Chiebot-Mirror
/
FFmpeg
mirror of https://github.com/FFmpeg/FFmpeg.git
8
0
Fork 0
Code Issues Projects Releases Wiki Activity

Move RV3/4-specific DSP functions into their own context

Browse Source 
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
pull/2/head
Kostya Shishkov 14 years ago committed by Ronald S. Bultje
parent 21d7037234
commit d241f51e0f
10 changed files with 253 additions and 178 deletions
Whitespace
Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL
Unified View
Diff Options
Show Stats Download Patch File Download Diff File
  1. 28
      libavcodec/dsputil.c
  2. 18
      libavcodec/dsputil.h
  3. 80
      libavcodec/rv30dsp.c
  4. 39
      libavcodec/rv34.c
  5. 2
      libavcodec/rv34.h
  6. 44
      libavcodec/rv34dsp.h
  7. 138
      libavcodec/rv40dsp.c
  8. 2
      libavcodec/x86/Makefile
  9. 20
      libavcodec/x86/dsputil_mmx.c
  10. 60
      libavcodec/x86/rv40dsp.c

28
libavcodec/dsputil.c
Unescape View File

@ -1280,16 +1280,16 @@ static void wmv2_mspel8_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int
} }
#if CONFIG_RV40_DECODER #if CONFIG_RV40_DECODER
static void put_rv40_qpel16_mc33_c(uint8_t *dst, uint8_t *src, int stride){ void ff_put_rv40_qpel16_mc33_c(uint8_t *dst, uint8_t *src, int stride){
put_pixels16_xy2_8_c(dst, src, stride, 16); put_pixels16_xy2_8_c(dst, src, stride, 16);
} }
static void avg_rv40_qpel16_mc33_c(uint8_t *dst, uint8_t *src, int stride){ void ff_avg_rv40_qpel16_mc33_c(uint8_t *dst, uint8_t *src, int stride){
avg_pixels16_xy2_8_c(dst, src, stride, 16); avg_pixels16_xy2_8_c(dst, src, stride, 16);
} }
static void put_rv40_qpel8_mc33_c(uint8_t *dst, uint8_t *src, int stride){ void ff_put_rv40_qpel8_mc33_c(uint8_t *dst, uint8_t *src, int stride){
put_pixels8_xy2_8_c(dst, src, stride, 8); put_pixels8_xy2_8_c(dst, src, stride, 8);
} }
static void avg_rv40_qpel8_mc33_c(uint8_t *dst, uint8_t *src, int stride){ void ff_avg_rv40_qpel8_mc33_c(uint8_t *dst, uint8_t *src, int stride){
avg_pixels8_xy2_8_c(dst, src, stride, 8); avg_pixels8_xy2_8_c(dst, src, stride, 8);
} }
#endif /* CONFIG_RV40_DECODER */ #endif /* CONFIG_RV40_DECODER */
@ -2903,16 +2903,6 @@ av_cold void dsputil_init(DSPContext* c, AVCodecContext *avctx)
#if CONFIG_WMV2_DECODER || CONFIG_VC1_DECODER #if CONFIG_WMV2_DECODER || CONFIG_VC1_DECODER
ff_intrax8dsp_init(c,avctx); ff_intrax8dsp_init(c,avctx);
#endif #endif
#if CONFIG_RV30_DECODER
ff_rv30dsp_init(c,avctx);
#endif
#if CONFIG_RV40_DECODER
ff_rv40dsp_init(c,avctx);
c->put_rv40_qpel_pixels_tab[0][15] = put_rv40_qpel16_mc33_c;
c->avg_rv40_qpel_pixels_tab[0][15] = avg_rv40_qpel16_mc33_c;
c->put_rv40_qpel_pixels_tab[1][15] = put_rv40_qpel8_mc33_c;
c->avg_rv40_qpel_pixels_tab[1][15] = avg_rv40_qpel8_mc33_c;
#endif
c->put_mspel_pixels_tab[0]= ff_put_pixels8x8_c; c->put_mspel_pixels_tab[0]= ff_put_pixels8x8_c;
c->put_mspel_pixels_tab[1]= put_mspel8_mc10_c; c->put_mspel_pixels_tab[1]= put_mspel8_mc10_c;
@ -3124,16 +3114,6 @@ av_cold void dsputil_init(DSPContext* c, AVCodecContext *avctx)
c->avg_2tap_qpel_pixels_tab[0][i]= c->avg_h264_qpel_pixels_tab[0][i]; c->avg_2tap_qpel_pixels_tab[0][i]= c->avg_h264_qpel_pixels_tab[0][i];
} }
c->put_rv30_tpel_pixels_tab[0][0] = c->put_h264_qpel_pixels_tab[0][0];
c->put_rv30_tpel_pixels_tab[1][0] = c->put_h264_qpel_pixels_tab[1][0];
c->avg_rv30_tpel_pixels_tab[0][0] = c->avg_h264_qpel_pixels_tab[0][0];
c->avg_rv30_tpel_pixels_tab[1][0] = c->avg_h264_qpel_pixels_tab[1][0];
c->put_rv40_qpel_pixels_tab[0][0] = c->put_h264_qpel_pixels_tab[0][0];
c->put_rv40_qpel_pixels_tab[1][0] = c->put_h264_qpel_pixels_tab[1][0];
c->avg_rv40_qpel_pixels_tab[0][0] = c->avg_h264_qpel_pixels_tab[0][0];
c->avg_rv40_qpel_pixels_tab[1][0] = c->avg_h264_qpel_pixels_tab[1][0];
switch(c->idct_permutation_type){ switch(c->idct_permutation_type){
case FF_NO_IDCT_PERM: case FF_NO_IDCT_PERM:
for(i=0; i<64; i++) for(i=0; i<64; i++)

