2002-05-26 00:45:33 +02:00
|
|
|
/*
|
|
|
|
* DSP utils
|
2009-01-19 16:46:40 +01:00
|
|
|
* Copyright (c) 2000, 2001, 2002 Fabrice Bellard
|
2004-01-10 17:04:55 +01:00
|
|
|
* Copyright (c) 2002-2004 Michael Niedermayer <michaelni@gmx.at>
|
2002-05-26 00:45:33 +02:00
|
|
|
*
|
2006-10-07 17:30:46 +02:00
|
|
|
* This file is part of FFmpeg.
|
|
|
|
*
|
|
|
|
* FFmpeg is free software; you can redistribute it and/or
|
2002-05-26 00:45:33 +02:00
|
|
|
* modify it under the terms of the GNU Lesser General Public
|
|
|
|
* License as published by the Free Software Foundation; either
|
2006-10-07 17:30:46 +02:00
|
|
|
* version 2.1 of the License, or (at your option) any later version.
|
2002-05-26 00:45:33 +02:00
|
|
|
*
|
2006-10-07 17:30:46 +02:00
|
|
|
* FFmpeg is distributed in the hope that it will be useful,
|
2002-05-26 00:45:33 +02:00
|
|
|
* 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 17:30:46 +02:00
|
|
|
* License along with FFmpeg; if not, write to the Free Software
|
2006-01-12 23:43:26 +01:00
|
|
|
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
|
2002-05-26 00:45:33 +02:00
|
|
|
*/
|
2003-03-06 11:08:15 +01:00
|
|
|
|
|
|
|
/**
|
2010-04-20 16:45:34 +02:00
|
|
|
* @file
|
2003-03-06 12:32:04 +01:00
|
|
|
* DSP utils.
|
2003-04-25 02:21:53 +02:00
|
|
|
* 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 11:08:15 +01:00
|
|
|
*/
|
|
|
|
|
2008-08-31 09:39:47 +02:00
|
|
|
#ifndef AVCODEC_DSPUTIL_H
|
|
|
|
#define AVCODEC_DSPUTIL_H
|
2001-07-22 16:18:56 +02:00
|
|
|
|
2009-01-11 23:19:48 +01:00
|
|
|
#include "libavutil/intreadwrite.h"
|
2002-02-26 23:14:27 +01:00
|
|
|
#include "avcodec.h"
|
2001-07-22 16:18:56 +02:00
|
|
|
|
2003-03-06 11:08:15 +01:00
|
|
|
|
2002-03-09 14:01:16 +01:00
|
|
|
//#define DEBUG
|
2001-07-22 16:18:56 +02:00
|
|
|
/* dct code */
|
|
|
|
typedef short DCTELEM;
|
|
|
|
|
2002-06-04 14:58:40 +02:00
|
|
|
void fdct_ifast (DCTELEM *data);
|
2003-10-25 02:32:54 +02:00
|
|
|
void fdct_ifast248 (DCTELEM *data);
|
2002-08-30 01:55:32 +02:00
|
|
|
void ff_jpeg_fdct_islow (DCTELEM *data);
|
2003-10-24 20:28:01 +02:00
|
|
|
void ff_fdct248_islow (DCTELEM *data);
|
2001-07-22 16:18:56 +02:00
|
|
|
|
|
|
|
void j_rev_dct (DCTELEM *data);
|
2004-09-26 01:18:58 +02:00
|
|
|
void j_rev_dct4 (DCTELEM *data);
|
2004-09-26 02:18:12 +02:00
|
|
|
void j_rev_dct2 (DCTELEM *data);
|
2004-09-26 12:01:40 +02:00
|
|
|
void j_rev_dct1 (DCTELEM *data);
|
2007-11-09 22:37:48 +01:00
|
|
|
void ff_wmv2_idct_c(DCTELEM *data);
|
2001-07-22 16:18:56 +02:00
|
|
|
|
2002-09-20 00:14:53 +02:00
|
|
|
void ff_fdct_mmx(DCTELEM *block);
|
2003-10-23 15:22:20 +02:00
|
|
|
void ff_fdct_mmx2(DCTELEM *block);
|
2004-01-29 02:15:25 +01:00
|
|
|
void ff_fdct_sse2(DCTELEM *block);
|
2001-07-22 16:18:56 +02:00
|
|
|
|
2011-03-29 17:48:59 +02:00
|
|
|
#define H264_IDCT(depth) \
|
|
|
|
void ff_h264_idct8_add_ ## depth ## _c(uint8_t *dst, DCTELEM *block, int stride);\
|
|
|
|
void ff_h264_idct_add_ ## depth ## _c(uint8_t *dst, DCTELEM *block, int stride);\
|
|
|
|
void ff_h264_idct8_dc_add_ ## depth ## _c(uint8_t *dst, DCTELEM *block, int stride);\
|
|
|
|
void ff_h264_idct_dc_add_ ## depth ## _c(uint8_t *dst, DCTELEM *block, int stride);\
|
|
|
|
void ff_h264_lowres_idct_add_ ## depth ## _c(uint8_t *dst, int stride, DCTELEM *block);\
|
|
|
|
void ff_h264_lowres_idct_put_ ## depth ## _c(uint8_t *dst, int stride, DCTELEM *block);\
|
|
|
|
void ff_h264_idct_add16_ ## depth ## _c(uint8_t *dst, const int *blockoffset, DCTELEM *block, int stride, const uint8_t nnzc[6*8]);\
|
|
|
|
void ff_h264_idct_add16intra_ ## depth ## _c(uint8_t *dst, const int *blockoffset, DCTELEM *block, int stride, const uint8_t nnzc[6*8]);\
|
|
|
|
void ff_h264_idct8_add4_ ## depth ## _c(uint8_t *dst, const int *blockoffset, DCTELEM *block, int stride, const uint8_t nnzc[6*8]);\
|
|
|
|
void ff_h264_idct_add8_ ## depth ## _c(uint8_t **dest, const int *blockoffset, DCTELEM *block, int stride, const uint8_t nnzc[6*8]);\
|
|
|
|
void ff_h264_luma_dc_dequant_idct_ ## depth ## _c(DCTELEM *output, DCTELEM *input, int qmul);\
|
|
|
|
void ff_h264_chroma_dc_dequant_idct_ ## depth ## _c(DCTELEM *block, int qmul);
|
|
|
|
|
|
|
|
H264_IDCT( 8)
|
|
|
|
H264_IDCT( 9)
|
|
|
|
H264_IDCT(10)
|
|
|
|
|
2011-01-15 01:41:18 +01:00
|
|
|
void ff_svq3_luma_dc_dequant_idct_c(DCTELEM *output, DCTELEM *input, int qp);
|
2011-01-14 22:34:25 +01:00
|
|
|
void ff_svq3_add_idct_c(uint8_t *dst, DCTELEM *block, int stride, int qp, int dc);
|
|
|
|
|
2001-08-08 00:43:19 +02:00
|
|
|
/* encoding scans */
|
2003-02-11 17:35:48 +01:00
|
|
|
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 20:28:01 +02:00
|
|
|
extern const uint8_t ff_zigzag248_direct[64];
|
2002-01-10 01:51:45 +01:00
|
|
|
|
2001-07-22 16:18:56 +02:00
|
|
|
/* pixel operations */
|
2004-06-21 03:01:44 +02:00
|
|
|
#define MAX_NEG_CROP 1024
|
2001-07-22 16:18:56 +02:00
|
|
|
|
|
|
|
/* temporary */
|
