13_old_c)
|
|
|
|
DEF_OLD_QPEL(qpel16_mc33_old_c)
|
|
|
|
DEF_OLD_QPEL(qpel8_mc11_old_c)
|
|
|
|
DEF_OLD_QPEL(qpel8_mc31_old_c)
|
|
|
|
DEF_OLD_QPEL(qpel8_mc12_old_c)
|
|
|
|
DEF_OLD_QPEL(qpel8_mc32_old_c)
|
|
|
|
DEF_OLD_QPEL(qpel8_mc13_old_c)
|
|
|
|
DEF_OLD_QPEL(qpel8_mc33_old_c)
|
2002-09-11 15:39:53 +03:00
|
|
|
|
|
|
|
#define CALL_2X_PIXELS(a, b, n)\
|
|
|
|
static void a(uint8_t *block, const uint8_t *pixels, int line_size, int h){\
|
|
|
|
b(block , pixels , line_size, h);\
|
|
|
|
b(block+n, pixels+n, line_size, h);\
|
|
|
|
}
|
2002-03-09 15:01:16 +02:00
|
|
|
|
2001-07-22 17:18:56 +03:00
|
|
|
/* motion estimation */
|
2003-12-30 21:19:45 +02:00
|
|
|
// h is limited to {width/2, width, 2*width} but never larger than 16 and never smaller then 2
|
2007-06-12 12:29:25 +03:00
|
|
|
// although currently h<4 is not used as functions with width <8 are neither used nor implemented
|
2003-12-30 18:07:57 +02:00
|
|
|
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))*/;
|
2002-12-28 01:51:46 +02:00
|
|
|
|
2003-04-04 17:42:28 +03:00
|
|
|
|
2006-03-16 21:18:18 +02:00
|
|
|
// for snow slices
|
|
|
|
typedef struct slice_buffer_s slice_buffer;
|
|
|
|
|
2008-03-04 12:26:56 +02:00
|
|
|
/**
|
|
|
|
* Scantable.
|
|
|
|
*/
|
|
|
|
typedef struct ScanTable{
|
|
|
|
const uint8_t *scantable;
|
|
|
|
uint8_t permutated[64];
|
|
|
|
uint8_t raster_end[64];
|
|
|
|
#ifdef ARCH_POWERPC
|
|
|
|
/** Used by dct_quantize_altivec to find last-non-zero */
|
2008-03-04 20:35:12 +02:00
|
|
|
DECLARE_ALIGNED(16, uint8_t, inverse[64]);
|
2008-03-04 12:26:56 +02:00
|
|
|
#endif
|
|
|
|
} ScanTable;
|
|
|
|
|
|
|
|
void ff_init_scantable(uint8_t *, ScanTable *st, const uint8_t *src_scantable);
|
|
|
|
|
2008-03-05 01:10:47 +02:00
|
|
|
void ff_emulated_edge_mc(uint8_t *buf, uint8_t *src, int linesize,
|
|
|
|
int block_w, int block_h,
|
|
|
|
int src_x, int src_y, int w, int h);
|
|
|
|
|
2003-03-06 12:08:15 +02:00
|
|
|
/**
|
|
|
|
* DSPContext.
|
|
|
|
*/
|
2002-11-11 11:40:17 +02:00
|
|
|
typedef struct DSPContext {
|
|
|
|
/* pixel ops : interface with DCT */
|
2003-02-11 18:35:48 +02:00
|
|
|
void (*get_pixels)(DCTELEM *block/*align 16*/, const uint8_t *pixels/*align 8*/, int line_size);
|
|
|
|
void (*diff_pixels)(DCTELEM *block/*align 16*/, const uint8_t *s1/*align 8*/, const uint8_t *s2/*align 8*/, int stride);
|
|
|
|
void (*put_pixels_clamped)(const DCTELEM *block/*align 16*/, uint8_t *pixels/*align 8*/, int line_size);
|
2004-04-27 06:58:06 +03:00
|
|
|
void (*put_signed_pixels_clamped)(const DCTELEM *block/*align 16*/, uint8_t *pixels/*align 8*/, int line_size);
|
2003-02-11 18:35:48 +02:00
|
|
|
void (*add_pixels_clamped)(const DCTELEM *block/*align 16*/, uint8_t *pixels/*align 8*/, int line_size);
|
2005-06-17 11:24:35 +03:00
|
|
|
void (*add_pixels8)(uint8_t *pixels, DCTELEM *block, int line_size);
|
|
|
|
void (*add_pixels4)(uint8_t *pixels, DCTELEM *block, int line_size);
|
2007-05-12 05:41:25 +03:00
|
|
|
int (*sum_abs_dctelem)(DCTELEM *block/*align 16*/);
|
2003-03-15 04:06:09 +02:00
|
|
|
/**
|
|
|
|
* translational global motion compensation.
|
|
|
|
*/
|
2003-02-11 18:35:48 +02:00
|
|
|
void (*gmc1)(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int srcStride, int h, int x16, int y16, int rounder);
|
2003-03-15 04:06:09 +02:00
|
|
|
/**
|
|
|
|
* global motion compensation.
|
|
|
|
*/
|
2003-02-11 18:35:48 +02:00
|
|
|
void (*gmc )(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int stride, int h, int ox, int oy,
|
2005-12-22 03:10:11 +02:00
|
|
|
int dxx, int dxy, int dyx, int dyy, int shift, int r, int width, int height);
|
2008-12-10 23:35:17 +02:00
|
|
|
void (*clear_block)(DCTELEM *block/*align 16*/);
|
2002-11-11 11:40:17 +02:00
|
|
|
void (*clear_blocks)(DCTELEM *blocks/*align 16*/);
|
2003-02-11 18:35:48 +02:00
|
|
|
int (*pix_sum)(uint8_t * pix, int line_size);
|
|
|
|
int (*pix_norm1)(uint8_t * pix, int line_size);
|
2003-12-30 18:07:57 +02:00
|
|
|
// 16x16 8x8 4x4 2x2 16x8 8x4 4x2 8x16 4x8 2x4
|
2005-12-17 20:14:38 +02:00
|
|
|
|
2004-01-06 00:57:07 +02:00
|
|
|
me_cmp_func sad[5]; /* identical to pix_absAxA except additional void * */
|
|
|
|
me_cmp_func sse[5];
|
|
|
|
me_cmp_func hadamard8_diff[5];
|
|
|
|
me_cmp_func dct_sad[5];
|
|
|
|
me_cmp_func quant_psnr[5];
|
|
|
|
me_cmp_func bit[5];
|
|
|
|
me_cmp_func rd[5];
|
|
|
|
me_cmp_func vsad[5];
|
|
|
|
me_cmp_func vsse[5];
|
2004-06-07 06:23:31 +03:00
|
|
|
me_cmp_func nsse[5];
|
2004-08-22 20:16:03 +03:00
|
|
|
me_cmp_func w53[5];
|
|
|
|
me_cmp_func w97[5];
|
2004-12-08 20:02:56 +02:00
|
|
|
me_cmp_func dct_max[5];
|
2006-01-04 18:31:23 +02:00
|
|
|
me_cmp_func dct264_sad[5];
|
2002-12-28 01:51:46 +02:00
|
|
|
|
2003-12-30 18:07:57 +02:00
|
|
|
me_cmp_func me_pre_cmp[5];
|
|
|
|
me_cmp_func me_cmp[5];
|
|
|
|
me_cmp_func me_sub_cmp[5];
|
|
|
|
me_cmp_func mb_cmp[5];
|
2004-01-06 00:57:07 +02:00
|
|
|
me_cmp_func ildct_cmp[5]; //only width 16 used
|
2004-12-08 20:02:56 +02:00
|
|
|
me_cmp_func frame_skip_cmp[5]; //only width 8 used
|
2002-11-11 11:40:17 +02:00
|
|
|
|
2007-07-09 02:15:00 +03:00
|
|
|
int (*ssd_int8_vs_int16)(const int8_t *pix1, const int16_t *pix2,
|
|
|
|
int size);
|
2007-03-30 22:15:31 +03:00
|
|
|
|
2003-03-15 04:06:09 +02:00
|
|
|
/**
|
|
|
|
* Halfpel motion compensation with rounding (a+b+1)>>1.