18
libavcodec/dsputil.h
Unescape View File

@ -114,6 +114,12 @@ void ff_vp3_h_loop_filter_c(uint8_t *src, int stride, int *bounding_values);
/* EA functions */ /* EA functions */
void ff_ea_idct_put_c(uint8_t *dest, int linesize, DCTELEM *block); void ff_ea_idct_put_c(uint8_t *dest, int linesize, DCTELEM *block);
/* RV40 functions */
void ff_put_rv40_qpel16_mc33_c(uint8_t *dst, uint8_t *src, int stride);
void ff_avg_rv40_qpel16_mc33_c(uint8_t *dst, uint8_t *src, int stride);
void ff_put_rv40_qpel8_mc33_c(uint8_t *dst, uint8_t *src, int stride);
void ff_avg_rv40_qpel8_mc33_c(uint8_t *dst, uint8_t *src, int stride);
/* 1/2^n downscaling functions from imgconvert.c */ /* 1/2^n downscaling functions from imgconvert.c */
void ff_shrink22(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_shrink44(uint8_t *dst, int dst_wrap, const uint8_t *src, int src_wrap, int width, int height);
@ -542,16 +548,6 @@ typedef struct DSPContext {
void (*vector_clip_int32)(int32_t *dst, const int32_t *src, int32_t min, void (*vector_clip_int32)(int32_t *dst, const int32_t *src, int32_t min,
int32_t max, unsigned int len); int32_t max, unsigned int len);
/* rv30 functions */
qpel_mc_func put_rv30_tpel_pixels_tab[4][16];
qpel_mc_func avg_rv30_tpel_pixels_tab[4][16];
/* 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];
op_fill_func fill_block_tab[2]; op_fill_func fill_block_tab[2];
} DSPContext; } DSPContext;
@ -626,8 +622,6 @@ void dsputil_init_sh4(DSPContext* c, AVCodecContext *avctx);
void dsputil_init_vis(DSPContext* c, AVCodecContext *avctx); void dsputil_init_vis(DSPContext* c, AVCodecContext *avctx);
void ff_dsputil_init_dwt(DSPContext *c); void ff_dsputil_init_dwt(DSPContext *c);
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_intrax8dsp_init(DSPContext* c, AVCodecContext *avctx);
void ff_mlp_init(DSPContext* c, AVCodecContext *avctx); void ff_mlp_init(DSPContext* c, AVCodecContext *avctx);
void ff_mlp_init_x86(DSPContext* c, AVCodecContext *avctx); void ff_mlp_init_x86(DSPContext* c, AVCodecContext *avctx);