2006-11-12 22:48:04 +01:00
|
|
|
extern uint32_t ff_squareTbl[512];
|
2006-11-12 21:08:09 +01:00
|
|
|
extern uint8_t ff_cropTbl[256 + 2 * MAX_NEG_CROP];
|
2001-07-22 16:18:56 +02:00
|
|
|
|
2011-03-29 17:48:59 +02:00
|
|
|
#define PUTAVG_PIXELS(depth)\
|
|
|
|
void ff_put_pixels8x8_ ## depth ## _c(uint8_t *dst, uint8_t *src, int stride);\
|
|
|
|
void ff_avg_pixels8x8_ ## depth ## _c(uint8_t *dst, uint8_t *src, int stride);\
|
|
|
|
void ff_put_pixels16x16_ ## depth ## _c(uint8_t *dst, uint8_t *src, int stride);\
|
|
|
|
void ff_avg_pixels16x16_ ## depth ## _c(uint8_t *dst, uint8_t *src, int stride);
|
|
|
|
|
|
|
|
PUTAVG_PIXELS( 8)
|
|
|
|
PUTAVG_PIXELS( 9)
|
|
|
|
PUTAVG_PIXELS(10)
|
|
|
|
|
|
|
|
#define ff_put_pixels8x8_c ff_put_pixels8x8_8_c
|
|
|
|
#define ff_avg_pixels8x8_c ff_avg_pixels8x8_8_c
|
|
|
|
#define ff_put_pixels16x16_c ff_put_pixels16x16_8_c
|
|
|
|
#define ff_avg_pixels16x16_c ff_avg_pixels16x16_8_c
|
2010-08-25 22:14:21 +02:00
|
|
|
|
2004-03-08 03:13:45 +01:00
|
|
|
/* VP3 DSP functions */
|
2005-05-17 13:12:04 +02:00
|
|
|
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*/);
|
2010-04-17 04:04:30 +02:00
|
|
|
void ff_vp3_idct_dc_add_c(uint8_t *dest/*align 8*/, int line_size, const DCTELEM *block/*align 16*/);
|
2001-07-22 16:18:56 +02:00
|
|
|
|
2008-10-04 12:26:17 +02:00
|
|
|
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);
|
|
|
|
|
2010-03-06 23:36:59 +01:00
|
|
|
/* Bink functions */
|
|
|
|
void ff_bink_idct_c (DCTELEM *block);
|
|
|
|
void ff_bink_idct_add_c(uint8_t *dest, int linesize, DCTELEM *block);
|
|
|
|
void ff_bink_idct_put_c(uint8_t *dest, int linesize, DCTELEM *block);
|
|
|
|
|
|
|
|
/* EA functions */
|
|
|
|
void ff_ea_idct_put_c(uint8_t *dest, int linesize, DCTELEM *block);
|
|
|
|
|
2006-03-31 22:43:37 +02:00
|
|
|
/* 1/2^n downscaling functions from imgconvert.c */
|
2010-09-07 23:23:45 +02:00
|
|
|
#if LIBAVCODEC_VERSION_MAJOR < 53
|
|
|
|
/**
|
|
|
|
* @deprecated Use av_image_copy_plane() instead.
|
|
|
|
*/
|
|
|
|
attribute_deprecated
|
2006-03-31 22:43:37 +02:00
|
|
|
void ff_img_copy_plane(uint8_t *dst, int dst_wrap, const uint8_t *src, int src_wrap, int width, int height);
|
2010-09-07 23:23:45 +02:00
|
|
|
#endif
|
|
|
|
|
2006-03-31 22:43:37 +02:00
|
|
|
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 11:23:45 +02:00
|
|
|
|
|
|
|
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 22:43:37 +02:00
|
|
|
|
2002-09-14 21:00:09 +02:00
|
|
|
/* minimum alignment rules ;)
|
2008-03-10 19:42:09 +01:00
|
|
|
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 00:58:46 +01:00
|
|
|
I (Michael) did not check them, these are just the alignments which I think
|
2008-03-10 19:42:09 +01:00
|
|
|
could be reached easily ...
|
2001-07-22 16:18:56 +02:00
|
|
|
|
2002-09-14 21:00:09 +02:00
|
|
|
!future video codecs might need functions with less strict alignment
|
|
|
|
*/
|
|
|
|
|
2002-11-11 10:40:17 +01:00
|
|
|
/*
|
2003-02-11 17:35:48 +01:00
|
|
|
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 14:41:04 +01:00
|
|
|
void clear_blocks_c(DCTELEM *blocks);
|
2002-11-11 10:40:17 +01:00
|
|
|
*/
|
2001-07-22 16:18:56 +02:00
|
|
|
|
|
|
|
/* add and put pixel (decoding) */
|
2002-09-14 21:00:09 +02:00
|
|
|
// blocksizes for op_pixels_func are 8x4,8x8 16x8 16x16
|
2003-12-30 20:19:45 +01:00
|
|
|
//h for op_pixels_func is limited to {width/2, width} but never larger than 16 and never smaller then 4
|
2003-02-11 17:35:48 +01:00
|
|
|
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 01:30:03 +02:00
|
|
|
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 17:35:48 +01:00
|
|
|
typedef void (*qpel_mc_func)(uint8_t *dst/*align width (8 or 16)*/, uint8_t *src/*align 1*/, int stride);
|
2003-04-04 16:42:28 +02:00
|
|
|
typedef void (*h264_chroma_mc_func)(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int srcStride, int h, int x, int y);
|
2002-09-11 14:39:53 +02:00
|
|
|
|
2010-02-21 14:28:46 +01:00
|
|
|
typedef void (*op_fill_func)(uint8_t *block/*align width (8 or 16)*/, uint8_t value, int line_size, int h);
|
|
|
|
|
2003-01-09 21:42:37 +01:00
|
|
|
#define DEF_OLD_QPEL(name)\
|
2003-02-11 17:35:48 +01:00
|
|
|
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 21:42:37 +01:00
|
|
|
|
|
|
|
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 14:39:53 +02:00
|
|
|
|
|
|
|
#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 14:01:16 +01:00
|
|
|
|
2001-07-22 16:18:56 +02:00
|
|
|
/* motion estimation */
|
2003-12-30 20:19:45 +01:00
|
|
|
// h is limited to {width/2, width, 2*width} but never larger than 16 and never smaller then 2
|
2007-06-12 11:29:25 +02:00
|
|
|
// although currently h<4 is not used as functions with width <8 are neither used nor implemented
|
2003-12-30 17:07:57 +01:00
|
|
|
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 00:51:46 +01:00
|
|
|
|
2008-03-04 11:26:56 +01:00
|
|
|
/**
|
|
|
|
* Scantable.