|
2007-03-30 23:56:32 +03:00
|
|
|
* this is an array[4][4] of motion compensation functions for 4
|
2004-01-01 20:02:30 +02:00
|
|
|
* horizontal blocksizes (8,16) and the 4 halfpel positions<br>
|
2003-04-25 03:21:53 +03:00
|
|
|
* *pixels_tab[ 0->16xH 1->8xH ][ xhalfpel + 2*yhalfpel ]
|
2003-03-15 04:06:09 +02:00
|
|
|
* @param block destination where the result is stored
|
|
|
|
* @param pixels source
|
|
|
|
* @param line_size number of bytes in a horizontal line of block
|
|
|
|
* @param h height
|
|
|
|
*/
|
2003-05-16 02:30:03 +03:00
|
|
|
op_pixels_func put_pixels_tab[4][4];
|
2003-03-15 04:06:09 +02:00
|
|
|
|
|
|
|
/**
|
|
|
|
* Halfpel motion compensation with rounding (a+b+1)>>1.
|
2005-12-17 20:14:38 +02:00
|
|
|
* This is an array[4][4] of motion compensation functions for 4
|
2004-01-01 20:02:30 +02:00
|
|
|
* horizontal blocksizes (8,16) and the 4 halfpel positions<br>
|
2003-04-25 03:21:53 +03:00
|
|
|
* *pixels_tab[ 0->16xH 1->8xH ][ xhalfpel + 2*yhalfpel ]
|
2003-03-15 04:06:09 +02:00
|
|
|
* @param block destination into which the result is averaged (a+b+1)>>1
|
|
|
|
* @param pixels source
|
|
|
|
* @param line_size number of bytes in a horizontal line of block
|
|
|
|
* @param h height
|
|
|
|
*/
|
2003-06-19 04:44:44 +03:00
|
|
|
op_pixels_func avg_pixels_tab[4][4];
|
2003-03-15 04:06:09 +02:00
|
|
|
|
|
|
|
/**
|
|
|
|
* Halfpel motion compensation with no rounding (a+b)>>1.
|
2007-03-30 23:56:32 +03:00
|
|
|
* this is an array[2][4] of motion compensation functions for 2
|
2003-05-05 13:00:57 +03:00
|
|
|
* horizontal blocksizes (8,16) and the 4 halfpel positions<br>
|
2003-04-25 03:21:53 +03:00
|
|
|
* *pixels_tab[ 0->16xH 1->8xH ][ xhalfpel + 2*yhalfpel ]
|
2003-03-15 04:06:09 +02:00
|
|
|
* @param block destination where the result is stored
|
|
|
|
* @param pixels source
|
|
|
|
* @param line_size number of bytes in a horizontal line of block
|
|
|
|
* @param h height
|
|
|
|
*/
|
2004-06-12 00:12:46 +03:00
|
|
|
op_pixels_func put_no_rnd_pixels_tab[4][4];
|
2003-03-15 04:06:09 +02:00
|
|
|
|
|
|
|
/**
|
|
|
|
* Halfpel motion compensation with no rounding (a+b)>>1.
|
2007-03-30 23:56:32 +03:00
|
|
|
* this is an array[2][4] of motion compensation functions for 2
|
2003-05-05 13:00:57 +03:00
|
|
|
* horizontal blocksizes (8,16) and the 4 halfpel positions<br>
|
2003-04-25 03:21:53 +03:00
|
|
|
* *pixels_tab[ 0->16xH 1->8xH ][ xhalfpel + 2*yhalfpel ]
|
2003-03-15 04:06:09 +02:00
|
|
|
* @param block destination into which the result is averaged (a+b)>>1
|
|
|
|
* @param pixels source
|
|
|
|
* @param line_size number of bytes in a horizontal line of block
|
|
|
|
* @param h height
|
|
|
|
*/
|
2004-06-12 00:12:46 +03:00
|
|
|
op_pixels_func avg_no_rnd_pixels_tab[4][4];
|
2005-12-17 20:14:38 +02:00
|
|
|
|
2004-03-05 20:44:04 +02:00
|
|
|
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);
|
2005-12-17 20:14:38 +02:00
|
|
|
|
2003-05-16 02:30:03 +03:00
|
|
|
/**
|
|
|
|
* Thirdpel motion compensation with rounding (a+b+1)>>1.
|
2007-03-30 23:56:32 +03:00
|
|
|
* this is an array[12] of motion compensation functions for the 9 thirdpe
|
|
|
|
* positions<br>
|
2003-05-16 02:30:03 +03:00
|
|
|
* *pixels_tab[ xthirdpel + 4*ythirdpel ]
|
|
|
|
* @param block destination where the result is stored
|
|
|
|
* @param pixels source
|
|
|
|
* @param line_size number of bytes in a horizontal line of block
|
|
|
|
* @param h height
|
|
|
|
*/
|
|
|
|
tpel_mc_func put_tpel_pixels_tab[11]; //FIXME individual func ptr per width?
|
2003-06-19 04:44:44 +03:00
|
|
|
tpel_mc_func avg_tpel_pixels_tab[11]; //FIXME individual func ptr per width?
