dsputil.h

This the source release kit for the following system configuration(s): - AMD Alchemy(TM) DBAu1200(

/*
 * DSP utils
 * Copyright (c) 2000, 2001, 2002 Fabrice Bellard.
 * Copyright (c) 2002-2004 Michael Niedermayer <[e-mail]>
 *
 * This file is part of FFmpeg.
 *
 * FFmpeg is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2.1 of the License, or (at your option) any later version.
 *
 * FFmpeg is distributed in the hope that it will be useful,
 * 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
 * License along with FFmpeg; if not, write to the Free Software
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
 */

/**
 * @file dsputil.h
 * DSP utils.
 * 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
 */

#ifndef DSPUTIL_H
#define DSPUTIL_H

#include "common.h"
#include "avcodec.h"


//#define DEBUG
/* dct code */
typedef short DCTELEM;
typedef int DWTELEM;

void ff_h264_idct8_add_c(uint8_t *dst, DCTELEM *block, int stride);
void ff_h264_idct_add_c(uint8_t *dst, DCTELEM *block, int stride);
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);
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);

/* pixel operations */
#define MAX_NEG_CROP 1024

/* temporary */
uint8_t cropTbl[256 + 2 * MAX_NEG_CROP];

/* minimum alignment rules ;)
if u 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-dev list, ...

!warning these alignments might not match reallity, (missing attribute((align)) stuff somewhere possible)
i (michael) didnt check them, these are just the alignents which i think could be reached easily ...

!future video codecs might need functions with less strict alignment
*/

/* add and put pixel (decoding) */
typedef void (*qpel_mc_func)(uint8_t *dst/*align width (8 or 16)*/, uint8_t *src/*align 1*/, int stride);
typedef void (*h264_chroma_mc_func)(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int srcStride, int h, int x, int y);
typedef void (*h264_weight_func)(uint8_t *block, int stride, int log2_denom, int weight, int offset);
typedef void (*h264_biweight_func)(uint8_t *dst, uint8_t *src, int stride, int log2_denom, int weightd, int weights, int offset);

/**
 * DSPContext.
 */
typedef struct DSPContext {
    /* pixel ops : interface with DCT */
    void (*add_pixels8)(uint8_t *pixels, DCTELEM *block, int line_size);
    void (*add_pixels4)(uint8_t *pixels, DCTELEM *block, int line_size);

    void (*clear_blocks)(DCTELEM *blocks/*align 16*/);

    /**
     * h264 Chram MC
     */
    h264_chroma_mc_func put_h264_chroma_pixels_tab[3];
    /* This is really one func used in VC-1 decoding */
    h264_chroma_mc_func put_no_rnd_h264_chroma_pixels_tab[3];
    h264_chroma_mc_func avg_h264_chroma_pixels_tab[3];

    qpel_mc_func put_h264_qpel_pixels_tab[4][16];
    qpel_mc_func avg_h264_qpel_pixels_tab[4][16];

    qpel_mc_func put_2tap_qpel_pixels_tab[4][16];
    qpel_mc_func avg_2tap_qpel_pixels_tab[4][16];

    h264_weight_func weight_h264_pixels_tab[10];
    h264_biweight_func biweight_h264_pixels_tab[10];

    void (*h264_v_loop_filter_luma)(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0);
    void (*h264_h_loop_filter_luma)(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0);
    void (*h264_v_loop_filter_chroma)(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0);
    void (*h264_h_loop_filter_chroma)(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0);
    void (*h264_v_loop_filter_chroma_intra)(uint8_t *pix, int stride, int alpha, int beta);
    void (*h264_h_loop_filter_chroma_intra)(uint8_t *pix, int stride, int alpha, int beta);
    // 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],
                                      int bidir, int edges, int step, int mask_mv0, int mask_mv1);

    void (*h264_idct_add)(uint8_t *dst, DCTELEM *block, int stride);
    void (*h264_idct8_add)(uint8_t *dst, DCTELEM *block, int stride);
    void (*h264_idct_dc_add)(uint8_t *dst, DCTELEM *block, int stride);
    void (*h264_idct8_dc_add)(uint8_t *dst, DCTELEM *block, int stride);

} DSPContext;

void dsputil_static_init(void);
void dsputil_init(DSPContext* p, AVCodecContext *avctx);

#define         BYTE_VEC32(c)   ((c)*0x01010101UL)

#define rnd_avg32(a, b)     (uint32_t)((a | b) - (((a ^ b) & ~BYTE_VEC32(0x01)) >> 1))
#define no_rnd_avg32(a, b)  (uint32_t)((a & b) + (((a ^ b) & ~BYTE_VEC32(0x01)) >> 1))

/**
 * Empty mmx state.
 * this must be called between any dsp function and float/double code.
 * for example sin(); dsp->idct_put(); emms_c(); cos()
 */
#define emms_c()

/* should be defined by architectures supporting
   one or more MultiMedia extension */
int mm_support(void);

#ifdef __GNUC__
  #define DECLARE_ALIGNED_16(t,v)       t v __attribute__ ((aligned (16)))
#else
  #define DECLARE_ALIGNED_16(t,v)      __declspec(align(16)) t v
#endif

#ifdef __GNUC__
  #define DECLARE_ALIGNED_8(t,v)       t v __attribute__ ((aligned (8)))
#else
  #define DECLARE_ALIGNED_8(t,v)      __declspec(align(8)) t v
#endif
#define STRIDE_ALIGN 8

#ifdef __GNUC__

struct unaligned_64 { uint64_t l; } __attribute__((packed));
struct unaligned_32 { uint32_t l; } __attribute__((packed));
struct unaligned_16 { uint16_t l; } __attribute__((packed));

#define LD16(a) (((const struct unaligned_16 *) (a))->l)
#define LD32(a) (((const struct unaligned_32 *) (a))->l)
#define LD64(a) (((const struct unaligned_64 *) (a))->l)

#define ST16(a, b) (((struct unaligned_16 *) (a))->l) = (b)
#define ST32(a, b) (((struct unaligned_32 *) (a))->l) = (b)

#else /* __GNUC__ */

#define LD16(a) (*((uint16_t __unaligned *)(a)))
#define LD32(a) (*((uint32_t __unaligned *)(a)))
#define LD64(a) (*((uint64_t __unaligned *)(a)))

#define ST16(a, b) *((uint16_t __unaligned *)(a)) = (b)
#define ST32(a, b) *((uint32_t __unaligned *)(a)) = (b)

#endif /* !__GNUC__ */

#endif