|
|
|
|
*/
|
|
|
|
typedef struct ScanTable{
|
|
|
|
const uint8_t *scantable;
|
|
|
|
uint8_t permutated[64];
|
|
|
|
uint8_t raster_end[64];
|
2009-01-14 00:44:16 +01:00
|
|
|
#if ARCH_PPC
|
2008-03-04 11:26:56 +01:00
|
|
|
/** Used by dct_quantize_altivec to find last-non-zero */
|
2010-01-22 04:25:11 +01:00
|
|
|
DECLARE_ALIGNED(16, uint8_t, inverse)[64];
|
2008-03-04 11:26:56 +01:00
|
|
|
#endif
|
|
|
|
} ScanTable;
|
|
|
|
|
|
|
|
void ff_init_scantable(uint8_t *, ScanTable *st, const uint8_t *src_scantable);
|
|
|
|
|
2011-03-29 17:48:59 +02:00
|
|
|
#define EMULATED_EDGE(depth) \
|
|
|
|
void ff_emulated_edge_mc_ ## depth (uint8_t *buf, const uint8_t *src, int linesize,\
|
|
|
|
int block_w, int block_h,\
|
2008-03-05 00:10:47 +01:00
|
|
|
int src_x, int src_y, int w, int h);
|
|
|
|
|
2011-03-29 17:48:59 +02:00
|
|
|
EMULATED_EDGE(8)
|
|
|
|
EMULATED_EDGE(9)
|
|
|
|
EMULATED_EDGE(10)
|
|
|
|
|
|
|
|
#define ff_emulated_edge_mc ff_emulated_edge_mc_8
|
|
|
|
|
2011-02-20 23:31:33 +01:00
|
|
|
void ff_add_pixels_clamped_c(const DCTELEM *block, uint8_t *dest, int linesize);
|
|
|
|
void ff_put_pixels_clamped_c(const DCTELEM *block, uint8_t *dest, int linesize);
|
|
|
|
void ff_put_signed_pixels_clamped_c(const DCTELEM *block, uint8_t *dest, int linesize);
|
|
|
|
|
2003-03-06 11:08:15 +01:00
|
|
|
/**
|
|
|
|
* DSPContext.
|
|
|
|
*/
|
2002-11-11 10:40:17 +01:00
|
|
|
typedef struct DSPContext {
|
|
|
|
/* pixel ops : interface with DCT */
|
2003-02-11 17:35:48 +01:00
|
|
|
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 05:58:06 +02:00
|
|
|
void (*put_signed_pixels_clamped)(const DCTELEM *block/*align 16*/, uint8_t *pixels/*align 8*/, int line_size);
|
2010-02-21 14:28:46 +01:00
|
|
|
void (*put_pixels_nonclamped)(const DCTELEM *block/*align 16*/, uint8_t *pixels/*align 8*/, int line_size);
|
2003-02-11 17:35:48 +01:00
|
|
|
void (*add_pixels_clamped)(const DCTELEM *block/*align 16*/, uint8_t *pixels/*align 8*/, int line_size);
|
2005-06-17 10:24:35 +02:00
|
|
|
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 04:41:25 +02:00
|
|
|
int (*sum_abs_dctelem)(DCTELEM *block/*align 16*/);
|
2011-01-27 04:35:56 +01:00
|
|
|
/**
|
|
|
|
* Motion estimation with emulated edge values.
|
|
|
|
* @param buf pointer to destination buffer (unaligned)
|
|
|
|
* @param src pointer to pixel source (unaligned)
|
|
|
|
* @param linesize width (in pixels) for src/buf
|
|
|
|
* @param block_w number of pixels (per row) to copy to buf
|
|
|
|
* @param block_h nummber of pixel rows to copy to buf
|
|
|
|
* @param src_x offset of src to start of row - this may be negative
|
|
|
|
* @param src_y offset of src to top of image - this may be negative
|
|
|
|
* @param w width of src in pixels
|
|
|
|
* @param h height of src in pixels
|
|
|
|
*/
|
|
|
|
void (*emulated_edge_mc)(uint8_t *buf, const uint8_t *src, int linesize,
|
|
|
|
int block_w, int block_h,
|
|
|
|
int src_x, int src_y, int w, int h);
|
2003-03-15 03:06:09 +01:00
|
|
|
/**
|
|
|
|
* translational global motion compensation.
|
|
|
|
*/
|
2003-02-11 17:35:48 +01:00
|
|
|
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 03:06:09 +01:00
|
|
|
/**
|
|
|
|
* global motion compensation.
|
|
|
|
*/
|
2003-02-11 17:35:48 +01:00
|
|
|
void (*gmc )(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int stride, int h, int ox, int oy,
|
2005-12-22 02:10:11 +01:00
|
|
|
int dxx, int dxy, int dyx, int dyy, int shift, int r, int width, int height);
|
2008-12-10 22:35:17 +01:00
|
|
|
void (*clear_block)(DCTELEM *block/*align 16*/);
|
2002-11-11 10:40:17 +01:00
|
|
|
void (*clear_blocks)(DCTELEM *blocks/*align 16*/);
|
2003-02-11 17:35:48 +01:00
|
|
|
int (*pix_sum)(uint8_t * pix, int line_size);
|
|
|
|
int (*pix_norm1)(uint8_t * pix, int line_size);
|
2003-12-30 17:07:57 +01:00
|
|
|
// 16x16 8x8 4x4 2x2 16x8 8x4 4x2 8x16 4x8 2x4
|
2005-12-17 19:14:38 +01:00
|
|
|
|
2009-02-19 01:28:08 +01:00
|
|
|
me_cmp_func sad[6]; /* identical to pix_absAxA except additional void * */
|
|
|
|
me_cmp_func sse[6];
|
|
|
|
me_cmp_func hadamard8_diff[6];
|
|
|
|
me_cmp_func dct_sad[6];
|
|
|
|
me_cmp_func quant_psnr[6];
|
|
|
|
me_cmp_func bit[6];
|
|
|
|
me_cmp_func rd[6];
|
|
|
|
me_cmp_func vsad[6];
|
|
|
|
me_cmp_func vsse[6];
|
|
|
|
me_cmp_func nsse[6];
|
|
|
|
me_cmp_func w53[6];
|
|
|
|
me_cmp_func w97[6];
|
|
|
|
me_cmp_func dct_max[6];
|
|
|
|
me_cmp_func dct264_sad[6];
|
|
|
|
|
|
|
|
me_cmp_func me_pre_cmp[6];
|
|
|
|
me_cmp_func me_cmp[6];
|
|
|
|
me_cmp_func me_sub_cmp[6];
|
|
|
|
me_cmp_func mb_cmp[6];
|
|
|
|
me_cmp_func ildct_cmp[6]; //only width 16 used
|
|
|
|
me_cmp_func frame_skip_cmp[6]; //only width 8 used
|
2002-11-11 10:40:17 +01:00
|
|
|
|
2007-07-09 01:15:00 +02:00
|
|
|
int (*ssd_int8_vs_int16)(const int8_t *pix1, const int16_t *pix2,
|
|
|
|
int size);
|
2007-03-30 21:15:31 +02:00
|
|
|
|
2003-03-15 03:06:09 +01:00
|
|
|
/**
|
|
|
|
* Halfpel motion compensation with rounding (a+b+1)>>1.