80
libavcodec/rv30dsp.c
Unescape View File

@ -26,6 +26,7 @@
#include "avcodec.h" #include "avcodec.h"
#include "dsputil.h" #include "dsputil.h"
#include "rv34dsp.h"
#define RV30_LOWPASS(OPNAME, OP) \ #define RV30_LOWPASS(OPNAME, OP) \
static av_unused void OPNAME ## rv30_tpel8_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, const int C1, const int C2){\ static av_unused void OPNAME ## rv30_tpel8_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, const int C1, const int C2){\
@ -251,41 +252,46 @@ RV30_MC(put_, 16)
RV30_MC(avg_, 8) RV30_MC(avg_, 8)
RV30_MC(avg_, 16) RV30_MC(avg_, 16)
av_cold void ff_rv30dsp_init(DSPContext* c, AVCodecContext *avctx) { av_cold void ff_rv30dsp_init(RV34DSPContext *c, DSPContext* dsp) {
c->put_rv30_tpel_pixels_tab[0][ 0] = c->put_h264_qpel_pixels_tab[0][0]; c->put_pixels_tab[0][ 0] = dsp->put_h264_qpel_pixels_tab[0][0];
c->put_rv30_tpel_pixels_tab[0][ 1] = put_rv30_tpel16_mc10_c; c->put_pixels_tab[0][ 1] = put_rv30_tpel16_mc10_c;
c->put_rv30_tpel_pixels_tab[0][ 2] = put_rv30_tpel16_mc20_c; c->put_pixels_tab[0][ 2] = put_rv30_tpel16_mc20_c;
c->put_rv30_tpel_pixels_tab[0][ 4] = put_rv30_tpel16_mc01_c; c->put_pixels_tab[0][ 4] = put_rv30_tpel16_mc01_c;
c->put_rv30_tpel_pixels_tab[0][ 5] = put_rv30_tpel16_mc11_c; c->put_pixels_tab[0][ 5] = put_rv30_tpel16_mc11_c;
c->put_rv30_tpel_pixels_tab[0][ 6] = put_rv30_tpel16_mc21_c; c->put_pixels_tab[0][ 6] = put_rv30_tpel16_mc21_c;
c->put_rv30_tpel_pixels_tab[0][ 8] = put_rv30_tpel16_mc02_c; c->put_pixels_tab[0][ 8] = put_rv30_tpel16_mc02_c;
c->put_rv30_tpel_pixels_tab[0][ 9] = put_rv30_tpel16_mc12_c; c->put_pixels_tab[0][ 9] = put_rv30_tpel16_mc12_c;
c->put_rv30_tpel_pixels_tab[0][10] = put_rv30_tpel16_mc22_c; c->put_pixels_tab[0][10] = put_rv30_tpel16_mc22_c;
c->avg_rv30_tpel_pixels_tab[0][ 0] = c->avg_h264_qpel_pixels_tab[0][0]; c->avg_pixels_tab[0][ 0] = dsp->avg_h264_qpel_pixels_tab[0][0];
c->avg_rv30_tpel_pixels_tab[0][ 1] = avg_rv30_tpel16_mc10_c; c->avg_pixels_tab[0][ 1] = avg_rv30_tpel16_mc10_c;
c->avg_rv30_tpel_pixels_tab[0][ 2] = avg_rv30_tpel16_mc20_c; c->avg_pixels_tab[0][ 2] = avg_rv30_tpel16_mc20_c;
c->avg_rv30_tpel_pixels_tab[0][ 4] = avg_rv30_tpel16_mc01_c; c->avg_pixels_tab[0][ 4] = avg_rv30_tpel16_mc01_c;
c->avg_rv30_tpel_pixels_tab[0][ 5] = avg_rv30_tpel16_mc11_c; c->avg_pixels_tab[0][ 5] = avg_rv30_tpel16_mc11_c;
c->avg_rv30_tpel_pixels_tab[0][ 6] = avg_rv30_tpel16_mc21_c; c->avg_pixels_tab[0][ 6] = avg_rv30_tpel16_mc21_c;
c->avg_rv30_tpel_pixels_tab[0][ 8] = avg_rv30_tpel16_mc02_c; c->avg_pixels_tab[0][ 8] = avg_rv30_tpel16_mc02_c;
