2013-10-01 06:03:30 +03:00
|
|
|
/*
|
|
|
|
* VP9 compatible video decoder
|
|
|
|
*
|
|
|
|
* Copyright (C) 2013 Ronald S. Bultje <rsbultje gmail com>
|
|
|
|
* Copyright (C) 2013 Clément Bœsch <u pkh me>
|
|
|
|
*
|
|
|
|
* This file is part of Libav.
|
|
|
|
*
|
|
|
|
* Libav 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.
|
|
|
|
*
|
|
|
|
* Libav 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 Libav; if not, write to the Free Software
|
|
|
|
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
|
|
|
|
*/
|
|
|
|
|
|
|
|
#ifndef AVCODEC_VP9_H
|
|
|
|
#define AVCODEC_VP9_H
|
|
|
|
|
|
|
|
#include <stddef.h>
|
|
|
|
#include <stdint.h>
|
|
|
|
|
2013-11-23 18:27:18 +03:00
|
|
|
#include "libavutil/buffer.h"
|
2013-10-01 06:03:30 +03:00
|
|
|
#include "libavutil/internal.h"
|
|
|
|
|
|
|
|
#include "avcodec.h"
|
2013-11-23 18:27:18 +03:00
|
|
|
#include "thread.h"
|
2013-10-01 06:03:30 +03:00
|
|
|
#include "vp56.h"
|
|
|
|
|
|
|
|
enum TxfmMode {
|
|
|
|
TX_4X4,
|
|
|
|
TX_8X8,
|
|
|
|
TX_16X16,
|
|
|
|
TX_32X32,
|
|
|
|
N_TXFM_SIZES,
|
|
|
|
TX_SWITCHABLE = N_TXFM_SIZES,
|
|
|
|
N_TXFM_MODES
|
|
|
|
};
|
|
|
|
|
|
|
|
enum TxfmType {
|
|
|
|
DCT_DCT,
|
|
|
|
DCT_ADST,
|
|
|
|
ADST_DCT,
|
|
|
|
ADST_ADST,
|
|
|
|
N_TXFM_TYPES
|
|
|
|
};
|
|
|
|
|
|
|
|
enum IntraPredMode {
|
|
|
|
VERT_PRED,
|
|
|
|
HOR_PRED,
|
|
|
|
DC_PRED,
|
|
|
|
DIAG_DOWN_LEFT_PRED,
|
|
|
|
DIAG_DOWN_RIGHT_PRED,
|
|
|
|
VERT_RIGHT_PRED,
|
|
|
|
HOR_DOWN_PRED,
|
|
|
|
VERT_LEFT_PRED,
|
|
|
|
HOR_UP_PRED,
|
|
|
|
TM_VP8_PRED,
|
|
|
|
LEFT_DC_PRED,
|
|
|
|
TOP_DC_PRED,
|
|
|
|
DC_128_PRED,
|
|
|
|
DC_127_PRED,
|
|
|
|
DC_129_PRED,
|
|
|
|
N_INTRA_PRED_MODES
|
|
|
|
};
|
|
|
|
|
|
|
|
enum FilterMode {
|
|
|
|
FILTER_8TAP_SMOOTH,
|
|
|
|
FILTER_8TAP_REGULAR,
|
|
|
|
FILTER_8TAP_SHARP,
|
|
|
|
FILTER_BILINEAR,
|
|
|
|
FILTER_SWITCHABLE,
|
|
|
|
};
|
|
|
|
|
|
|
|
enum BlockPartition {
|
|
|
|
PARTITION_NONE, // [ ] <-.