|
2007-03-30 22:56:32 +02:00
|
|
|
* this is an array[4][4] of motion compensation functions for 4
|
2004-01-01 19:02:30 +01:00
|
|
|
* horizontal blocksizes (8,16) and the 4 halfpel positions<br>
|
2003-04-25 02:21:53 +02:00
|
|
|
* *pixels_tab[ 0->16xH 1->8xH ][ xhalfpel + 2*yhalfpel ]
|
2003-03-15 03:06:09 +01:00
|
|
|
* @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 01:30:03 +02:00
|
|
|
op_pixels_func put_pixels_tab[4][4];
|
2003-03-15 03:06:09 +01:00
|
|
|
|
|
|
|
/**
|
|
|
|
* Halfpel motion compensation with rounding (a+b+1)>>1.
|
2005-12-17 19:14:38 +01:00
|
|
|
* This is an array[4][4] of motion compensation functions for 4
|
2004-01-01 19:02:30 +01:00
|
|
|
* horizontal blocksizes (8,16) and the 4 halfpel positions<br>
|
2003-04-25 02:21:53 +02:00
|
|
|
* *pixels_tab[ 0->16xH 1->8xH ][ xhalfpel + 2*yhalfpel ]
|
2003-03-15 03:06:09 +01:00
|
|
|
* @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 03:44:44 +02:00
|
|
|
op_pixels_func avg_pixels_tab[4][4];
|
2003-03-15 03:06:09 +01:00
|
|
|
|
|
|
|
/**
|
|
|
|
* Halfpel motion compensation with no rounding (a+b)>>1.
|
2007-03-30 22:56:32 +02:00
|
|
|
* this is an array[2][4] of motion compensation functions for 2
|
2003-05-05 12:00:57 +02:00
|
|
|
* horizontal blocksizes (8,16) and the 4 halfpel positions<br>
|
2003-04-25 02:21:53 +02:00
|
|
|
* *pixels_tab[ 0->16xH 1->8xH ][ xhalfpel + 2*yhalfpel ]
|
2003-03-15 03:06:09 +01:00
|
|
|
* @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
|
|
|
|
*/
|
2004-06-11 23:12:46 +02:00
|
|
|
op_pixels_func put_no_rnd_pixels_tab[4][4];
|
2003-03-15 03:06:09 +01:00
|
|
|
|
|
|
|
/**
|
|
|
|
* Halfpel motion compensation with no rounding (a+b)>>1.
|
2007-03-30 22:56:32 +02:00
|
|
|
* this is an array[2][4] of motion compensation functions for 2
|
2003-05-05 12:00:57 +02:00
|
|
|
* horizontal blocksizes (8,16) and the 4 halfpel positions<br>
|
2003-04-25 02:21:53 +02:00
|
|
|
* *pixels_tab[ 0->16xH 1->8xH ][ xhalfpel + 2*yhalfpel ]
|
2003-03-15 03:06:09 +01:00
|
|
|
* @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
|
|
|
|
*/
|
2004-06-11 23:12:46 +02:00
|
|
|
op_pixels_func avg_no_rnd_pixels_tab[4][4];
|
2005-12-17 19:14:38 +01:00
|
|
|
|
2004-03-05 19:44:04 +01:00
|
|
|
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 19:14:38 +01:00
|
|
|
|
2003-05-16 01:30:03 +02:00
|
|
|
/**
|
|
|
|
* Thirdpel motion compensation with rounding (a+b+1)>>1.
|
2007-03-30 22:56:32 +02:00
|
|
|
* this is an array[12] of motion compensation functions for the 9 thirdpe
|
|
|
|
* positions<br>
|
2003-05-16 01:30:03 +02:00
|
|
|
* *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 03:44:44 +02:00
|
|
|
tpel_mc_func avg_tpel_pixels_tab[11]; //FIXME individual func ptr per width?