|
|
|
|
|
2002-11-11 11:40:17 +02:00
|
|
|
qpel_mc_func put_qpel_pixels_tab[2][16];
|
|
|
|
qpel_mc_func avg_qpel_pixels_tab[2][16];
|
|
|
|
qpel_mc_func put_no_rnd_qpel_pixels_tab[2][16];
|
|
|
|
qpel_mc_func avg_no_rnd_qpel_pixels_tab[2][16];
|
2002-12-28 01:51:46 +02:00
|
|
|
qpel_mc_func put_mspel_pixels_tab[8];
|
2005-12-17 20:14:38 +02:00
|
|
|
|
2003-04-04 17:42:28 +03:00
|
|
|
/**
|
2007-03-30 23:56:32 +03:00
|
|
|
* h264 Chroma MC
|
2003-04-04 17:42:28 +03:00
|
|
|
*/
|
|
|
|
h264_chroma_mc_func put_h264_chroma_pixels_tab[3];
|
2006-09-02 07:52:00 +03:00
|
|
|
/* This is really one func used in VC-1 decoding */
|
|
|
|
h264_chroma_mc_func put_no_rnd_h264_chroma_pixels_tab[3];
|
2003-04-04 17:42:28 +03:00
|
|
|
h264_chroma_mc_func avg_h264_chroma_pixels_tab[3];
|
2002-11-11 11:40:17 +02:00
|
|
|
|
2006-01-08 03:50:34 +02:00
|
|
|
qpel_mc_func put_h264_qpel_pixels_tab[4][16];
|
|
|
|
qpel_mc_func avg_h264_qpel_pixels_tab[4][16];
|
2005-12-17 20:14:38 +02:00
|
|
|
|
2006-10-02 00:25:17 +03:00
|
|
|
qpel_mc_func put_2tap_qpel_pixels_tab[4][16];
|
|
|
|
qpel_mc_func avg_2tap_qpel_pixels_tab[4][16];
|
|
|
|
|
2005-01-07 07:56:07 +02:00
|
|
|
h264_weight_func weight_h264_pixels_tab[10];
|
|
|
|
h264_biweight_func biweight_h264_pixels_tab[10];
|
2005-12-17 20:14:38 +02:00
|
|
|
|
2006-07-03 03:16:45 +03:00
|
|
|
/* AVS specific */
|
|
|
|
qpel_mc_func put_cavs_qpel_pixels_tab[2][16];
|
|
|
|
qpel_mc_func avg_cavs_qpel_pixels_tab[2][16];
|
|
|
|
void (*cavs_filter_lv)(uint8_t *pix, int stride, int alpha, int beta, int tc, int bs1, int bs2);
|
|
|
|
void (*cavs_filter_lh)(uint8_t *pix, int stride, int alpha, int beta, int tc, int bs1, int bs2);
|
|
|
|
void (*cavs_filter_cv)(uint8_t *pix, int stride, int alpha, int beta, int tc, int bs1, int bs2);
|
|
|
|
void (*cavs_filter_ch)(uint8_t *pix, int stride, int alpha, int beta, int tc, int bs1, int bs2);
|
|
|
|
void (*cavs_idct8_add)(uint8_t *dst, DCTELEM *block, int stride);
|
|
|
|
|
2003-12-30 18:07:57 +02:00
|
|
|
me_cmp_func pix_abs[2][4];
|
2005-12-17 20:14:38 +02:00
|
|
|
|
2002-11-14 21:20:04 +02:00
|
|
|
/* huffyuv specific */
|
|
|
|
void (*add_bytes)(uint8_t *dst/*align 16*/, uint8_t *src/*align 16*/, int w);
|
2008-02-21 09:10:46 +02:00
|
|
|
void (*add_bytes_l2)(uint8_t *dst/*align 16*/, uint8_t *src1/*align 16*/, uint8_t *src2/*align 16*/, int w);
|
2002-12-28 01:51:46 +02:00
|
|
|
void (*diff_bytes)(uint8_t *dst/*align 16*/, uint8_t *src1/*align 16*/, uint8_t *src2/*align 1*/,int w);
|
2003-10-13 20:27:30 +03:00
|
|
|
/**
|
|
|
|
* subtract huffyuv's variant of median prediction
|
|
|
|
* note, this might read from src1[-1], src2[-1]
|
|
|
|
*/
|
|
|
|
void (*sub_hfyu_median_prediction)(uint8_t *dst, uint8_t *src1, uint8_t *src2, int w, int *left, int *left_top);
|
2008-02-21 09:10:46 +02:00
|
|
|
/* this might write to dst[w] */
|
|
|
|
void (*add_png_paeth_prediction)(uint8_t *dst, uint8_t *src, uint8_t *top, int w, int bpp);
|
2008-02-01 15:49:15 +02:00
|
|
|
void (*bswap_buf)(uint32_t *dst, const uint32_t *src, int w);
|
2005-04-25 04:01:41 +03:00
|
|
|
|
2008-08-15 03:14:07 +03:00
|
|
|
void (*h264_v_loop_filter_luma)(uint8_t *pix/*align 16*/, int stride, int alpha, int beta, int8_t *tc0);
|
|
|
|
void (*h264_h_loop_filter_luma)(uint8_t *pix/*align 4 */, int stride, int alpha, int beta, int8_t *tc0);
|
|
|
|
/* v/h_loop_filter_luma_intra: align 16 */
|
2008-12-19 02:44:51 +02:00
|
|
|
void (*h264_v_loop_filter_luma_intra)(uint8_t *pix, int stride, int alpha, int beta);
|
|
|
|
void (*h264_h_loop_filter_luma_intra)(uint8_t *pix, int stride, int alpha, int beta);
|
2008-08-15 03:14:07 +03:00
|
|
|
void (*h264_v_loop_filter_chroma)(uint8_t *pix/*align 8*/, int stride, int alpha, int beta, int8_t *tc0);
|
|
|
|
void (*h264_h_loop_filter_chroma)(uint8_t *pix/*align 4*/, int stride, int alpha, int beta, int8_t *tc0);
|
|
|
|
void (*h264_v_loop_filter_chroma_intra)(uint8_t *pix/*align 8*/, int stride, int alpha, int beta);
|
|
|
|
void (*h264_h_loop_filter_chroma_intra)(uint8_t *pix/*align 8*/, int stride, int alpha, int beta);
|
2006-08-28 12:33:01 +03:00
|
|
|
// h264_loop_filter_strength: simd only. the C version is inlined in h264.c
|
|
|
|
void (*h264_loop_filter_strength)(int16_t bS[2][4][4], uint8_t nnz[40], int8_t ref[2][40], int16_t mv[2][40][2],
|
2008-07-20 00:53:54 +03:00
|
|
|
int bidir, int edges, int step, int mask_mv0, int mask_mv1, int field);
|
2005-12-17 20:14:38 +02:00
|
|
|
|
2003-12-01 17:23:14 +02:00
|
|
|
void (*h263_v_loop_filter)(uint8_t *src, int stride, int qscale);
|
|
|
|
void (*h263_h_loop_filter)(uint8_t *src, int stride, int qscale);
|
|
|
|
|
2004-05-31 00:21:13 +03:00
|
|
|
void (*h261_loop_filter)(uint8_t *src, int stride);
|
2004-05-30 23:37:15 +03:00
|
|
|
|
2007-11-09 23:37:48 +02:00
|
|
|
void (*x8_v_loop_filter)(uint8_t *src, int stride, int qscale);
|
|
|
|
void (*x8_h_loop_filter)(uint8_t *src, int stride, int qscale);
|
|
|
|
|
2008-10-04 13:26:17 +03:00
|
|
|
void (*vp3_v_loop_filter)(uint8_t *src, int stride, int *bounding_values);
|
|
|
|
void (*vp3_h_loop_filter)(uint8_t *src, int stride, int *bounding_values);
|
|
|
|
|
2006-08-10 22:06:25 +03:00
|
|
|
/* assume len is a multiple of 4, and arrays are 16-byte aligned */
|
2006-08-03 06:18:47 +03:00
|
|
|