|
|
|
|
PARTITION_H, // [-] |
|
|
|
|
PARTITION_V, // [|] |
|
|
|
|
PARTITION_SPLIT, // [+] --'
|
|
|
|
};
|
|
|
|
|
|
|
|
enum InterPredMode {
|
|
|
|
NEARESTMV = 10,
|
|
|
|
NEARMV = 11,
|
|
|
|
ZEROMV = 12,
|
|
|
|
NEWMV = 13,
|
|
|
|
};
|
|
|
|
|
|
|
|
enum MVJoint {
|
|
|
|
MV_JOINT_ZERO,
|
|
|
|
MV_JOINT_H,
|
|
|
|
MV_JOINT_V,
|
|
|
|
MV_JOINT_HV,
|
|
|
|
};
|
|
|
|
|
|
|
|
typedef struct ProbContext {
|
|
|
|
uint8_t y_mode[4][9];
|
|
|
|
uint8_t uv_mode[10][9];
|
|
|
|
uint8_t filter[4][2];
|
|
|
|
uint8_t mv_mode[7][3];
|
|
|
|
uint8_t intra[4];
|
|
|
|
uint8_t comp[5];
|
|
|
|
uint8_t single_ref[5][2];
|
|
|
|
uint8_t comp_ref[5];
|
|
|
|
uint8_t tx32p[2][3];
|
|
|
|
uint8_t tx16p[2][2];
|
|
|
|
uint8_t tx8p[2];
|
|
|
|
uint8_t skip[3];
|
|
|
|
uint8_t mv_joint[3];
|
|
|
|
struct {
|
|
|
|
uint8_t sign;
|
|
|
|
uint8_t classes[10];
|
|
|
|
uint8_t class0;
|
|
|
|
uint8_t bits[10];
|
|
|
|
uint8_t class0_fp[2][3];
|
|
|
|
uint8_t fp[3];
|
|
|
|
uint8_t class0_hp;
|
|
|
|
uint8_t hp;
|
|
|
|
} mv_comp[2];
|
|
|
|
uint8_t partition[4][4][3];
|
|
|
|
} ProbContext;
|
|
|
|
|
2016-11-03 09:12:02 +02:00
|
|
|
typedef void (*vp9_mc_func)(uint8_t *dst, ptrdiff_t dst_stride,
|
|
|
|
const uint8_t *ref, ptrdiff_t ref_stride,
|
2013-10-01 06:03:30 +03:00
|
|
|
int h, int mx, int my);
|
|
|
|
|
|
|
|
typedef struct VP9DSPContext {
|
|
|
|
/*
|
|
|
|
* dimension 1: 0=4x4, 1=8x8, 2=16x16, 3=32x32
|
|
|
|
* dimension 2: intra prediction modes
|
|
|
|
*
|
|
|
|
* dst/left/top is aligned by transform-size (i.e. 4, 8, 16 or 32 pixels)
|
|
|
|
* stride is aligned by 16 pixels
|
|
|
|
* top[-1] is top/left; top[4,7] is top-right for 4x4
|
|
|
|
*/
|
|
|
|
// FIXME(rbultje) maybe replace left/top pointers with HAVE_TOP/
|
|
|
|
// HAVE_LEFT/HAVE_TOPRIGHT flags instead, and then handle it in-place?
|
2016-04-27 19:45:23 +02:00
|
|
|
// also needs to fit in with what H.264/VP8/etc do
|
2013-10-01 06:03:30 +03:00
|
|
|
void (*intra_pred[N_TXFM_SIZES][N_INTRA_PRED_MODES])(uint8_t *dst,
|
|
|
|
ptrdiff_t stride,
|
|
|
|
const uint8_t *left,
|
|
|
|
const uint8_t *top);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* dimension 1: 0=4x4, 1=8x8, 2=16x16, 3=32x32, 4=lossless (3-4=dct only)
|
|
|
|
* dimension 2: 0=dct/dct, 1=dct/adst, 2=adst/dct, 3=adst/adst
|
|
|
|
*
|
|
|
|
* dst is aligned by transform-size (i.e. 4, 8, 16 or 32 pixels)
|
|
|
|
* stride is aligned by 16 pixels
|
|
|
|
* block is 16-byte aligned
|
|
|
|
* eob indicates the position (+1) of the last non-zero coefficient,
|
|
|
|
* in scan-order. This can be used to write faster versions, e.g. a
|
|
|
|
* dc-only 4x4/8x8/16x16/32x32, or a 4x4-only (eob<10) 8x8/16x16/32x32,
|
|
|
|
* etc.