|
|
|
|
|
2002-11-11 10:40:17 +01:00
|
|
|
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 00:51:46 +01:00
|
|
|
qpel_mc_func put_mspel_pixels_tab[8];
|
2005-12-17 19:14:38 +01:00
|
|
|
|
2003-04-04 16:42:28 +02:00
|
|
|
/**
|
2007-03-30 22:56:32 +02:00
|
|
|
* h264 Chroma MC
|
2003-04-04 16:42:28 +02:00
|
|
|
*/
|
|
|
|
h264_chroma_mc_func put_h264_chroma_pixels_tab[3];
|
|
|
|
h264_chroma_mc_func avg_h264_chroma_pixels_tab[3];
|
2002-11-11 10:40:17 +01:00
|
|
|
|
2006-01-08 02:50:34 +01:00
|
|
|
qpel_mc_func put_h264_qpel_pixels_tab[4][16];
|
|
|
|
qpel_mc_func avg_h264_qpel_pixels_tab[4][16];
|
2005-12-17 19:14:38 +01:00
|
|
|
|
2006-10-01 23:25:17 +02:00
|
|
|
qpel_mc_func put_2tap_qpel_pixels_tab[4][16];
|
|
|
|
qpel_mc_func avg_2tap_qpel_pixels_tab[4][16];
|
|
|
|
|
2003-12-30 17:07:57 +01:00
|
|
|
me_cmp_func pix_abs[2][4];
|
2005-12-17 19:14:38 +01:00
|
|
|
|
2002-11-14 20:20:04 +01:00
|
|
|
/* huffyuv specific */
|
|
|
|
void (*add_bytes)(uint8_t *dst/*align 16*/, uint8_t *src/*align 16*/, int w);
|
2008-02-21 08:10:46 +01:00
|
|
|
void (*add_bytes_l2)(uint8_t *dst/*align 16*/, uint8_t *src1/*align 16*/, uint8_t *src2/*align 16*/, int w);
|
2002-12-28 00:51:46 +01:00
|
|
|
void (*diff_bytes)(uint8_t *dst/*align 16*/, uint8_t *src1/*align 16*/, uint8_t *src2/*align 1*/,int w);
|
2003-10-13 19:27:30 +02:00
|
|
|
/**
|
|
|
|
* subtract huffyuv's variant of median prediction
|
|
|
|
* note, this might read from src1[-1], src2[-1]
|
|
|
|
*/
|
2009-10-18 22:47:25 +02:00
|
|
|
void (*sub_hfyu_median_prediction)(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, int w, int *left, int *left_top);
|
|
|
|
void (*add_hfyu_median_prediction)(uint8_t *dst, const uint8_t *top, const uint8_t *diff, int w, int *left, int *left_top);
|
2009-10-18 22:10:10 +02:00
|
|
|
int (*add_hfyu_left_prediction)(uint8_t *dst, const uint8_t *src, int w, int left);
|
2010-01-14 02:32:49 +01:00
|
|
|
void (*add_hfyu_left_prediction_bgr32)(uint8_t *dst, const uint8_t *src, int w, int *red, int *green, int *blue, int *alpha);
|
2008-02-21 08:10:46 +01:00
|
|
|
/* 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 14:49:15 +01:00
|
|
|
void (*bswap_buf)(uint32_t *dst, const uint32_t *src, int w);
|
2010-04-21 01:09:00 +02:00
|
|
|
void (*bswap16_buf)(uint16_t *dst, const uint16_t *src, int len);
|
2005-04-25 03:01:41 +02:00
|
|
|
|
2003-12-01 16:23:14 +01:00
|
|
|
void (*h263_v_loop_filter)(uint8_t *src, int stride, int qscale);
|
|
|
|
void (*h263_h_loop_filter)(uint8_t *src, int stride, int qscale);
|
|
|
|
|
2004-05-30 23:21:13 +02:00
|
|
|
void (*h261_loop_filter)(uint8_t *src, int stride);
|
2004-05-30 22:37:15 +02:00
|
|
|
|
2007-11-09 22:37:48 +01:00
|
|
|
void (*x8_v_loop_filter)(uint8_t *src, int stride, int qscale);
|
|
|
|
void (*x8_h_loop_filter)(uint8_t *src, int stride, int qscale);
|
|
|
|
|
2010-04-17 04:04:30 +02:00
|
|
|
void (*vp3_idct_dc_add)(uint8_t *dest/*align 8*/, int line_size, const DCTELEM *block/*align 16*/);
|
2008-10-04 12:26:17 +02:00
|
|
|
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 21:06:25 +02:00
|
|
|
/* assume len is a multiple of 4, and arrays are 16-byte aligned */
|
2006-08-03 05:18:47 +02:00
|
|
|
void (*vorbis_inverse_coupling)(float *mag, float *ang, int blocksize);
|
2008-08-14 01:33:48 +02:00
|
|
|
void (*ac3_downmix)(float (*samples)[256], float (*matrix)[2], int out_ch, int in_ch, int len);
|
2006-08-10 21:06:25 +02:00
|
|
|
/* assume len is a multiple of 8, and arrays are 16-byte aligned */
|
2011-01-13 21:28:06 +01:00
|
|
|
void (*vector_fmul)(float *dst, const float *src0, const float *src1, int len);
|
2006-08-10 21:06:25 +02:00
|
|
|
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 */
|
2009-09-27 18:51:54 +02:00
|
|
|
void (*vector_fmul_add)(float *dst, const float *src0, const float *src1, const float *src2, int len);
|
2008-07-13 16:56:01 +02:00
|
|
|
/* assume len is a multiple of 4, and arrays are 16-byte aligned */
|
2011-01-31 20:26:02 +01:00
|
|
|
void (*vector_fmul_window)(float *dst, const float *src0, const float *src1, const float *win, int len);
|
2008-08-14 01:35:40 +02:00
|
|
|
/* assume len is a multiple of 8, and arrays are 16-byte aligned */
|
2009-08-27 17:38:59 +02:00
|
|
|
void (*vector_clipf)(float *dst /* align 16 */, const float *src /* align 16 */, float min, float max, int len /* align 16 */);
|
2009-09-22 02:48:41 +02:00
|
|
|
/**
|
|
|
|
* Multiply a vector of floats by a scalar float. Source and
|
|
|
|
* destination vectors must overlap exactly or not at all.
|
|
|
|
* @param dst result vector, 16-byte aligned
|
|
|
|
* @param src input vector, 16-byte aligned
|
|
|
|
* @param mul scalar value
|
|
|
|
* @param len length of vector, multiple of 4
|
|
|
|
*/
|
|
|
|
void (*vector_fmul_scalar)(float *dst, const float *src, float mul,
|
|
|
|
int len);
|
|
|
|
/**
|
|
|
|
* Multiply a vector of floats by concatenated short vectors of
|
|
|
|
* floats and by a scalar float. Source and destination vectors
|
|
|
|
* must overlap exactly or not at all.
|
|
|
|
* [0]: short vectors of length 2, 8-byte aligned
|
|
|
|
* [1]: short vectors of length 4, 16-byte aligned
|
|
|
|
* @param dst output vector, 16-byte aligned
|
|
|
|
* @param src input vector, 16-byte aligned
|
|
|
|
* @param sv array of pointers to short vectors
|
|
|
|
* @param mul scalar value
|
|
|
|
* @param len number of elements in src and dst, multiple of 4
|
|
|
|
*/
|
|
|
|
void (*vector_fmul_sv_scalar[2])(float *dst, const float *src,
|
|
|
|
const float **sv, float mul, int len);
|
|
|
|
/**
|
|
|
|
* Multiply short vectors of floats by a scalar float, store
|
|
|
|
* concatenated result.
|
|
|
|
* [0]: short vectors of length 2, 8-byte aligned
|
|
|
|
* [1]: short vectors of length 4, 16-byte aligned
|
|
|
|
* @param dst output vector, 16-byte aligned
|
|
|
|
* @param sv array of pointers to short vectors
|
|
|
|
* @param mul scalar value
|
|
|
|
* @param len number of output elements, multiple of 4
|
|
|
|
*/
|
|
|
|
void (*sv_fmul_scalar[2])(float *dst, const float **sv,
|
|
|
|
float mul, int len);
|
|
|
|
/**
|
|
|
|
* Calculate the scalar product of two vectors of floats.
|
|
|
|
* @param v1 first vector, 16-byte aligned
|
|
|
|
* @param v2 second vector, 16-byte aligned
|
|
|
|
* @param len length of vectors, multiple of 4
|
|
|
|
*/
|
|
|
|
float (*scalarproduct_float)(const float *v1, const float *v2, int len);
|
|
|
|
/**
|
|
|
|
* Calculate the sum and difference of two vectors of floats.