void (*vorbis_inverse_coupling)(float *mag, float *ang, int blocksize);
|
2008-08-14 02:33:48 +03:00
|
|
|
void (*ac3_downmix)(float (*samples)[256], float (*matrix)[2], int out_ch, int in_ch, int len);
|
2007-09-30 01:31:18 +03:00
|
|
|
/* no alignment needed */
|
|
|
|
void (*flac_compute_autocorr)(const int32_t *data, int len, int lag, double *autoc);
|
2006-08-10 22:06:25 +03:00
|
|
|
/* assume len is a multiple of 8, and arrays are 16-byte aligned */
|
2006-08-11 21:19:37 +03:00
|
|
|
void (*vector_fmul)(float *dst, const float *src, int len);
|
2006-08-10 22:06:25 +03:00
|
|
|
void (*vector_fmul_reverse)(float *dst, const float *src0, const float *src1, int len);
|
|
|
|
/* assume len is a multiple of 8, and src arrays are 16-byte aligned */
|
|
|
|
void (*vector_fmul_add_add)(float *dst, const float *src0, const float *src1, const float *src2, int src3, int len, int step);
|
2008-07-13 17:56:01 +03:00
|
|
|
/* assume len is a multiple of 4, and arrays are 16-byte aligned */
|
|
|
|
void (*vector_fmul_window)(float *dst, const float *src0, const float *src1, const float *win, float add_bias, int len);
|
2008-08-14 02:35:40 +03:00
|
|
|
/* assume len is a multiple of 8, and arrays are 16-byte aligned */
|
|
|
|
void (*int32_to_float_fmul_scalar)(float *dst, const int *src, float mul, int len);
|
2006-08-10 22:06:25 +03:00
|
|
|
|
|
|
|
/* C version: convert floats from the range [384.0,386.0] to ints in [-32768,32767]
|
2006-08-31 22:14:00 +03:00
|
|
|
* simd versions: convert floats from [-32768.0,32767.0] without rescaling and arrays are 16byte aligned */
|
2008-07-08 00:04:29 +03:00
|
|
|
void (*float_to_int16)(int16_t *dst, const float *src, long len);
|
2008-07-16 03:50:12 +03:00
|
|
|
void (*float_to_int16_interleave)(int16_t *dst, const float **src, long len, int channels);
|
2006-08-03 06:18:47 +03:00
|
|
|
|
2003-03-03 16:54:00 +02:00
|
|
|
/* (I)DCT */
|
|
|
|
void (*fdct)(DCTELEM *block/* align 16*/);
|
2003-10-24 21:28:01 +03:00
|
|
|
void (*fdct248)(DCTELEM *block/* align 16*/);
|
2005-12-17 20:14:38 +02:00
|
|
|
|
2003-06-21 04:11:31 +03:00
|
|
|
/* IDCT really*/
|
|
|
|
void (*idct)(DCTELEM *block/* align 16*/);
|
2005-12-17 20:14:38 +02:00
|
|
|
|
2003-03-06 12:08:15 +02:00
|
|
|
/**
|
2003-03-06 12:26:19 +02:00
|
|
|
* block -> idct -> clip to unsigned 8 bit -> dest.
|
2003-03-06 12:08:15 +02:00
|
|
|
* (-1392, 0, 0, ...) -> idct -> (-174, -174, ...) -> put -> (0, 0, ...)
|
2007-09-27 14:29:21 +03:00
|
|
|
* @param line_size size in bytes of a horizontal line of dest
|
2003-03-06 12:08:15 +02:00
|
|
|
*/
|
2003-03-03 16:54:00 +02:00
|
|
|
void (*idct_put)(uint8_t *dest/*align 8*/, int line_size, DCTELEM *block/*align 16*/);
|
2005-12-17 20:14:38 +02:00
|
|
|
|
2003-03-06 12:08:15 +02:00
|
|
|
/**
|
|
|
|
* block -> idct -> add dest -> clip to unsigned 8 bit -> dest.
|
2007-09-27 14:29:21 +03:00
|
|
|
* @param line_size size in bytes of a horizontal line of dest
|
2003-03-06 12:08:15 +02:00
|
|
|
*/
|
2003-03-03 16:54:00 +02:00
|
|
|
void (*idct_add)(uint8_t *dest/*align 8*/, int line_size, DCTELEM *block/*align 16*/);
|
2005-12-17 20:14:38 +02:00
|
|
|
|
2003-03-06 12:08:15 +02:00
|
|
|
/**
|
2003-03-06 12:26:19 +02:00
|
|
|
* idct input permutation.
|
2003-05-11 13:19:05 +03:00
|
|
|
* several optimized IDCTs need a permutated input (relative to the normal order of the reference
|
|
|
|
* IDCT)
|
|
|
|
* this permutation must be performed before the idct_put/add, note, normally this can be merged
|
|
|
|
* with the zigzag/alternate scan<br>
|
2003-03-06 12:08:15 +02:00
|
|
|
* an example to avoid confusion:
|
|
|
|
* - (->decode coeffs -> zigzag reorder -> dequant -> reference idct ->...)
|
|
|
|
* - (x -> referece dct -> reference idct -> x)
|
|
|
|
* - (x -> referece dct -> simple_mmx_perm = idct_permutation -> simple_idct_mmx -> x)
|
|
|
|
* - (->decode coeffs -> zigzag reorder -> simple_mmx_perm -> dequant -> simple_idct_mmx ->...)
|
|
|
|
*/
|
2003-03-03 16:54:00 +02:00
|
|
|
uint8_t idct_permutation[64];
|
|
|
|
int idct_permutation_type;
|
|
|
|
#define FF_NO_IDCT_PERM 1
|
|
|
|
#define FF_LIBMPEG2_IDCT_PERM 2
|
|
|
|
#define FF_SIMPLE_IDCT_PERM 3
|
|
|
|
#define FF_TRANSPOSE_IDCT_PERM 4
|
2005-05-17 21:28:40 +03:00
|
|
|
#define FF_PARTTRANS_IDCT_PERM 5
|
2008-04-16 04:34:55 +03:00
|
|
|
#define FF_SSE2_IDCT_PERM 6
|
2003-03-03 16:54:00 +02:00
|
|
|
|
2004-02-06 04:12:37 +02:00
|
|
|
int (*try_8x8basis)(int16_t rem[64], int16_t weight[64], int16_t basis[64], int scale);
|
|
|
|
void (*add_8x8basis)(int16_t rem[64], int16_t basis[64], int scale);
|
|
|
|
#define BASIS_SHIFT 16
|
|
|
|
#define RECON_SHIFT 6
|
2005-12-17 20:14:38 +02:00
|
|
|
|
2008-03-04 02:07:41 +02:00
|
|
|
void (*draw_edges)(uint8_t *buf, int wrap, int width, int height, int w);
|
2008-03-05 02:38:14 +02:00
|
|
|
#define EDGE_WIDTH 16
|
2008-03-04 02:07:41 +02:00
|
|
|
|
2006-12-05 23:57:31 +02:00
|
|
|
/* h264 functions */
|
2008-12-19 11:34:18 +02:00
|
|
|
/* NOTE!!! if you implement any of h264_idct8_add, h264_idct8_add4 then you must implement all of them
|
|
|
|
NOTE!!! if you implement any of h264_idct_add, h264_idct_add16, h264_idct_add16intra, h264_idct_add8 then you must implement all of them
|
|
|
|
The reason for above, is that no 2 out of one list may use a different permutation.