|
|
|
|
*/
|
|
|
|
// FIXME also write idct_add_block() versions for whole (inter) pred
|
|
|
|
// blocks, so we can do 2 4x4s at once
|
|
|
|
void (*itxfm_add[N_TXFM_SIZES + 1][N_TXFM_TYPES])(uint8_t *dst,
|
|
|
|
ptrdiff_t stride,
|
|
|
|
int16_t *block, int eob);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* dimension 1: width of filter (0=4, 1=8, 2=16)
|
|
|
|
* dimension 2: 0=col-edge filter (h), 1=row-edge filter (v)
|
|
|
|
*
|
|
|
|
* dst/stride are aligned by 8
|
|
|
|
*/
|
|
|
|
void (*loop_filter_8[3][2])(uint8_t *dst, ptrdiff_t stride,
|
|
|
|
int mb_lim, int lim, int hev_thr);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* dimension 1: 0=col-edge filter (h), 1=row-edge filter (v)
|
|
|
|
*
|
|
|
|
* The width of filter is assumed to be 16; dst/stride are aligned by 16
|
|
|
|
*/
|
|
|
|
void (*loop_filter_16[2])(uint8_t *dst, ptrdiff_t stride,
|
|
|
|
int mb_lim, int lim, int hev_thr);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* dimension 1/2: width of filter (0=4, 1=8) for each filter half
|
|
|
|
* dimension 3: 0=col-edge filter (h), 1=row-edge filter (v)
|
|
|
|
*
|
|
|
|
* dst/stride are aligned by operation size
|
|
|
|
* this basically calls loop_filter[d1][d3][0](), followed by
|
|
|
|
* loop_filter[d2][d3][0]() on the next 8 pixels
|
|
|
|
* mb_lim/lim/hev_thr contain two values in the lowest two bytes of the
|
|
|
|
* integer.
|
|
|
|
*/
|
|
|
|
// FIXME perhaps a mix4 that operates on 32px (for AVX2)
|
|
|
|
void (*loop_filter_mix2[2][2][2])(uint8_t *dst, ptrdiff_t stride,
|
|
|
|
int mb_lim, int lim, int hev_thr);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* dimension 1: hsize (0: 64, 1: 32, 2: 16, 3: 8, 4: 4)
|
|
|
|
* dimension 2: filter type (0: smooth, 1: regular, 2: sharp, 3: bilin)
|
|
|
|
* dimension 3: averaging type (0: put, 1: avg)
|
|
|
|
* dimension 4: x subpel interpolation (0: none, 1: 8tap/bilin)
|
|
|
|
* dimension 5: y subpel interpolation (1: none, 1: 8tap/bilin)
|
|
|
|
*
|
|
|
|
* dst/stride are aligned by hsize
|
|
|
|
*/
|
|
|
|
vp9_mc_func mc[5][4][2][2][2];
|
|
|
|
} VP9DSPContext;
|
|
|
|
|
|
|
|
enum CompPredMode {
|
|
|
|
PRED_SINGLEREF,
|
|
|
|
PRED_COMPREF,
|
|
|
|
PRED_SWITCHABLE,
|
|
|
|
};
|
|
|
|
|
|
|
|
typedef struct VP9MVRefPair {
|
|
|
|
VP56mv mv[2];
|
|
|
|
int8_t ref[2];
|
|
|
|
} VP9MVRefPair;
|
|
|
|
|
|
|
|
typedef struct VP9Filter {
|
|
|
|
uint8_t level[8 * 8];
|
|
|
|
uint8_t /* bit=col */ mask[2 /* 0=y, 1=uv */][2 /* 0=col, 1=row */]
|
|
|
|
[8 /* rows */][4 /* 0=16, 1=8, 2=4, 3=inner4 */];
|
|
|
|
} VP9Filter;
|
|
|
|
|
2013-11-23 18:27:18 +03:00