c->avg_rv30_tpel_pixels_tab[0][ 9] = avg_rv30_tpel16_mc12_c; c->avg_pixels_tab[0][ 9] = avg_rv30_tpel16_mc12_c;
c->avg_rv30_tpel_pixels_tab[0][10] = avg_rv30_tpel16_mc22_c; c->avg_pixels_tab[0][10] = avg_rv30_tpel16_mc22_c;
c->put_rv30_tpel_pixels_tab[1][ 0] = c->put_h264_qpel_pixels_tab[1][0]; c->put_pixels_tab[1][ 0] = dsp->put_h264_qpel_pixels_tab[1][0];
c->put_rv30_tpel_pixels_tab[1][ 1] = put_rv30_tpel8_mc10_c; c->put_pixels_tab[1][ 1] = put_rv30_tpel8_mc10_c;
c->put_rv30_tpel_pixels_tab[1][ 2] = put_rv30_tpel8_mc20_c; c->put_pixels_tab[1][ 2] = put_rv30_tpel8_mc20_c;
c->put_rv30_tpel_pixels_tab[1][ 4] = put_rv30_tpel8_mc01_c; c->put_pixels_tab[1][ 4] = put_rv30_tpel8_mc01_c;
c->put_rv30_tpel_pixels_tab[1][ 5] = put_rv30_tpel8_mc11_c; c->put_pixels_tab[1][ 5] = put_rv30_tpel8_mc11_c;
c->put_rv30_tpel_pixels_tab[1][ 6] = put_rv30_tpel8_mc21_c; c->put_pixels_tab[1][ 6] = put_rv30_tpel8_mc21_c;
c->put_rv30_tpel_pixels_tab[1][ 8] = put_rv30_tpel8_mc02_c; c->put_pixels_tab[1][ 8] = put_rv30_tpel8_mc02_c;
c->put_rv30_tpel_pixels_tab[1][ 9] = put_rv30_tpel8_mc12_c; c->put_pixels_tab[1][ 9] = put_rv30_tpel8_mc12_c;
c->put_rv30_tpel_pixels_tab[1][10] = put_rv30_tpel8_mc22_c; c->put_pixels_tab[1][10] = put_rv30_tpel8_mc22_c;
c->avg_rv30_tpel_pixels_tab[1][ 0] = c->avg_h264_qpel_pixels_tab[1][0]; c->avg_pixels_tab[1][ 0] = dsp->avg_h264_qpel_pixels_tab[1][0];
c->avg_rv30_tpel_pixels_tab[1][ 1] = avg_rv30_tpel8_mc10_c; c->avg_pixels_tab[1][ 1] = avg_rv30_tpel8_mc10_c;
c->avg_rv30_tpel_pixels_tab[1][ 2] = avg_rv30_tpel8_mc20_c; c->avg_pixels_tab[1][ 2] = avg_rv30_tpel8_mc20_c;
c->avg_rv30_tpel_pixels_tab[1][ 4] = avg_rv30_tpel8_mc01_c; c->avg_pixels_tab[1][ 4] = avg_rv30_tpel8_mc01_c;
c->avg_rv30_tpel_pixels_tab[1][ 5] = avg_rv30_tpel8_mc11_c; c->avg_pixels_tab[1][ 5] = avg_rv30_tpel8_mc11_c;
c->avg_rv30_tpel_pixels_tab[1][ 6] = avg_rv30_tpel8_mc21_c; c->avg_pixels_tab[1][ 6] = avg_rv30_tpel8_mc21_c;
c->avg_rv30_tpel_pixels_tab[1][ 8] = avg_rv30_tpel8_mc02_c; c->avg_pixels_tab[1][ 8] = avg_rv30_tpel8_mc02_c;
c->avg_rv30_tpel_pixels_tab[1][ 9] = avg_rv30_tpel8_mc12_c; c->avg_pixels_tab[1][ 9] = avg_rv30_tpel8_mc12_c;
c->avg_rv30_tpel_pixels_tab[1][10] = avg_rv30_tpel8_mc22_c; c->avg_pixels_tab[1][10] = avg_rv30_tpel8_mc22_c;
c->put_chroma_pixels_tab[0] = dsp->put_h264_chroma_pixels_tab[0];
c->put_chroma_pixels_tab[1] = dsp->put_h264_chroma_pixels_tab[1];
c->avg_chroma_pixels_tab[0] = dsp->avg_h264_chroma_pixels_tab[0];
c->avg_chroma_pixels_tab[1] = dsp->avg_h264_chroma_pixels_tab[1];
} }