|
|
|
|
*/
|
2008-08-24 07:03:02 +03:00
|
|
|
void (*h264_idct_add)(uint8_t *dst/*align 4*/, DCTELEM *block/*align 16*/, int stride);
|
|
|
|
void (*h264_idct8_add)(uint8_t *dst/*align 8*/, DCTELEM *block/*align 16*/, int stride);
|
|
|
|
void (*h264_idct_dc_add)(uint8_t *dst/*align 4*/, DCTELEM *block/*align 16*/, int stride);
|
|
|
|
void (*h264_idct8_dc_add)(uint8_t *dst/*align 8*/, DCTELEM *block/*align 16*/, int stride);
|
2006-12-07 18:09:33 +02:00
|
|
|
void (*h264_dct)(DCTELEM block[4][4]);
|
2008-12-18 04:36:48 +02:00
|
|
|
void (*h264_idct_add16)(uint8_t *dst/*align 16*/, const int *blockoffset, DCTELEM *block/*align 16*/, int stride, const uint8_t nnzc[6*8]);
|
|
|
|
void (*h264_idct8_add4)(uint8_t *dst/*align 16*/, const int *blockoffset, DCTELEM *block/*align 16*/, int stride, const uint8_t nnzc[6*8]);
|
|
|
|
void (*h264_idct_add8)(uint8_t **dst/*align 16*/, const int *blockoffset, DCTELEM *block/*align 16*/, int stride, const uint8_t nnzc[6*8]);
|
|
|
|
void (*h264_idct_add16intra)(uint8_t *dst/*align 16*/, const int *blockoffset, DCTELEM *block/*align 16*/, int stride, const uint8_t nnzc[6*8]);
|
2006-03-16 21:18:18 +02:00
|
|
|
|
|
|
|
/* snow wavelet */
|
2007-08-25 06:00:51 +03:00
|
|
|
void (*vertical_compose97i)(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2, IDWTELEM *b3, IDWTELEM *b4, IDWTELEM *b5, int width);
|
|
|
|
void (*horizontal_compose97i)(IDWTELEM *b, int width);
|
2007-01-30 12:31:34 +02:00
|
|
|
void (*inner_add_yblock)(const uint8_t *obmc, const int obmc_stride, uint8_t * * block, int b_w, int b_h, int src_x, int src_y, int src_stride, slice_buffer * sb, int add, uint8_t * dst8);
|
2006-03-23 22:16:36 +02:00
|
|
|
|
|
|
|
void (*prefetch)(void *mem, int stride, int h);
|
2006-03-31 23:43:37 +03:00
|
|
|
|
|
|
|
void (*shrink[4])(uint8_t *dst, int dst_wrap, const uint8_t *src, int src_wrap, int width, int height);
|
2006-07-30 06:45:02 +03:00
|
|
|
|
|
|
|
/* vc1 functions */
|
|
|
|
void (*vc1_inv_trans_8x8)(DCTELEM *b);
|
2007-12-08 12:41:18 +02:00
|
|
|
void (*vc1_inv_trans_8x4)(uint8_t *dest, int line_size, DCTELEM *block);
|
|
|
|
void (*vc1_inv_trans_4x8)(uint8_t *dest, int line_size, DCTELEM *block);
|
|
|
|
void (*vc1_inv_trans_4x4)(uint8_t *dest, int line_size, DCTELEM *block);
|
2006-11-26 06:57:31 +02:00
|
|
|
void (*vc1_v_overlap)(uint8_t* src, int stride);
|
|
|
|
void (*vc1_h_overlap)(uint8_t* src, int stride);
|
2006-07-30 06:45:02 +03:00
|
|
|
/* put 8x8 block with bicubic interpolation and quarterpel precision
|
|
|
|
* last argument is actually round value instead of height
|
|
|
|
*/
|
|
|
|
op_pixels_func put_vc1_mspel_pixels_tab[16];
|
2007-11-09 23:37:48 +02:00
|
|
|
|
|
|
|
/* intrax8 functions */
|
2007-11-10 14:24:02 +02:00
|
|
|
void (*x8_spatial_compensation[12])(uint8_t *src , uint8_t *dst, int linesize);
|
|
|
|
void (*x8_setup_spatial_compensation)(uint8_t *src, uint8_t *dst, int linesize,
|
2007-11-09 23:37:48 +02:00
|
|
|
int * range, int * sum, int edges);
|
|
|
|
|
2008-07-06 09:06:55 +03:00
|
|
|
/* ape functions */
|
|
|
|
/**
|
|
|
|
* Add contents of the second vector to the first one.
|
2008-07-10 09:15:13 +03:00
|
|
|
* @param len length of vectors, should be multiple of 16
|
2008-07-06 09:06:55 +03:00
|
|
|
*/
|
|
|
|
void (*add_int16)(int16_t *v1/*align 16*/, int16_t *v2, int len);
|
|
|
|
/**
|
|
|
|
* Add contents of the second vector to the first one.
|
2008-07-10 09:15:13 +03:00
|
|
|
* @param len length of vectors, should be multiple of 16
|
2008-07-06 09:06:55 +03:00
|
|
|
*/
|
|
|
|
void (*sub_int16)(int16_t *v1/*align 16*/, int16_t *v2, int len);
|
|
|
|
/**
|
|
|
|
* Calculate scalar product of two vectors.
|
2008-07-10 09:15:13 +03:00
|
|
|
* @param len length of vectors, should be multiple of 16
|
2008-07-06 09:06:55 +03:00
|
|
|
* @param shift number of bits to discard from product
|
|
|
|
*/
|
|
|
|
int32_t (*scalarproduct_int16)(int16_t *v1, int16_t *v2/*align 16*/, int len, int shift);
|
2008-12-01 08:36:22 +02:00
|
|
|
|
2008-12-01 08:40:36 +02:00
|
|
|
/* rv30 functions */
|
|
|
|
qpel_mc_func put_rv30_tpel_pixels_tab[4][16];
|
|
|
|
qpel_mc_func avg_rv30_tpel_pixels_tab[4][16];
|
|
|
|
|
2008-12-01 08:36:22 +02:00
|
|
|
/* rv40 functions */
|
|
|
|
qpel_mc_func put_rv40_qpel_pixels_tab[4][16];
|
|
|
|
qpel_mc_func avg_rv40_qpel_pixels_tab[4][16];
|
|
|
|
h264_chroma_mc_func put_rv40_chroma_pixels_tab[3];
|
|
|
|
h264_chroma_mc_func avg_rv40_chroma_pixels_tab[3];
|
2002-11-11 11:40:17 +02:00
|
|
|
} DSPContext;
|
|
|
|
|
2006-11-14 05:18:09 +02:00
|
|
|
void dsputil_static_init(void);
|
2003-03-03 16:54:00 +02:00
|
|
|
void dsputil_init(DSPContext* p, AVCodecContext *avctx);
|
2001-07-22 17:18:56 +03:00
|
|
|
|
2006-12-07 20:33:00 +02:00
|
|
|
int ff_check_alignment(void);
|
|
|
|
|
2002-10-23 18:11:07 +03:00
|
|
|
/**
|
|
|
|
* permute block according to permuatation.
