annotate vp56.h @ 12197:fbf4d5b1b664 libavcodec

Remove FF_MM_SSE2/3 flags for CPUs where this is generally not faster than regular MMX code. Examples of this are the Core1 CPU. Instead, set a new flag, FF_MM_SSE2/3SLOW, which can be checked for particular SSE2/3 functions that have been checked specifically on such CPUs and are actually faster than their MMX counterparts. In addition, use this flag to enable particular VP8 and LPC SSE2 functions that are faster than their MMX counterparts. Based on a patch by Loren Merritt <lorenm AT u washington edu>.

/**
 * @file
 * VP5 and VP6 compatible video decoder (common features)
 *
 * Copyright (C) 2006  Aurelien Jacobs <[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
 */

#ifndef AVCODEC_VP56_H
#define AVCODEC_VP56_H

#include "vp56data.h"
#include "dsputil.h"
#include "get_bits.h"
#include "bytestream.h"
#include "cabac.h"
#include "vp56dsp.h"

typedef struct vp56_context VP56Context;
typedef struct vp56_mv VP56mv;

typedef void (*VP56ParseVectorAdjustment)(VP56Context *s,
                                          VP56mv *vect);
typedef void (*VP56Filter)(VP56Context *s, uint8_t *dst, uint8_t *src,
                           int offset1, int offset2, int stride,
                           VP56mv mv, int mask, int select, int luma);
typedef void (*VP56ParseCoeff)(VP56Context *s);
typedef void (*VP56DefaultModelsInit)(VP56Context *s);
typedef void (*VP56ParseVectorModels)(VP56Context *s);
typedef void (*VP56ParseCoeffModels)(VP56Context *s);
typedef int  (*VP56ParseHeader)(VP56Context *s, const uint8_t *buf,
                                int buf_size, int *golden_frame);

typedef struct {
    int high;
    int bits; /* stored negated (i.e. negative "bits" is a positive number of
                 bits left) in order to eliminate a negate in cache refilling */
    const uint8_t *buffer;
    const uint8_t *end;
    unsigned long code_word;
} VP56RangeCoder;

typedef struct {
    uint8_t not_null_dc;
    VP56Frame ref_frame;
    DCTELEM dc_coeff;
} VP56RefDc;

struct vp56_mv {
    int x;
    int y;
};

typedef struct {
    uint8_t type;
    VP56mv mv;
} VP56Macroblock;

typedef struct {
    uint8_t coeff_reorder[64];       /* used in vp6 only */
    uint8_t coeff_index_to_pos[64];  /* used in vp6 only */
    uint8_t vector_sig[2];           /* delta sign */
    uint8_t vector_dct[2];           /* delta coding types */
    uint8_t vector_pdi[2][2];        /* predefined delta init */
    uint8_t vector_pdv[2][7];        /* predefined delta values */
    uint8_t vector_fdv[2][8];        /* 8 bit delta value definition */
    uint8_t coeff_dccv[2][11];       /* DC coeff value */
    uint8_t coeff_ract[2][3][6][11]; /* Run/AC coding type and AC coeff value */
    uint8_t coeff_acct[2][3][3][6][5];/* vp5 only AC coding type for coding group < 3 */
    uint8_t coeff_dcct[2][36][5];    /* DC coeff coding type */
    uint8_t coeff_runv[2][14];       /* run value (vp6 only) */
    uint8_t mb_type[3][10][10];      /* model for decoding MB type */
    uint8_t mb_types_stats[3][10][2];/* contextual, next MB type stats */
} VP56Model;

struct vp56_context {
    AVCodecContext *avctx;
    DSPContext dsp;
    VP56DSPContext vp56dsp;
    ScanTable scantable;
    AVFrame frames[4];
    AVFrame *framep[6];
    uint8_t *edge_emu_buffer_alloc;
    uint8_t *edge_emu_buffer;
    VP56RangeCoder c;
    VP56RangeCoder cc;
    VP56RangeCoder *ccp;
    int sub_version;

    /* frame info */
    int plane_width[4];
    int plane_height[4];
    int mb_width;   /* number of horizontal MB */
    int mb_height;  /* number of vertical MB */
    int block_offset[6];

    int quantizer;
    uint16_t dequant_dc;
    uint16_t dequant_ac;
    int8_t *qscale_table;

    /* DC predictors management */
    VP56RefDc *above_blocks;
    VP56RefDc left_block[4];
    int above_block_idx[6];
    DCTELEM prev_dc[3][3];    /* [plan][ref_frame] */

    /* blocks / macroblock */
    VP56mb mb_type;
    VP56Macroblock *macroblocks;
    DECLARE_ALIGNED(16, DCTELEM, block_coeff)[6][64];

    /* motion vectors */
    VP56mv mv[6];  /* vectors for each block in MB */
    VP56mv vector_candidate[2];
    int vector_candidate_pos;

    /* filtering hints */
    int filter_header;               /* used in vp6 only */
    int deblock_filtering;
    int filter_selection;
    int filter_mode;
    int max_vector_length;
    int sample_variance_threshold;

    uint8_t coeff_ctx[4][64];              /* used in vp5 only */
    uint8_t coeff_ctx_last[4];             /* used in vp5 only */

    int has_alpha;

    /* upside-down flipping hints */
    int flip;  /* are we flipping ? */
    int frbi;  /* first row block index in MB */
    int srbi;  /* second row block index in MB */
    int stride[4];  /* stride for each plan */

    const uint8_t *vp56_coord_div;
    VP56ParseVectorAdjustment parse_vector_adjustment;
    VP56Filter filter;
    VP56ParseCoeff parse_coeff;
    VP56DefaultModelsInit default_models_init;
    VP56ParseVectorModels parse_vector_models;
    VP56ParseCoeffModels parse_coeff_models;
    VP56ParseHeader parse_header;