|
|
|
|
* @param v1 first input vector, sum output, 16-byte aligned
|
|
|
|
* @param v2 second input vector, difference output, 16-byte aligned
|
|
|
|
* @param len length of vectors, multiple of 4
|
|
|
|
*/
|
|
|
|
void (*butterflies_float)(float *restrict v1, float *restrict v2, int len);
|
2006-08-10 21:06:25 +02:00
|
|
|
|
2003-03-03 15:54:00 +01:00
|
|
|
/* (I)DCT */
|
|
|
|
void (*fdct)(DCTELEM *block/* align 16*/);
|
2003-10-24 20:28:01 +02:00
|
|
|
void (*fdct248)(DCTELEM *block/* align 16*/);
|
2005-12-17 19:14:38 +01:00
|
|
|
|
2003-06-21 03:11:31 +02:00
|
|
|
/* IDCT really*/
|
|
|
|
void (*idct)(DCTELEM *block/* align 16*/);
|
2005-12-17 19:14:38 +01:00
|
|
|
|
2003-03-06 11:08:15 +01:00
|
|
|
/**
|
2003-03-06 11:26:19 +01:00
|
|
|
* block -> idct -> clip to unsigned 8 bit -> dest.
|
2003-03-06 11:08:15 +01:00
|
|
|
* (-1392, 0, 0, ...) -> idct -> (-174, -174, ...) -> put -> (0, 0, ...)
|
2007-09-27 13:29:21 +02:00
|
|
|
* @param line_size size in bytes of a horizontal line of dest
|
2003-03-06 11:08:15 +01:00
|
|
|
*/
|
2003-03-03 15:54:00 +01:00
|
|
|
void (*idct_put)(uint8_t *dest/*align 8*/, int line_size, DCTELEM *block/*align 16*/);
|
2005-12-17 19:14:38 +01:00
|
|
|
|
2003-03-06 11:08:15 +01:00
|
|
|
/**
|
|
|
|
* block -> idct -> add dest -> clip to unsigned 8 bit -> dest.
|
2007-09-27 13:29:21 +02:00
|
|
|
* @param line_size size in bytes of a horizontal line of dest
|
2003-03-06 11:08:15 +01:00
|
|
|
*/
|
2003-03-03 15:54:00 +01:00
|
|
|
void (*idct_add)(uint8_t *dest/*align 8*/, int line_size, DCTELEM *block/*align 16*/);
|
2005-12-17 19:14:38 +01:00
|
|
|
|
2003-03-06 11:08:15 +01:00
|
|
|
/**
|
2003-03-06 11:26:19 +01:00
|
|
|
* idct input permutation.
|
2003-05-11 12:19:05 +02:00
|
|
|
* 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 11:08:15 +01:00
|
|
|
* 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 15:54:00 +01:00
|
|
|
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 20:28:40 +02:00
|
|
|
#define FF_PARTTRANS_IDCT_PERM 5
|
2008-04-16 03:34:55 +02:00
|
|
|
#define FF_SSE2_IDCT_PERM 6
|
2003-03-03 15:54:00 +01:00
|
|
|
|
2004-02-06 03:12:37 +01:00
|
|
|
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 19:14:38 +01:00
|
|
|
|
2011-03-22 22:36:57 +01:00
|
|
|
void (*draw_edges)(uint8_t *buf, int wrap, int width, int height, int w, int sides);
|
2008-03-05 01:38:14 +01:00
|
|
|
#define EDGE_WIDTH 16
|
2011-03-22 22:36:57 +01:00
|
|
|
#define EDGE_TOP 1
|
|
|
|
#define EDGE_BOTTOM 2
|
2008-03-04 01:07:41 +01:00
|
|
|
|
2006-03-23 21:16:36 +01:00
|
|
|
void (*prefetch)(void *mem, int stride, int h);
|
2006-03-31 22:43:37 +02:00
|
|
|
|
|
|
|
void (*shrink[4])(uint8_t *dst, int dst_wrap, const uint8_t *src, int src_wrap, int width, int height);
|
2006-07-30 05:45:02 +02:00
|
|
|
|
2009-04-30 14:30:05 +02:00
|
|
|
/* mlp/truehd functions */
|
2009-05-15 17:34:22 +02:00
|
|
|
void (*mlp_filter_channel)(int32_t *state, const int32_t *coeff,
|
|
|
|
int firorder, int iirorder,
|
2009-04-30 14:30:05 +02:00
|
|
|
unsigned int filter_shift, int32_t mask, int blocksize,
|
|
|
|
int32_t *sample_buffer);
|
|
|
|
|
2007-11-09 22:37:48 +01:00
|
|
|
/* intrax8 functions */
|
2007-11-10 13:24:02 +01:00
|
|
|
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 22:37:48 +01:00
|
|
|
int * range, int * sum, int edges);
|
|
|
|
|
2008-07-06 08:06:55 +02:00
|
|
|
/**
|
|
|
|
* Calculate scalar product of two vectors.
|
2008-07-10 08:15:13 +02:00
|
|
|
* @param len length of vectors, should be multiple of 16
|
2008-07-06 08:06:55 +02:00
|
|
|
* @param shift number of bits to discard from product
|
|
|
|
*/
|
2010-06-27 17:11:38 +02:00
|
|
|
int32_t (*scalarproduct_int16)(const int16_t *v1, const int16_t *v2/*align 16*/, int len, int shift);
|
2009-12-05 16:09:10 +01:00
|
|
|
/* ape functions */
|
|
|
|
/**
|
|
|
|
* Calculate scalar product of v1 and v2,
|
|
|
|
* and v1[i] += v3[i] * mul
|
|
|
|
* @param len length of vectors, should be multiple of 16
|
|
|
|
*/
|
2010-06-27 17:11:38 +02:00
|
|
|
int32_t (*scalarproduct_and_madd_int16)(int16_t *v1/*align 16*/, const int16_t *v2, const int16_t *v3, int len, int mul);
|
2008-12-01 07:36:22 +01:00
|
|
|
|
2011-03-20 18:31:36 +01:00
|
|
|
/**
|
|
|
|
* Apply symmetric window in 16-bit fixed-point.