|
|
|
|
* @param last last non zero element in scantable order
|
|
|
|
*/
|
2003-02-11 18:35:48 +02:00
|
|
|
void ff_block_permute(DCTELEM *block, uint8_t *permutation, const uint8_t *scantable, int last);
|
2001-08-08 01:43:19 +03:00
|
|
|
|
2004-01-06 00:57:07 +02:00
|
|
|
void ff_set_cmp(DSPContext* c, me_cmp_func *cmp, int type);
|
|
|
|
|
2005-12-22 03:10:11 +02:00
|
|
|
#define BYTE_VEC32(c) ((c)*0x01010101UL)
|
2003-05-15 02:08:01 +03:00
|
|
|
|
|
|
|
static inline uint32_t rnd_avg32(uint32_t a, uint32_t b)
|
|
|
|
{
|
|
|
|
return (a | b) - (((a ^ b) & ~BYTE_VEC32(0x01)) >> 1);
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline uint32_t no_rnd_avg32(uint32_t a, uint32_t b)
|
|
|
|
{
|
|
|
|
return (a & b) + (((a ^ b) & ~BYTE_VEC32(0x01)) >> 1);
|
|
|
|
}
|
|
|
|
|
2004-08-22 20:16:03 +03:00
|
|
|
static inline int get_penalty_factor(int lambda, int lambda2, int type){
|
|
|
|
switch(type&0xFF){
|
|
|
|
default:
|
|
|
|
case FF_CMP_SAD:
|
|
|
|
return lambda>>FF_LAMBDA_SHIFT;
|
|
|
|
case FF_CMP_DCT:
|
|
|
|
return (3*lambda)>>(FF_LAMBDA_SHIFT+1);
|
|
|
|
case FF_CMP_W53:
|
|
|
|
return (4*lambda)>>(FF_LAMBDA_SHIFT);
|
|
|
|
case FF_CMP_W97:
|
|
|
|
return (2*lambda)>>(FF_LAMBDA_SHIFT);
|
|
|
|
case FF_CMP_SATD:
|
2006-01-04 18:31:23 +02:00
|
|
|
case FF_CMP_DCT264:
|
2004-08-22 20:16:03 +03:00
|
|
|
return (2*lambda)>>FF_LAMBDA_SHIFT;
|
|
|
|
case FF_CMP_RD:
|
|
|
|
case FF_CMP_PSNR:
|
|
|
|
case FF_CMP_SSE:
|
|
|
|
case FF_CMP_NSSE:
|
|
|
|
return lambda2>>FF_LAMBDA_SHIFT;
|
|
|
|
case FF_CMP_BIT:
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2003-03-06 12:08:15 +02:00
|
|
|
/**
|
2003-03-06 12:26:19 +02:00
|
|
|
* Empty mmx state.
|
2003-03-06 12:08:15 +02:00
|
|
|
* this must be called between any dsp function and float/double code.
|
|
|
|
* for example sin(); dsp->idct_put(); emms_c(); cos()
|
|
|
|
*/
|
2002-11-11 11:40:17 +02:00
|
|
|
#define emms_c()
|
|
|
|
|
2003-01-11 22:51:03 +02:00
|
|
|
/* should be defined by architectures supporting
|
|
|
|
one or more MultiMedia extension */
|
|
|
|
int mm_support(void);
|
|
|
|
|
2007-06-16 13:44:46 +03:00
|
|
|
void dsputil_init_alpha(DSPContext* c, AVCodecContext *avctx);
|
2008-12-17 02:54:54 +02:00
|
|
|
void dsputil_init_arm(DSPContext* c, AVCodecContext *avctx);
|
2007-06-16 13:44:46 +03:00
|
|
|
void dsputil_init_bfin(DSPContext* c, AVCodecContext *avctx);
|
|
|
|
void dsputil_init_mlib(DSPContext* c, AVCodecContext *avctx);
|
|
|
|
void dsputil_init_mmi(DSPContext* c, AVCodecContext *avctx);
|
|
|
|
void dsputil_init_mmx(DSPContext* c, AVCodecContext *avctx);
|
|
|
|
void dsputil_init_ppc(DSPContext* c, AVCodecContext *avctx);
|
|
|
|
void dsputil_init_sh4(DSPContext* c, AVCodecContext *avctx);
|
|
|
|
void dsputil_init_vis(DSPContext* c, AVCodecContext *avctx);
|
|
|
|
|
2007-05-13 17:01:15 +03:00
|
|
|
#define DECLARE_ALIGNED_16(t, v) DECLARE_ALIGNED(16, t, v)
|
2004-04-25 07:06:09 +03:00
|
|
|
|
2001-08-14 00:45:36 +03:00
|
|
|
#if defined(HAVE_MMX)
|
2001-07-22 17:18:56 +03:00
|
|
|
|
2002-11-12 17:00:04 +02:00
|
|
|
#undef emms_c
|
2002-11-11 11:40:17 +02:00
|
|
|
|
2006-11-14 05:18:09 +02:00
|
|
|
extern int mm_flags;
|
|
|
|
|
2003-02-11 18:35:48 +02:00
|
|
|
void add_pixels_clamped_mmx(const DCTELEM *block, uint8_t *pixels, int line_size);
|
|
|
|
void put_pixels_clamped_mmx(const DCTELEM *block, uint8_t *pixels, int line_size);
|
2004-04-27 06:58:06 +03:00
|
|
|
void put_signed_pixels_clamped_mmx(const DCTELEM *block, uint8_t *pixels, int line_size);
|
2001-07-22 17:18:56 +03:00
|
|
|
|
|
|
|
static inline void emms(void)
|
|
|
|
{
|
2008-10-16 16:34:09 +03:00
|
|
|
__asm__ volatile ("emms;":::"memory");
|
2001-07-24 23:42:03 +03:00
|
|
|
}
|
|
|
|
|
2002-12-28 01:51:46 +02:00
|
|
|
|
2001-07-24 23:42:03 +03:00
|
|
|
#define emms_c() \
|
|
|
|
{\
|
2008-11-03 20:08:00 +02:00
|
|
|
if (mm_flags & FF_MM_MMX)\
|
2001-07-24 23:42:03 +03:00
|
|
|
emms();\
|
2001-07-22 17:18:56 +03:00
|
|
|
}
|
|
|
|
|
2003-03-03 16:54:00 +02:00
|
|
|
void dsputil_init_pix_mmx(DSPContext* c, AVCodecContext *avctx);
|
2003-02-11 18:57:55 +02:00
|
|
|
|
2008-12-17 02:54:54 +02:00
|
|
|
#elif defined(ARCH_ARM)
|
2001-08-14 00:45:36 +03:00
|
|
|
|
2006-11-14 05:18:09 +02:00
|
|
|
extern int mm_flags;
|
|
|
|
|
2008-08-25 04:59:06 +03:00
|
|
|
#ifdef HAVE_NEON
|
|
|
|
# define DECLARE_ALIGNED_8(t, v) DECLARE_ALIGNED(16, t, v)
|
|
|
|
# define STRIDE_ALIGN 16
|
|
|
|
#endif
|
|
|
|
|
2002-08-28 16:14:36 +03:00
|
|
|
#elif defined(ARCH_POWERPC)
|
|
|
|
|
2006-11-14 05:18:09 +02:00
|
|
|
extern int mm_flags;