    VP56Model *modelp;
    VP56Model models[2];

    /* huffman decoding */
    int use_huffman;
    GetBitContext gb;
    VLC dccv_vlc[2];
    VLC runv_vlc[2];
    VLC ract_vlc[2][3][6];
    unsigned int nb_null[2][2];       /* number of consecutive NULL DC/AC */
};


void vp56_init(AVCodecContext *avctx, int flip, int has_alpha);
int vp56_free(AVCodecContext *avctx);
void vp56_init_dequant(VP56Context *s, int quantizer);
int vp56_decode_frame(AVCodecContext *avctx, void *data, int *data_size,
                      AVPacket *avpkt);


/**
 * vp56 specific range coder implementation
 */

static inline void vp56_init_range_decoder(VP56RangeCoder *c,
                                           const uint8_t *buf, int buf_size)
{
    c->high = 255;
    c->bits = -8;
    c->buffer = buf;
    c->end = buf + buf_size;
    c->code_word = bytestream_get_be16(&c->buffer);
}

static inline int vp56_rac_get_prob(VP56RangeCoder *c, uint8_t prob)
{
    /* Don't put c->high in a local variable; if we do that, gcc gets
     * the stupids and turns the code below into a branch again. */
    int bits = c->bits;
    unsigned long code_word = c->code_word;
    unsigned int low = 1 + (((c->high - 1) * prob) >> 8);
    unsigned int low_shift = low << 8;
    int bit = code_word >= low_shift;
    int shift;

    /* Incantation to convince GCC to turn these into conditional moves
     * instead of branches -- faster, as this branch is basically
     * unpredictable. */
    c->high = bit ? c->high - low : low;
    code_word = bit ? code_word - low_shift : code_word;

    /* normalize */
    shift = ff_h264_norm_shift[c->high] - 1;
    c->high   <<= shift;
    code_word <<= shift;
    bits       += shift;
    if(bits >= 0 && c->buffer < c->end) {
        code_word |= *c->buffer++ << bits;
        bits -= 8;
    }
    c->bits = bits;
    c->code_word = code_word;
    return bit;
}

static inline int vp56_rac_get(VP56RangeCoder *c)
{
    /* equiprobable */
    int low = (c->high + 1) >> 1;
    unsigned int low_shift = low << 8;
    int bit = c->code_word >= low_shift;
    if (bit) {
        c->high = (c->high - low) << 1;
        c->code_word -= low_shift;
    } else {
        c->high = low << 1;
    }

    /* normalize */
    c->code_word <<= 1;
    if (++c->bits == 0 && c->buffer < c->end) {
        c->bits = -8;
        c->code_word |= *c->buffer++;
    }
    return bit;
}

// rounding is different than vp56_rac_get, is vp56_rac_get wrong?
static inline int vp8_rac_get(VP56RangeCoder *c)
{
    return vp56_rac_get_prob(c, 128);
}

static inline int vp56_rac_gets(VP56RangeCoder *c, int bits)
{
    int value = 0;

    while (bits--) {
        value = (value << 1) | vp56_rac_get(c);
    }

    return value;
}

static inline int vp8_rac_get_uint(VP56RangeCoder *c, int bits)
{
    int value = 0;

    while (bits--) {
        value = (value << 1) | vp8_rac_get(c);
    }

    return value;
}

// fixme: add 1 bit to all the calls to this?
static inline int vp8_rac_get_sint(VP56RangeCoder *c, int bits)
{
    int v;

    if (!vp8_rac_get(c))
        return 0;

    v = vp8_rac_get_uint(c, bits);

    if (vp8_rac_get(c))
        v = -v;

    return v;
}

// P(7)
static inline int vp56_rac_gets_nn(VP56RangeCoder *c, int bits)
{
    int v = vp56_rac_gets(c, 7) << 1;
    return v + !v;
}

static inline int vp8_rac_get_nn(VP56RangeCoder *c)
{
    int v = vp8_rac_get_uint(c, 7) << 1;
    return v + !v;
}

static inline int vp56_rac_get_tree(VP56RangeCoder *c,
                                    const VP56Tree *tree,
                                    const uint8_t *probs)
{
    while (tree->val > 0) {
        if (vp56_rac_get_prob(c, probs[tree->prob_idx]))
            tree += tree->val;
        else
            tree++;
    }
    return -tree->val;
}

/**
 * This is identical to vp8_rac_get_tree except for the possibility of starting
 * on a node other than the root node, needed for coeff decode where this is
 * used to save a bit after a 0 token (by disallowing EOB to immediately follow.)
 */
static inline int vp8_rac_get_tree_with_offset(VP56RangeCoder *c, const int8_t (*tree)[2],
                                               const uint8_t *probs, int i)
{
    do {
        i = tree[i][vp56_rac_get_prob(c, probs[i])];
    } while (i > 0);

    return -i;
}

// how probabilities are associated with decisions is different I think
// well, the new scheme fits in the old but this way has one fewer branches per decision
static inline int vp8_rac_get_tree(VP56RangeCoder *c, const int8_t (*tree)[2],
                                   const uint8_t *probs)
{
    return vp8_rac_get_tree_with_offset(c, tree, probs, 0);
}

// DCTextra
static inline int vp8_rac_get_coeff(VP56RangeCoder *c, const uint8_t *prob)
{
    int v = 0;

    do {
        v = (v<<1) + vp56_rac_get_prob(c, *prob++);
    } while (*prob);

    return v;
}

#endif /* AVCODEC_VP56_H */