39
libavcodec/rv34.c
Unescape View File

@ -809,24 +809,18 @@ static void rv34_mc_1mv(RV34DecContext *r, const int block_type,
const int width, const int height, int dir) const int width, const int height, int dir)
{ {
rv34_mc(r, block_type, xoff, yoff, mv_off, width, height, dir, r->rv30, rv34_mc(r, block_type, xoff, yoff, mv_off, width, height, dir, r->rv30,
r->rv30 ? r->s.dsp.put_rv30_tpel_pixels_tab r->rdsp.put_pixels_tab,
: r->s.dsp.put_rv40_qpel_pixels_tab, r->rdsp.put_chroma_pixels_tab);
r->rv30 ? r->s.dsp.put_h264_chroma_pixels_tab
: r->s.dsp.put_rv40_chroma_pixels_tab);
} }
static void rv34_mc_2mv(RV34DecContext *r, const int block_type) static void rv34_mc_2mv(RV34DecContext *r, const int block_type)
{ {
rv34_mc(r, block_type, 0, 0, 0, 2, 2, 0, r->rv30, rv34_mc(r, block_type, 0, 0, 0, 2, 2, 0, r->rv30,
r->rv30 ? r->s.dsp.put_rv30_tpel_pixels_tab r->rdsp.put_pixels_tab,
: r->s.dsp.put_rv40_qpel_pixels_tab, r->rdsp.put_chroma_pixels_tab);
r->rv30 ? r->s.dsp.put_h264_chroma_pixels_tab
: r->s.dsp.put_rv40_chroma_pixels_tab);
rv34_mc(r, block_type, 0, 0, 0, 2, 2, 1, r->rv30, rv34_mc(r, block_type, 0, 0, 0, 2, 2, 1, r->rv30,
r->rv30 ? r->s.dsp.avg_rv30_tpel_pixels_tab r->rdsp.avg_pixels_tab,
: r->s.dsp.avg_rv40_qpel_pixels_tab, r->rdsp.avg_chroma_pixels_tab);
r->rv30 ? r->s.dsp.avg_h264_chroma_pixels_tab
: r->s.dsp.avg_rv40_chroma_pixels_tab);
} }
static void rv34_mc_2mv_skip(RV34DecContext *r) static void rv34_mc_2mv_skip(RV34DecContext *r)
@ -835,15 +829,11 @@ static void rv34_mc_2mv_skip(RV34DecContext *r)
for(j = 0; j < 2; j++) for(j = 0; j < 2; j++)
for(i = 0; i < 2; i++){ for(i = 0; i < 2; i++){
rv34_mc(r, RV34_MB_P_8x8, i*8, j*8, i+j*r->s.b8_stride, 1, 1, 0, r->rv30, rv34_mc(r, RV34_MB_P_8x8, i*8, j*8, i+j*r->s.b8_stride, 1, 1, 0, r->rv30,
r->rv30 ? r->s.dsp.put_rv30_tpel_pixels_tab r->rdsp.put_pixels_tab,
: r->s.dsp.put_rv40_qpel_pixels_tab, r->rdsp.put_chroma_pixels_tab);
r->rv30 ? r->s.dsp.put_h264_chroma_pixels_tab
: r->s.dsp.put_rv40_chroma_pixels_tab);
rv34_mc(r, RV34_MB_P_8x8, i*8, j*8, i+j*r->s.b8_stride, 1, 1, 1, r->rv30, rv34_mc(r, RV34_MB_P_8x8, i*8, j*8, i+j*r->s.b8_stride, 1, 1, 1, r->rv30,
r->rv30 ? r->s.dsp.avg_rv30_tpel_pixels_tab r->rdsp.avg_pixels_tab,
: r->s.dsp.avg_rv40_qpel_pixels_tab, r->rdsp.avg_chroma_pixels_tab);
r->rv30 ? r->s.dsp.avg_h264_chroma_pixels_tab
: r->s.dsp.avg_rv40_chroma_pixels_tab);
} }
} }
@ -1363,6 +1353,15 @@ av_cold int ff_rv34_decode_init(AVCodecContext *avctx)
ff_h264_pred_init(&r->h, CODEC_ID_RV40, 8); ff_h264_pred_init(&r->h, CODEC_ID_RV40, 8);
#if CONFIG_RV30_DECODER
if (avctx->codec_id == CODEC_ID_RV30)
ff_rv30dsp_init(&r->rdsp, &r->s.dsp);
#endif
#if CONFIG_RV40_DECODER
if (avctx->codec_id == CODEC_ID_RV40)
ff_rv40dsp_init(&r->rdsp, &r->s.dsp);
#endif
r->intra_types_stride = 4*s->mb_stride + 4; r->intra_types_stride = 4*s->mb_stride + 4;
r->intra_types_hist = av_malloc(r->intra_types_stride * 4 * 2 * sizeof(*r->intra_types_hist)); r->intra_types_hist = av_malloc(r->intra_types_stride * 4 * 2 * sizeof(*r->intra_types_hist));
r->intra_types = r->intra_types_hist + r->intra_types_stride * 4; r->intra_types = r->intra_types_hist + r->intra_types_stride * 4;