|
|
|
typedef struct VP9Frame {
|
|
|
|
ThreadFrame tf;
|
|
|
|
|
|
|
|
uint8_t *segmentation_map;
|
|
|
|
VP9MVRefPair *mv;
|
|
|
|
|
|
|
|
AVBufferRef *segmentation_map_buf;
|
|
|
|
AVBufferRef *mv_buf;
|
|
|
|
} VP9Frame;
|
|
|
|
|
2013-10-01 06:03:30 +03:00
|
|
|
enum BlockLevel {
|
|
|
|
BL_64X64,
|
|
|
|
BL_32X32,
|
|
|
|
BL_16X16,
|
|
|
|
BL_8X8,
|
|
|
|
};
|
|
|
|
|
|
|
|
enum BlockSize {
|
|
|
|
BS_64x64,
|
|
|
|
BS_64x32,
|
|
|
|
BS_32x64,
|
|
|
|
BS_32x32,
|
|
|
|
BS_32x16,
|
|
|
|
BS_16x32,
|
|
|
|
BS_16x16,
|
|
|
|
BS_16x8,
|
|
|
|
BS_8x16,
|
|
|
|
BS_8x8,
|
|
|
|
BS_8x4,
|
|
|
|
BS_4x8,
|
|
|
|
BS_4x4,
|
|
|
|
N_BS_SIZES,
|
|
|
|
};
|
|
|
|
|
|
|
|
typedef struct VP9Block {
|
|
|
|
uint8_t seg_id, intra, comp, ref[2], mode[4], uvmode, skip;
|
|
|
|
enum FilterMode filter;
|
|
|
|
VP56mv mv[4 /* b_idx */][2 /* ref */];
|
|
|
|
enum BlockSize bs;
|
|
|
|
enum TxfmMode tx, uvtx;
|
|
|
|
|
|
|
|
int row, row7, col, col7;
|
|
|
|
uint8_t *dst[3];
|
|
|
|
ptrdiff_t y_stride, uv_stride;
|
2016-08-02 07:55:31 +02:00
|
|
|
|
|
|
|
enum BlockLevel bl;
|
|
|
|
enum BlockPartition bp;
|
2013-10-01 06:03:30 +03:00
|
|
|
} VP9Block;
|
|
|
|
|
|
|
|
typedef struct VP9Context {
|
|
|
|
VP9DSPContext dsp;
|
|
|
|
VideoDSPContext vdsp;
|
|
|
|
GetBitContext gb;
|
|
|
|
VP56RangeCoder c;
|
|
|
|
VP56RangeCoder *c_b;
|
|
|
|
unsigned c_b_size;
|
2013-11-23 20:10:12 +03:00
|
|
|
VP9Block *b;
|
|
|
|
VP9Block *b_base;
|
2013-10-01 06:03:30 +03:00
|
|
|
|
2016-08-02 07:55:31 +02:00
|
|
|
int alloc_width;
|
|
|
|
int alloc_height;
|
|
|
|
|
|
|
|
int pass;
|
|
|
|
int uses_2pass;
|
|
|
|
int last_uses_2pass;
|
|
|
|
int setup_finished;
|
|
|
|
|
2013-10-01 06:03:30 +03:00
|
|
|
// bitstream header
|
|
|
|
uint8_t profile;
|
|
|
|
uint8_t keyframe, last_keyframe;
|
|
|
|
uint8_t invisible;
|
|
|
|
uint8_t use_last_frame_mvs;
|
|
|
|
uint8_t errorres;
|
|
|
|
uint8_t colorspace;
|
2015-05-27 16:06:57 +02:00
|
|
|
uint8_t sub_x;
|
|
|
|
uint8_t sub_y;
|
2013-10-01 06:03:30 +03:00
|
|
|
uint8_t fullrange;
|
|
|
|
uint8_t intraonly;
|
|
|
|
uint8_t resetctx;
|
|
|
|
uint8_t refreshrefmask;
|
|
|
|
uint8_t highprecisionmvs;
|
|
|
|
enum FilterMode filtermode;
|
|
|
|
uint8_t allowcompinter;
|
|
|
|
uint8_t fixcompref;
|
|
|
|
uint8_t refreshctx;
|
|
|
|
uint8_t parallelmode;
|
|
|
|
uint8_t framectxid;
|
|
|
|
uint8_t refidx[3];
|
|
|
|
uint8_t signbias[3];
|
|
|
|
uint8_t varcompref[2];
|
2013-11-23 18:27:18 +03:00
|
|
|
|
|
|
|
ThreadFrame refs[8];
|
|
|
|
|
|
|
|
#define CUR_FRAME 0
|
|
|
|
#define LAST_FRAME 1
|
|
|
|
VP9Frame frames[2];
|