|
|
|
|
* @param output destination array
|
|
|
|
* constraints: 16-byte aligned
|
|
|
|
* @param input source array
|
|
|
|
* constraints: 16-byte aligned
|
|
|
|
* @param window window array
|
|
|
|
* constraints: 16-byte aligned, at least len/2 elements
|
|
|
|
* @param len full window length
|
|
|
|
* constraints: multiple of ? greater than zero
|
|
|
|
*/
|
|
|
|
void (*apply_window_int16)(int16_t *output, const int16_t *input,
|
|
|
|
const int16_t *window, unsigned int len);
|
|
|
|
|
2008-12-01 07:40:36 +01:00
|
|
|
/* 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 07:36:22 +01:00
|
|
|
/* 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];
|
2010-02-21 14:28:46 +01:00
|
|
|
|
|
|
|
/* bink functions */
|
|
|
|
op_fill_func fill_block_tab[2];
|
|
|
|
void (*scale_block)(const uint8_t src[64]/*align 8*/, uint8_t *dst/*align 8*/, int linesize);
|
2002-11-11 10:40:17 +01:00
|
|
|
} DSPContext;
|
|
|
|
|
2006-11-14 04:18:09 +01:00
|
|
|
void dsputil_static_init(void);
|
2003-03-03 15:54:00 +01:00
|
|
|
void dsputil_init(DSPContext* p, AVCodecContext *avctx);
|
2001-07-22 16:18:56 +02:00
|
|
|
|
2006-12-07 19:33:00 +01:00
|
|
|
int ff_check_alignment(void);
|
|
|
|
|
2002-10-23 17:11:07 +02:00
|
|
|
/**
|
|
|
|
* permute block according to permuatation.
|
|
|
|
* @param last last non zero element in scantable order
|
|
|
|
*/
|
2003-02-11 17:35:48 +01:00
|
|
|
void ff_block_permute(DCTELEM *block, uint8_t *permutation, const uint8_t *scantable, int last);
|
2001-08-08 00:43:19 +02:00
|
|
|
|
2004-01-05 23:57:07 +01:00
|
|
|
void ff_set_cmp(DSPContext* c, me_cmp_func *cmp, int type);
|
|
|
|
|
2005-12-22 02:10:11 +01:00
|
|
|
#define BYTE_VEC32(c) ((c)*0x01010101UL)
|
2011-03-29 17:48:59 +02:00
|
|
|
#define BYTE_VEC64(c) ((c)*0x0001000100010001UL)
|
2003-05-15 01:08:01 +02:00
|
|
|
|
|
|
|
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);
|
|
|
|
}
|
|
|
|
|
2011-03-29 17:48:59 +02:00
|
|
|
static inline uint64_t rnd_avg64(uint64_t a, uint64_t b)
|
|
|
|
{
|
|
|
|
return (a | b) - (((a ^ b) & ~BYTE_VEC64(0x01)) >> 1);
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline uint64_t no_rnd_avg64(uint64_t a, uint64_t b)
|
|
|
|
{
|
|
|
|
return (a & b) + (((a ^ b) & ~BYTE_VEC64(0x01)) >> 1);
|
|
|
|
}
|
|
|
|
|
2004-08-22 19:16:03 +02:00
|
|
|
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 17:31:23 +01:00
|
|
|
case FF_CMP_DCT264:
|
2004-08-22 19:16:03 +02:00
|
|
|
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 11:08:15 +01:00
|
|
|
/**
|
2003-03-06 11:26:19 +01:00
|
|
|
* Empty mmx state.
|
2003-03-06 11:08:15 +01:00
|
|
|
* this must be called between any dsp function and float/double code.
|
|
|
|
* for example sin(); dsp->idct_put(); emms_c(); cos()
|
|
|
|
*/
|
2002-11-11 10:40:17 +01:00
|
|
|
#define emms_c()
|
|
|
|
|
2007-06-16 12:44:46 +02:00
|
|
|
void dsputil_init_alpha(DSPContext* c, AVCodecContext *avctx);
|
2008-12-17 01:54:54 +01:00
|
|
|
void dsputil_init_arm(DSPContext* c, AVCodecContext *avctx);
|
2007-06-16 12:44:46 +02:00
|
|
|
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);
|
|
|
|
|
2010-03-14 18:50:12 +01:00
|
|
|
void ff_dsputil_init_dwt(DSPContext *c);
|
2010-03-06 23:36:49 +01:00
|
|
|
void ff_rv30dsp_init(DSPContext* c, AVCodecContext *avctx);
|
|
|
|
void ff_rv40dsp_init(DSPContext* c, AVCodecContext *avctx);
|
|
|
|
void ff_intrax8dsp_init(DSPContext* c, AVCodecContext *avctx);
|
|
|
|
void ff_mlp_init(DSPContext* c, AVCodecContext *avctx);
|
|
|
|
void ff_mlp_init_x86(DSPContext* c, AVCodecContext *avctx);
|
|
|
|
|
2009-01-14 00:44:16 +01:00
|
|
|
#if HAVE_MMX
|
2001-07-22 16:18:56 +02:00
|
|
|
|
2002-11-12 16:00:04 +01:00
|
|
|
#undef emms_c
|
2002-11-11 10:40:17 +01:00
|
|
|
|
2001-07-22 16:18:56 +02:00
|
|
|
static inline void emms(void)
|
|
|
|
{
|
2008-10-16 15:34:09 +02:00
|
|
|
__asm__ volatile ("emms;":::"memory");
|
2001-07-24 22:42:03 +02:00
|
|
|
}
|
|
|
|
|
2011-02-26 22:51:56 +01:00
|
|
|
#define emms_c() \
|
|
|
|
{\
|
|
|
|
if(av_get_cpu_flags() & AV_CPU_FLAG_MMX)\
|
|
|
|
emms();\
|
|
|
|
}
|
2001-07-22 16:18:56 +02:00
|
|
|
|
2009-01-14 00:44:16 +01:00
|
|
|
#elif ARCH_ARM
|
2001-08-13 23:45:36 +02:00
|
|
|
|
2009-01-14 00:44:16 +01:00
|
|
|
#if HAVE_NEON
|
2008-08-25 03:59:06 +02:00
|
|
|
# define STRIDE_ALIGN 16
|
|
|
|
#endif
|
|
|
|
|
2009-01-14 00:44:16 +01:00
|
|
|
#elif ARCH_PPC
|
2002-08-28 15:14:36 +02:00
|
|
|
|
2004-10-26 14:47:56 +02:00
|
|
|
#define STRIDE_ALIGN 16
|
2002-08-28 15:14:36 +02:00
|
|
|
|
2009-01-14 00:44:16 +01:00
|
|
|
#elif HAVE_MMI
|
2002-09-23 10:44:24 +02:00
|
|
|
|
2004-10-26 14:47:56 +02:00
|
|
|
#define STRIDE_ALIGN 16
|
2002-09-23 10:44:24 +02:00
|
|
|
|