2
libavcodec/rv34.h
Unescape View File

@ -32,6 +32,7 @@
#include "mpegvideo.h" #include "mpegvideo.h"
#include "h264pred.h" #include "h264pred.h"
#include "rv34dsp.h"
#define MB_TYPE_SEPARATE_DC 0x01000000 #define MB_TYPE_SEPARATE_DC 0x01000000
#define IS_SEPARATE_DC(a) ((a) & MB_TYPE_SEPARATE_DC) #define IS_SEPARATE_DC(a) ((a) & MB_TYPE_SEPARATE_DC)
@ -83,6 +84,7 @@ typedef struct SliceInfo{
/** decoder context */ /** decoder context */
typedef struct RV34DecContext{ typedef struct RV34DecContext{
MpegEncContext s; MpegEncContext s;
RV34DSPContext rdsp;
int8_t *intra_types_hist;///< old block types, used for prediction int8_t *intra_types_hist;///< old block types, used for prediction
int8_t *intra_types; ///< block types int8_t *intra_types; ///< block types
int intra_types_stride;///< block types array stride int intra_types_stride;///< block types array stride

44
libavcodec/rv34dsp.h
Unescape View File

@ -0,0 +1,44 @@
/*
* RV30/40 decoder motion compensation functions
* Copyright (c) 2008 Konstantin Shishkov
*
* 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
*/
/**
* @file
* RV30/40 decoder motion compensation functions
*/
#ifndef AVCODEC_RV34DSP_H
#define AVCODEC_RV34DSP_H
#include "dsputil.h"
typedef struct RV34DSPContext {
qpel_mc_func put_pixels_tab[4][16];
qpel_mc_func avg_pixels_tab[4][16];
h264_chroma_mc_func put_chroma_pixels_tab[3];
h264_chroma_mc_func avg_chroma_pixels_tab[3];
} RV34DSPContext;
void ff_rv30dsp_init(RV34DSPContext *c, DSPContext* dsp);
void ff_rv40dsp_init(RV34DSPContext *c, DSPContext* dsp);
void ff_rv40dsp_init_x86(RV34DSPContext *c, DSPContext *dsp);
#endif /* AVCODEC_RV34DSP_H */