2013-10-01 06:03:30 +03:00
|
|
|
|
|
|
|
struct {
|
|
|
|
uint8_t level;
|
|
|
|
int8_t sharpness;
|
|
|
|
uint8_t lim_lut[64];
|
|
|
|
uint8_t mblim_lut[64];
|
|
|
|
} filter;
|
|
|
|
struct {
|
|
|
|
uint8_t enabled;
|
|
|
|
int8_t mode[2];
|
|
|
|
int8_t ref[4];
|
|
|
|
} lf_delta;
|
|
|
|
uint8_t yac_qi;
|
|
|
|
int8_t ydc_qdelta, uvdc_qdelta, uvac_qdelta;
|
|
|
|
uint8_t lossless;
|
|
|
|
struct {
|
|
|
|
uint8_t enabled;
|
|
|
|
uint8_t temporal;
|
|
|
|
uint8_t absolute_vals;
|
|
|
|
uint8_t update_map;
|
2014-11-11 14:27:01 +02:00
|
|
|
#define MAX_SEGMENT 8
|
2013-10-01 06:03:30 +03:00
|
|
|
struct {
|
|
|
|
uint8_t q_enabled;
|
|
|
|
uint8_t lf_enabled;
|
|
|
|
uint8_t ref_enabled;
|
|
|
|
uint8_t skip_enabled;
|
|
|
|
uint8_t ref_val;
|
|
|
|
int16_t q_val;
|
|
|
|
int8_t lf_val;
|
|
|
|
int16_t qmul[2][2];
|
|
|
|
uint8_t lflvl[4][2];
|
2014-11-11 14:27:01 +02:00
|
|
|
} feat[MAX_SEGMENT];
|
2013-10-01 06:03:30 +03:00
|
|
|
} segmentation;
|
|
|
|
struct {
|
|
|
|
unsigned log2_tile_cols, log2_tile_rows;
|
|
|
|
unsigned tile_cols, tile_rows;
|
|
|
|
unsigned tile_row_start, tile_row_end, tile_col_start, tile_col_end;
|
|
|
|
} tiling;
|
|
|
|
unsigned sb_cols, sb_rows, rows, cols;
|
|
|
|
struct {
|
|
|
|
ProbContext p;
|
|
|
|
uint8_t coef[4][2][2][6][6][3];
|
|
|
|
} prob_ctx[4];
|
|
|
|
struct {
|
|
|
|
ProbContext p;
|
|
|
|
uint8_t coef[4][2][2][6][6][11];
|
|
|
|
uint8_t seg[7];
|
|
|
|
uint8_t segpred[3];
|
|
|
|
} prob;
|
|
|
|
struct {
|
|
|
|
unsigned y_mode[4][10];
|
|
|
|
unsigned uv_mode[10][10];
|
|
|
|
unsigned filter[4][3];
|
|
|
|
unsigned mv_mode[7][4];
|
|
|
|
unsigned intra[4][2];
|
|
|
|
unsigned comp[5][2];
|
|
|
|
unsigned single_ref[5][2][2];
|
|
|
|
unsigned comp_ref[5][2];
|
|
|
|
unsigned tx32p[2][4];
|
|
|
|
unsigned tx16p[2][3];
|
|
|
|
unsigned tx8p[2][2];
|
|
|
|
unsigned skip[3][2];
|
|
|
|
unsigned mv_joint[4];
|
|
|
|
struct {
|
|
|
|
unsigned sign[2];
|
|
|
|
unsigned classes[11];
|
|
|
|
unsigned class0[2];
|
|
|
|
unsigned bits[10][2];
|
|
|
|
unsigned class0_fp[2][4];
|
|
|
|
unsigned fp[4];
|
|
|
|
unsigned class0_hp[2];
|
|
|
|
unsigned hp[2];
|
|
|
|
} mv_comp[2];
|
|
|
|
unsigned partition[4][4][4];
|
|
|
|
unsigned coef[4][2][2][6][6][3];
|
|
|
|
unsigned eob[4][2][2][6][6][2];
|
|
|
|
} counts;
|
|
|
|
enum TxfmMode txfmmode;
|
|
|
|
enum CompPredMode comppredmode;
|
|
|
|
|
|
|
|
// contextual (left/above) cache
|
|
|
|
uint8_t left_partition_ctx[8], *above_partition_ctx;
|
|
|
|
uint8_t left_mode_ctx[16], *above_mode_ctx;
|
|
|
|
// FIXME maybe merge some of the below in a flags field?