|
|
|
|
|
2007-05-13 17:01:15 +03:00
|
|
|
#define DECLARE_ALIGNED_8(t, v) DECLARE_ALIGNED(16, t, v)
|
2004-10-26 15:47:56 +03:00
|
|
|
#define STRIDE_ALIGN 16
|
2002-08-28 16:14:36 +03:00
|
|
|
|
2002-09-23 11:44:24 +03:00
|
|
|
#elif defined(HAVE_MMI)
|
|
|
|
|
2007-05-13 17:01:15 +03:00
|
|
|
#define DECLARE_ALIGNED_8(t, v) DECLARE_ALIGNED(16, t, v)
|
2004-10-26 15:47:56 +03:00
|
|
|
#define STRIDE_ALIGN 16
|
2002-09-23 11:44:24 +03:00
|
|
|
|
2008-06-25 02:51:17 +03:00
|
|
|
#else
|
|
|
|
|
|
|
|
#define mm_flags 0
|
|
|
|
#define mm_support() 0
|
|
|
|
|
2008-02-16 21:18:30 +02:00
|
|
|
#endif
|
2001-07-22 17:18:56 +03:00
|
|
|
|
2008-02-16 21:18:30 +02:00
|
|
|
#ifndef DECLARE_ALIGNED_8
|
|
|
|
# define DECLARE_ALIGNED_8(t, v) DECLARE_ALIGNED(8, t, v)
|
|
|
|
#endif
|
2001-07-22 17:18:56 +03:00
|
|
|
|
2008-02-16 21:18:30 +02:00
|
|
|
#ifndef STRIDE_ALIGN
|
|
|
|
# define STRIDE_ALIGN 8
|
2001-07-22 17:18:56 +03:00
|
|
|
#endif
|
|
|
|
|
2002-02-27 00:14:27 +02:00
|
|
|
/* PSNR */
|
2003-02-11 18:35:48 +02:00
|
|
|
void get_psnr(uint8_t *orig_image[3], uint8_t *coded_image[3],
|
2002-02-27 00:14:27 +02:00
|
|
|
int orig_linesize[3], int coded_linesize,
|
|
|
|
AVCodecContext *avctx);
|
2002-10-28 02:34:08 +02:00
|
|
|
|
|
|
|
/* FFT computation */
|
|
|
|
|
|
|
|
/* NOTE: soon integer code will be added, so you must use the
|
|
|
|
FFTSample type */
|
|
|
|
typedef float FFTSample;
|
|
|
|
|
2006-08-08 07:01:04 +03:00
|
|
|
struct MDCTContext;
|
|
|
|
|
2002-10-28 02:34:08 +02:00
|
|
|
typedef struct FFTComplex {
|
|
|
|
FFTSample re, im;
|
|
|
|
} FFTComplex;
|
|
|
|
|
|
|
|
typedef struct FFTContext {
|
|
|
|
int nbits;
|
|
|
|
int inverse;
|
|
|
|
uint16_t *revtab;
|
|
|
|
FFTComplex *exptab;
|
|
|
|
FFTComplex *exptab1; /* only used by SSE code */
|
2008-08-12 03:26:58 +03:00
|
|
|
FFTComplex *tmp_buf;
|
|
|
|
void (*fft_permute)(struct FFTContext *s, FFTComplex *z);
|
2002-10-28 02:34:08 +02:00
|
|
|
void (*fft_calc)(struct FFTContext *s, FFTComplex *z);
|
2008-08-12 03:36:36 +03:00
|
|
|
void (*imdct_calc)(struct MDCTContext *s, FFTSample *output, const FFTSample *input);
|
|
|
|
void (*imdct_half)(struct MDCTContext *s, FFTSample *output, const FFTSample *input);
|
2002-10-28 02:34:08 +02:00
|
|
|
} FFTContext;
|
|
|
|
|
2004-03-13 23:43:24 +02:00
|
|
|
int ff_fft_init(FFTContext *s, int nbits, int inverse);
|
2008-08-12 03:26:58 +03:00
|
|
|
void ff_fft_permute_c(FFTContext *s, FFTComplex *z);
|
|
|
|
void ff_fft_permute_sse(FFTContext *s, FFTComplex *z);
|
2004-03-13 23:43:24 +02:00
|
|
|
void ff_fft_calc_c(FFTContext *s, FFTComplex *z);
|
|
|
|
void ff_fft_calc_sse(FFTContext *s, FFTComplex *z);
|
2006-03-08 06:13:55 +02:00
|
|
|
void ff_fft_calc_3dn(FFTContext *s, FFTComplex *z);
|
|
|
|
void ff_fft_calc_3dn2(FFTContext *s, FFTComplex *z);
|
2004-03-13 23:43:24 +02:00
|
|
|
void ff_fft_calc_altivec(FFTContext *s, FFTComplex *z);
|
2003-01-07 20:15:48 +02:00
|
|
|
|
2008-08-12 03:26:58 +03:00
|
|
|
static inline void ff_fft_permute(FFTContext *s, FFTComplex *z)
|
|
|
|
{
|
|
|
|
s->fft_permute(s, z);
|
|
|
|
}
|
2004-03-13 23:43:24 +02:00
|
|
|
static inline void ff_fft_calc(FFTContext *s, FFTComplex *z)
|
2002-10-28 02:34:08 +02:00
|
|
|
{
|
|
|
|
s->fft_calc(s, z);
|
|
|
|
}
|
2004-03-13 23:43:24 +02:00
|
|
|
void ff_fft_end(FFTContext *s);
|
2002-10-28 02:34:08 +02:00
|
|
|
|
|
|
|
/* MDCT computation */
|
|
|
|
|
|
|
|
typedef struct MDCTContext {
|
|
|
|
int n; /* size of MDCT (i.e. number of input data * 2) */
|
|
|
|
int nbits; /* n = 2^nbits */
|
|
|
|
/* pre/post rotation tables */
|
|
|
|
FFTSample *tcos;
|
|
|
|
FFTSample *tsin;
|
|
|
|
FFTContext fft;
|
|
|
|
} MDCTContext;
|
|
|
|
|
2008-08-12 03:38:30 +03:00
|
|
|
static inline void ff_imdct_calc(MDCTContext *s, FFTSample *output, const FFTSample *input)
|
|
|
|
{
|
|
|
|
s->fft.imdct_calc(s, output, input);
|
|
|
|
}
|
|
|
|
static inline void ff_imdct_half(MDCTContext *s, FFTSample *output, const FFTSample *input)
|
|
|
|
{
|
|
|
|
s->fft.imdct_half(s, output, input);
|
|
|
|
}
|
|
|
|
|
2008-01-12 13:11:19 +02:00
|
|
|
/**
|
|
|
|
* Generate a Kaiser-Bessel Derived Window.
|
|
|
|
* @param window pointer to half window
|
2008-01-13 13:02:08 +02:00
|
|
|
* @param alpha determines window shape
|
|
|
|
* @param n size of half window
|
2008-01-12 13:11:19 +02:00
|
|
|
*/
|
2008-01-13 13:02:08 +02:00
|
|
|
void ff_kbd_window_init(float *window, float alpha, int n);
|
2008-01-12 13:11:19 +02:00
|
|
|
|
2008-06-22 18:12:27 +03:00
|
|
|
/**
|
|
|
|
* Generate a sine window.