2008-02-16 20:18:30 +01:00
|
|
|
#endif
|
2001-07-22 16:18:56 +02:00
|
|
|
|
2008-02-16 20:18:30 +01:00
|
|
|
#ifndef STRIDE_ALIGN
|
|
|
|
# define STRIDE_ALIGN 8
|
2001-07-22 16:18:56 +02:00
|
|
|
#endif
|
|
|
|
|
2011-01-20 03:04:54 +01:00
|
|
|
#define LOCAL_ALIGNED_A(a, t, v, s, o, ...) \
|
|
|
|
uint8_t la_##v[sizeof(t s o) + (a)]; \
|
|
|
|
t (*v) o = (void *)FFALIGN((uintptr_t)la_##v, a)
|
|
|
|
|
|
|
|
#define LOCAL_ALIGNED_D(a, t, v, s, o, ...) DECLARE_ALIGNED(a, t, v) s o
|
|
|
|
|
|
|
|
#define LOCAL_ALIGNED(a, t, v, ...) LOCAL_ALIGNED_A(a, t, v, __VA_ARGS__,,)
|
2010-02-17 21:36:12 +01:00
|
|
|
|
|
|
|
#if HAVE_LOCAL_ALIGNED_8
|
2011-01-20 03:04:54 +01:00
|
|
|
# define LOCAL_ALIGNED_8(t, v, ...) LOCAL_ALIGNED_D(8, t, v, __VA_ARGS__,,)
|
2010-02-17 21:36:12 +01:00
|
|
|
#else
|
2011-01-20 03:04:54 +01:00
|
|
|
# define LOCAL_ALIGNED_8(t, v, ...) LOCAL_ALIGNED(8, t, v, __VA_ARGS__)
|
2010-02-17 21:36:12 +01:00
|
|
|
#endif
|
|
|
|
|
|
|
|
#if HAVE_LOCAL_ALIGNED_16
|
2011-01-20 03:04:54 +01:00
|
|
|
# define LOCAL_ALIGNED_16(t, v, ...) LOCAL_ALIGNED_D(16, t, v, __VA_ARGS__,,)
|
2010-02-17 21:36:12 +01:00
|
|
|
#else
|
2011-01-20 03:04:54 +01:00
|
|
|
# define LOCAL_ALIGNED_16(t, v, ...) LOCAL_ALIGNED(16, t, v, __VA_ARGS__)
|
2010-02-17 21:36:12 +01:00
|
|
|
#endif
|
|
|
|
|
2002-02-26 23:14:27 +01:00
|
|
|
/* PSNR */
|
2003-02-11 17:35:48 +01:00
|
|
|
void get_psnr(uint8_t *orig_image[3], uint8_t *coded_image[3],
|
2002-02-26 23:14:27 +01:00
|
|
|
int orig_linesize[3], int coded_linesize,
|
|
|
|
AVCodecContext *avctx);
|
2002-10-28 01:34:08 +01:00
|
|
|
|
2007-12-21 13:38:41 +01:00
|
|
|
#define WRAPPER8_16(name8, name16)\
|
2003-12-30 17:07:57 +01:00
|
|
|
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 13:38:41 +01:00
|
|
|
#define WRAPPER8_16_SQ(name8, name16)\
|
2003-12-30 17:07:57 +01:00
|
|
|
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 00:51:46 +01:00
|
|
|
}
|
|
|
|
|
2006-11-26 19:25:07 +01:00
|
|
|
|
2009-04-14 06:07:35 +02:00
|
|
|
static inline void copy_block2(uint8_t *dst, const uint8_t *src, int dstStride, int srcStride, int h)
|
2006-11-26 19:25:07 +01:00
|
|
|
{
|
|
|
|
int i;
|
|
|
|
for(i=0; i<h; i++)
|
|
|
|
{
|
2007-08-09 19:01:15 +02:00
|
|
|
AV_WN16(dst , AV_RN16(src ));
|
2006-11-26 19:25:07 +01:00
|
|
|
dst+=dstStride;
|
|
|
|
src+=srcStride;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2009-04-14 06:07:35 +02:00
|
|
|
static inline void copy_block4(uint8_t *dst, const uint8_t *src, int dstStride, int srcStride, int h)
|
2006-11-26 19:25:07 +01:00
|
|
|
{
|
|
|
|
int i;
|
|
|
|
for(i=0; i<h; i++)
|
|
|
|
{
|
2007-08-09 19:01:15 +02:00
|
|
|
AV_WN32(dst , AV_RN32(src ));
|
2006-11-26 19:25:07 +01:00
|
|
|
dst+=dstStride;
|
|
|
|
src+=srcStride;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2009-04-14 06:07:35 +02:00
|
|
|
static inline void copy_block8(uint8_t *dst, const uint8_t *src, int dstStride, int srcStride, int h)
|
2006-11-26 19:25:07 +01:00
|
|
|
{
|
|
|
|
int i;
|
|
|
|
for(i=0; i<h; i++)
|
|
|
|
{
|
2007-08-09 19:01:15 +02:00
|
|
|
AV_WN32(dst , AV_RN32(src ));
|
|
|
|
AV_WN32(dst+4 , AV_RN32(src+4 ));
|
2006-11-26 19:25:07 +01:00
|
|
|
dst+=dstStride;
|
|
|
|
src+=srcStride;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2009-04-14 06:07:35 +02:00
|
|
|
static inline void copy_block9(uint8_t *dst, const uint8_t *src, int dstStride, int srcStride, int h)
|
2006-11-26 19:25:07 +01:00
|
|
|
{
|
|
|
|
int i;
|
|
|
|
for(i=0; i<h; i++)
|
|
|
|
{
|
2007-08-09 19:01:15 +02:00
|
|
|
AV_WN32(dst , AV_RN32(src ));
|
|
|
|
AV_WN32(dst+4 , AV_RN32(src+4 ));
|
2006-11-26 19:25:07 +01:00
|
|
|
dst[8]= src[8];
|
|
|
|
dst+=dstStride;
|
|
|
|
src+=srcStride;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2009-04-14 06:07:35 +02:00
|
|
|
static inline void copy_block16(uint8_t *dst, const uint8_t *src, int dstStride, int srcStride, int h)
|
2006-11-26 19:25:07 +01:00
|
|
|
{
|
|
|
|
int i;
|
|
|
|
for(i=0; i<h; i++)
|
|
|
|
{
|
2007-08-09 19:01:15 +02:00
|
|
|
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 19:25:07 +01:00
|
|
|
dst+=dstStride;
|
|
|
|
src+=srcStride;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2009-04-14 06:07:35 +02:00
|
|
|
static inline void copy_block17(uint8_t *dst, const uint8_t *src, int dstStride, int srcStride, int h)
|
2006-11-26 19:25:07 +01:00
|
|
|
{
|
|
|
|
int i;
|
|
|
|
for(i=0; i<h; i++)
|
|
|
|
{
|
2007-08-09 19:01:15 +02:00
|
|
|
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 19:25:07 +01:00
|
|
|
dst[16]= src[16];
|
|
|
|
dst+=dstStride;
|
|
|
|
src+=srcStride;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2008-08-31 09:39:47 +02:00
|
|
|
#endif /* AVCODEC_DSPUTIL_H */
|