|
|
|
|
uint8_t left_y_nnz_ctx[16], *above_y_nnz_ctx;
|
|
|
|
uint8_t left_uv_nnz_ctx[2][8], *above_uv_nnz_ctx[2];
|
|
|
|
uint8_t left_skip_ctx[8], *above_skip_ctx; // 1bit
|
|
|
|
uint8_t left_txfm_ctx[8], *above_txfm_ctx; // 2bit
|
|
|
|
uint8_t left_segpred_ctx[8], *above_segpred_ctx; // 1bit
|
|
|
|
uint8_t left_intra_ctx[8], *above_intra_ctx; // 1bit
|
|
|
|
uint8_t left_comp_ctx[8], *above_comp_ctx; // 1bit
|
|
|
|
uint8_t left_ref_ctx[8], *above_ref_ctx; // 2bit
|
|
|
|
uint8_t left_filter_ctx[8], *above_filter_ctx;
|
|
|
|
VP56mv left_mv_ctx[16][2], (*above_mv_ctx)[2];
|
|
|
|
|
|
|
|
// whole-frame cache
|
|
|
|
uint8_t *intra_pred_data[3];
|
|
|
|
VP9Filter *lflvl;
|
arm: vp9: Add NEON optimizations of VP9 MC functions
This work is sponsored by, and copyright, Google.
The filter coefficients are signed values, where the product of the
multiplication with one individual filter coefficient doesn't
overflow a 16 bit signed value (the largest filter coefficient is
127). But when the products are accumulated, the resulting sum can
overflow the 16 bit signed range. Instead of accumulating in 32 bit,
we accumulate the largest product (either index 3 or 4) last with a
saturated addition.
(The VP8 MC asm does something similar, but slightly simpler, by
accumulating each half of the filter separately. In the VP9 MC
filters, each half of the filter can also overflow though, so the
largest component has to be handled individually.)
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_avg4_neon: 1.71 1.15 1.42 1.49
vp9_avg8_neon: 2.51 3.63 3.14 2.58
vp9_avg16_neon: 2.95 6.76 3.01 2.84
vp9_avg32_neon: 3.29 6.64 2.85 3.00
vp9_avg64_neon: 3.47 6.67 3.14 2.80
vp9_avg_8tap_smooth_4h_neon: 3.22 4.73 2.76 4.67
vp9_avg_8tap_smooth_4hv_neon: 3.67 4.76 3.28 4.71
vp9_avg_8tap_smooth_4v_neon: 5.52 7.60 4.60 6.31
vp9_avg_8tap_smooth_8h_neon: 6.22 9.04 5.12 9.32
vp9_avg_8tap_smooth_8hv_neon: 6.38 8.21 5.72 8.17
vp9_avg_8tap_smooth_8v_neon: 9.22 12.66 8.15 11.10
vp9_avg_8tap_smooth_64h_neon: 7.02 10.23 5.54 11.58
vp9_avg_8tap_smooth_64hv_neon: 6.76 9.46 5.93 9.40
vp9_avg_8tap_smooth_64v_neon: 10.76 14.13 9.46 13.37
vp9_put4_neon: 1.11 1.47 1.00 1.21
vp9_put8_neon: 1.23 2.17 1.94 1.48
vp9_put16_neon: 1.63 4.02 1.73 1.97
vp9_put32_neon: 1.56 4.92 2.00 1.96
vp9_put64_neon: 2.10 5.28 2.03 2.35
vp9_put_8tap_smooth_4h_neon: 3.11 4.35 2.63 4.35
vp9_put_8tap_smooth_4hv_neon: 3.67 4.69 3.25 4.71
vp9_put_8tap_smooth_4v_neon: 5.45 7.27 4.49 6.52
vp9_put_8tap_smooth_8h_neon: 5.97 8.18 4.81 8.56
vp9_put_8tap_smooth_8hv_neon: 6.39 7.90 5.64 8.15
vp9_put_8tap_smooth_8v_neon: 9.03 11.84 8.07 11.51
vp9_put_8tap_smooth_64h_neon: 6.78 9.48 4.88 10.89
vp9_put_8tap_smooth_64hv_neon: 6.99 8.87 5.94 9.56
vp9_put_8tap_smooth_64v_neon: 10.69 13.30 9.43 14.34
For the larger 8tap filters, the speedup vs C code is around 5-14x.