|
|
|
|
* @param window pointer to half window
|
|
|
|
* @param n size of half window
|
|
|
|
*/
|
|
|
|
void ff_sine_window_init(float *window, int n);
|
2008-08-14 17:40:16 +03:00
|
|
|
extern float ff_sine_128 [ 128];
|
|
|
|
extern float ff_sine_256 [ 256];
|
|
|
|
extern float ff_sine_512 [ 512];
|
|
|
|
extern float ff_sine_1024[1024];
|
|
|
|
extern float ff_sine_2048[2048];
|
|
|
|
extern float *ff_sine_windows[5];
|
2008-06-22 18:12:27 +03:00
|
|
|
|
2002-10-30 00:26:01 +02:00
|
|
|
int ff_mdct_init(MDCTContext *s, int nbits, int inverse);
|
2008-08-12 03:38:30 +03:00
|
|
|
void ff_imdct_calc_c(MDCTContext *s, FFTSample *output, const FFTSample *input);
|
|
|
|
void ff_imdct_half_c(MDCTContext *s, FFTSample *output, const FFTSample *input);
|
2008-08-12 03:36:36 +03:00
|
|
|
void ff_imdct_calc_3dn(MDCTContext *s, FFTSample *output, const FFTSample *input);
|
2008-08-12 03:33:34 +03:00
|
|
|
void ff_imdct_half_3dn(MDCTContext *s, FFTSample *output, const FFTSample *input);
|
2008-08-12 03:36:36 +03:00
|
|
|
void ff_imdct_calc_3dn2(MDCTContext *s, FFTSample *output, const FFTSample *input);
|
2008-08-12 03:33:34 +03:00
|
|
|
void ff_imdct_half_3dn2(MDCTContext *s, FFTSample *output, const FFTSample *input);
|
2008-08-12 03:36:36 +03:00
|
|
|
void ff_imdct_calc_sse(MDCTContext *s, FFTSample *output, const FFTSample *input);
|
2008-08-12 03:33:34 +03:00
|
|
|
void ff_imdct_half_sse(MDCTContext *s, FFTSample *output, const FFTSample *input);
|
2008-08-12 03:36:36 +03:00
|
|
|
void ff_mdct_calc(MDCTContext *s, FFTSample *out, const FFTSample *input);
|
2002-10-30 00:26:01 +02:00
|
|
|
void ff_mdct_end(MDCTContext *s);
|
2002-10-28 02:34:08 +02:00
|
|
|
|
2007-12-21 14:38:41 +02:00
|
|
|
#define WRAPPER8_16(name8, name16)\
|
2003-12-30 18:07:57 +02:00
|
|
|
static int name16(void /*MpegEncContext*/ *s, uint8_t *dst, uint8_t *src, int stride, int h){\
|
|
|
|
return name8(s, dst , src , stride, h)\
|
|
|
|
+name8(s, dst+8 , src+8 , stride, h);\
|
|
|
|
}
|
|
|
|
|
2007-12-21 14:38:41 +02:00
|
|
|
#define WRAPPER8_16_SQ(name8, name16)\
|
2003-12-30 18:07:57 +02:00
|
|
|
static int name16(void /*MpegEncContext*/ *s, uint8_t *dst, uint8_t *src, int stride, int h){\
|
|
|
|
int score=0;\
|
|
|
|
score +=name8(s, dst , src , stride, 8);\
|
|
|
|
score +=name8(s, dst+8 , src+8 , stride, 8);\
|
|
|
|
if(h==16){\
|
|
|
|
dst += 8*stride;\
|
|
|
|
src += 8*stride;\
|
|
|
|
score +=name8(s, dst , src , stride, 8);\
|
|
|
|
score +=name8(s, dst+8 , src+8 , stride, 8);\
|
|
|
|
}\
|
|
|
|
return score;\
|
2002-12-28 01:51:46 +02:00
|
|
|
}
|
|
|
|
|
2006-11-26 20:25:07 +02:00
|
|
|
|
|
|
|
static inline void copy_block2(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h)
|
|
|
|
{
|
|
|
|
int i;
|
|
|
|
for(i=0; i<h; i++)
|
|
|
|
{
|
2007-08-09 20:01:15 +03:00
|
|
|
AV_WN16(dst , AV_RN16(src ));
|
2006-11-26 20:25:07 +02:00
|
|
|
dst+=dstStride;
|
|
|
|
src+=srcStride;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline void copy_block4(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h)
|
|
|
|
{
|
|
|
|
int i;
|
|
|
|
for(i=0; i<h; i++)
|
|
|
|
{
|
2007-08-09 20:01:15 +03:00
|
|
|
AV_WN32(dst , AV_RN32(src ));
|
2006-11-26 20:25:07 +02:00
|
|
|
dst+=dstStride;
|
|
|
|
src+=srcStride;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline void copy_block8(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h)
|
|
|
|
{
|
|
|
|
int i;
|
|
|
|
for(i=0; i<h; i++)
|
|
|
|
{
|
2007-08-09 20:01:15 +03:00
|
|
|
AV_WN32(dst , AV_RN32(src ));
|
|
|
|
AV_WN32(dst+4 , AV_RN32(src+4 ));
|
2006-11-26 20:25:07 +02:00
|
|
|
dst+=dstStride;
|
|
|
|
src+=srcStride;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline void copy_block9(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h)
|
|
|
|
{
|
|
|
|
int i;
|
|
|
|
for(i=0; i<h; i++)
|
|
|
|
{
|
2007-08-09 20:01:15 +03:00
|
|
|
AV_WN32(dst , AV_RN32(src ));
|
|
|
|
AV_WN32(dst+4 , AV_RN32(src+4 ));
|
2006-11-26 20:25:07 +02:00
|
|
|
dst[8]= src[8];
|
|
|
|
dst+=dstStride;
|
|
|
|
src+=srcStride;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline void copy_block16(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h)
|
|
|
|
{
|
|
|
|
int i;
|
|
|
|
for(i=0; i<h; i++)
|
|
|
|
{
|
2007-08-09 20:01:15 +03:00
|
|
|
AV_WN32(dst , AV_RN32(src ));
|
|
|
|
AV_WN32(dst+4 , AV_RN32(src+4 ));
|
|
|
|
AV_WN32(dst+8 , AV_RN32(src+8 ));
|
|
|
|
AV_WN32(dst+12, AV_RN32(src+12));
|
2006-11-26 20:25:07 +02:00
|
|
|
dst+=dstStride;
|
|
|
|
src+=srcStride;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline void copy_block17(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h)
|
|
|
|
{
|
|
|
|
int i;
|
|
|
|
for(i=0; i<h; i++)
|
|
|
|
{
|
2007-08-09 20:01:15 +03:00
|
|
|
AV_WN32(dst , AV_RN32(src ));
|
|
|
|
AV_WN32(dst+4 , AV_RN32(src+4 ));
|
|
|
|
AV_WN32(dst+8 , AV_RN32(src+8 ));
|
|
|
|
AV_WN32(dst+12, AV_RN32(src+12));
|
2006-11-26 20:25:07 +02:00
|
|
|
dst[16]= src[16];
|
|
|
|
dst+=dstStride;
|
|
|
|
src+=srcStride;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2008-08-31 10:39:47 +03:00
|
|
|
#endif /* AVCODEC_DSPUTIL_H */
|