This is significantly faster than libvpx's implementation of the same
functions, at least when comparing the put_8tap_smooth_64 functions
(compared to vpx_convolve8_horiz_neon and vpx_convolve8_vert_neon from
libvpx).
Absolute runtimes from checkasm:
Cortex A7 A8 A9 A53
vp9_put_8tap_smooth_64h_neon: 20150.3 14489.4 19733.6 10863.7
libvpx vpx_convolve8_horiz_neon: 52623.3 19736.4 21907.7 25027.7
vp9_put_8tap_smooth_64v_neon: 14455.0 12303.9 13746.4 9628.9
libvpx vpx_convolve8_vert_neon: 42090.0 17706.2 17659.9 16941.2
Thus, on the A9, the horizontal filter is only marginally faster than
libvpx, while our version is significantly faster on the other cores,
and the vertical filter is significantly faster on all cores. The
difference is especially large on the A7.
The libvpx implementation does the accumulation in 32 bit, which
probably explains most of the differences.
Signed-off-by: Martin Storsjö <martin@martin.st>
2016-11-03 09:12:08 +02:00
|
|
|
// This requires 64 + 8 rows, with 80 bytes stride
|
|
|
|
DECLARE_ALIGNED(32, uint8_t, edge_emu_buffer)[72 * 80];
|
2013-10-01 06:03:30 +03:00
|
|
|
|
|
|
|
// block reconstruction intermediates
|
2013-11-23 20:10:12 +03:00
|
|
|
int16_t *block_base, *block, *uvblock_base[2], *uvblock[2];
|
|
|
|
uint8_t *eob_base, *uveob_base[2], *eob, *uveob[2];
|
2014-01-07 15:24:03 +03:00
|
|
|
struct { int x, y; } min_mv, max_mv;
|
2013-10-01 06:03:30 +03:00
|
|
|
DECLARE_ALIGNED(32, uint8_t, tmp_y)[64 * 64];
|
|
|
|
DECLARE_ALIGNED(32, uint8_t, tmp_uv)[2][32 * 32];
|
|
|
|
} VP9Context;
|
|
|
|
|
2016-11-05 13:18:53 +02:00
|
|
|
extern const int8_t ff_vp9_subpel_filters[3][15][8];
|
|
|
|
|
2013-10-01 06:03:30 +03:00
|
|
|
void ff_vp9dsp_init(VP9DSPContext *dsp);
|
|
|
|
|
2016-11-10 11:06:26 +02:00
|
|
|
void ff_vp9dsp_init_aarch64(VP9DSPContext *dsp);
|
arm: vp9: Add NEON optimizations of VP9 MC functions
This work is sponsored by, and copyright, Google.
The filter coefficients are signed values, where the product of the
multiplication with one individual filter coefficient doesn't
overflow a 16 bit signed value (the largest filter coefficient is
127). But when the products are accumulated, the resulting sum can
overflow the 16 bit signed range. Instead of accumulating in 32 bit,
we accumulate the largest product (either index 3 or 4) last with a
saturated addition.
(The VP8 MC asm does something similar, but slightly simpler, by
accumulating each half of the filter separately. In the VP9 MC
filters, each half of the filter can also overflow though, so the
largest component has to be handled individually.)
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_avg4_neon: 1.71 1.15 1.42 1.49
vp9_avg8_neon: 2.51 3.63 3.14 2.58
vp9_avg16_neon: 2.95 6.76 3.01 2.84
vp9_avg32_neon: 3.29 6.64 2.85 3.00
vp9_avg64_neon: 3.47 6.67 3.14 2.80
vp9_avg_8tap_smooth_4h_neon: 3.22 4.73 2.76 4.67
vp9_avg_8tap_smooth_4hv_neon: 3.67 4.76 3.28 4.71
vp9_avg_8tap_smooth_4v_neon: 5.52 7.60 4.60 6.31
vp9_avg_8tap_smooth_8h_neon: 6.22 9.04 5.12 9.32
vp9_avg_8tap_smooth_8hv_neon: 6.38 8.21 5.72 8.17
vp9_avg_8tap_smooth_8v_neon: 9.22 12.66 8.15 11.10
vp9_avg_8tap_smooth_64h_neon: 7.02 10.23 5.54 11.58
vp9_avg_8tap_smooth_64hv_neon: 6.76 9.46 5.93 9.40
vp9_avg_8tap_smooth_64v_neon: 10.76 14.13 9.46 13.37
vp9_put4_neon: 1.11 1.47 1.00 1.21
vp9_put8_neon: 1.23 2.17 1.94 1.48
vp9_put16_neon: 1.63 4.02 1.73 1.97
vp9_put32_neon: 1.56 4.92 2.00 1.96
vp9_put64_neon: 2.10 5.28 2.03 2.35
vp9_put_8tap_smooth_4h_neon: 3.11 4.35 2.63 4.35
vp9_put_8tap_smooth_4hv_neon: 3.67 4.69 3.25 4.71
vp9_put_8tap_smooth_4v_neon: 5.45 7.27 4.49 6.52
vp9_put_8tap_smooth_8h_neon: 5.97 8.18 4.81 8.56
vp9_put_8tap_smooth_8hv_neon: 6.39 7.90 5.64 8.15
vp9_put_8tap_smooth_8v_neon: 9.03 11.84 8.07 11.51
vp9_put_8tap_smooth_64h_neon: 6.78 9.48 4.88 10.89
vp9_put_8tap_smooth_64hv_neon: 6.99 8.87 5.94 9.56
vp9_put_8tap_smooth_64v_neon: 10.69 13.30 9.43 14.34
For the larger 8tap filters, the speedup vs C code is around 5-14x.
This is significantly faster than libvpx's implementation of the same
functions, at least when comparing the put_8tap_smooth_64 functions
(compared to vpx_convolve8_horiz_neon and vpx_convolve8_vert_neon from
libvpx).
Absolute runtimes from checkasm:
Cortex A7 A8 A9 A53
vp9_put_8tap_smooth_64h_neon: 20150.3 14489.4 19733.6 10863.7
libvpx vpx_convolve8_horiz_neon: 52623.3 19736.4 21907.7 25027.7
vp9_put_8tap_smooth_64v_neon: 14455.0 12303.9 13746.4 9628.9
libvpx vpx_convolve8_vert_neon: 42090.0 17706.2 17659.9 16941.2
Thus, on the A9, the horizontal filter is only marginally faster than
libvpx, while our version is significantly faster on the other cores,
and the vertical filter is significantly faster on all cores. The
difference is especially large on the A7.
The libvpx implementation does the accumulation in 32 bit, which
probably explains most of the differences.
Signed-off-by: Martin Storsjö <martin@martin.st>
2016-11-03 09:12:08 +02:00
|
|
|
void ff_vp9dsp_init_arm(VP9DSPContext *dsp);
|
2013-10-01 06:03:30 +03:00
|
|
|
void ff_vp9dsp_init_x86(VP9DSPContext *dsp);
|
|
|
|
|
|
|
|
void ff_vp9_fill_mv(VP9Context *s, VP56mv *mv, int mode, int sb);
|
|
|
|
|
|
|
|
void ff_vp9_adapt_probs(VP9Context *s);
|
|
|
|
|
|
|
|
int ff_vp9_decode_block(AVCodecContext *avctx, int row, int col,
|
|
|
|
VP9Filter *lflvl, ptrdiff_t yoff, ptrdiff_t uvoff,
|
|
|
|
enum BlockLevel bl, enum BlockPartition bp);
|
|
|
|
|
|
|
|
#endif /* AVCODEC_VP9_H */
|