138
libavcodec/rv40dsp.c
Unescape View File

@ -26,6 +26,7 @@
#include "avcodec.h" #include "avcodec.h"
#include "dsputil.h" #include "dsputil.h"
#include "rv34dsp.h"
#define RV40_LOWPASS(OPNAME, OP) \ #define RV40_LOWPASS(OPNAME, OP) \
static av_unused void OPNAME ## rv40_qpel8_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride,\ static av_unused void OPNAME ## rv40_qpel8_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride,\
@ -284,70 +285,77 @@ static void OPNAME ## rv40_chroma_mc8_c(uint8_t *dst/*align 8*/, uint8_t *src/*a
RV40_CHROMA_MC(put_, op_put) RV40_CHROMA_MC(put_, op_put)
RV40_CHROMA_MC(avg_, op_avg) RV40_CHROMA_MC(avg_, op_avg)
void ff_rv40dsp_init(DSPContext* c, AVCodecContext *avctx) { av_cold void ff_rv40dsp_init(RV34DSPContext *c, DSPContext* dsp) {
c->put_rv40_qpel_pixels_tab[0][ 0] = c->put_h264_qpel_pixels_tab[0][0]; c->put_pixels_tab[0][ 0] = dsp->put_h264_qpel_pixels_tab[0][0];
c->put_rv40_qpel_pixels_tab[0][ 1] = put_rv40_qpel16_mc10_c; c->put_pixels_tab[0][ 1] = put_rv40_qpel16_mc10_c;
c->put_rv40_qpel_pixels_tab[0][ 2] = put_rv40_qpel16_mc20_c; c->put_pixels_tab[0][ 2] = put_rv40_qpel16_mc20_c;
c->put_rv40_qpel_pixels_tab[0][ 3] = put_rv40_qpel16_mc30_c; c->put_pixels_tab[0][ 3] = put_rv40_qpel16_mc30_c;
c->put_rv40_qpel_pixels_tab[0][ 4] = put_rv40_qpel16_mc01_c; c->put_pixels_tab[0][ 4] = put_rv40_qpel16_mc01_c;
c->put_rv40_qpel_pixels_tab[0][ 5] = put_rv40_qpel16_mc11_c; c->put_pixels_tab[0][ 5] = put_rv40_qpel16_mc11_c;
c->put_rv40_qpel_pixels_tab[0][ 6] = put_rv40_qpel16_mc21_c; c->put_pixels_tab[0][ 6] = put_rv40_qpel16_mc21_c;
c->put_rv40_qpel_pixels_tab[0][ 7] = put_rv40_qpel16_mc31_c; c->put_pixels_tab[0][ 7] = put_rv40_qpel16_mc31_c;
c->put_rv40_qpel_pixels_tab[0][ 8] = put_rv40_qpel16_mc02_c; c->put_pixels_tab[0][ 8] = put_rv40_qpel16_mc02_c;
c->put_rv40_qpel_pixels_tab[0][ 9] = put_rv40_qpel16_mc12_c; c->put_pixels_tab[0][ 9] = put_rv40_qpel16_mc12_c;
c->put_rv40_qpel_pixels_tab[0][10] = put_rv40_qpel16_mc22_c; c->put_pixels_tab[0][10] = put_rv40_qpel16_mc22_c;
c->put_rv40_qpel_pixels_tab[0][11] = put_rv40_qpel16_mc32_c; c->put_pixels_tab[0][11] = put_rv40_qpel16_mc32_c;
c->put_rv40_qpel_pixels_tab[0][12] = put_rv40_qpel16_mc03_c; c->put_pixels_tab[0][12] = put_rv40_qpel16_mc03_c;
c->put_rv40_qpel_pixels_tab[0][13] = put_rv40_qpel16_mc13_c; c->put_pixels_tab[0][13] = put_rv40_qpel16_mc13_c;
c->put_rv40_qpel_pixels_tab[0][14] = put_rv40_qpel16_mc23_c; c->put_pixels_tab[0][14] = put_rv40_qpel16_mc23_c;
c->avg_rv40_qpel_pixels_tab[0][ 0] = c->avg_h264_qpel_pixels_tab[0][0]; c->put_pixels_tab[0][15] = ff_put_rv40_qpel16_mc33_c;
c->avg_rv40_qpel_pixels_tab[0][ 1] = avg_rv40_qpel16_mc10_c; c->avg_pixels_tab[0][ 0] = dsp->avg_h264_qpel_pixels_tab[0][0];
c->avg_rv40_qpel_pixels_tab[0][ 2] = avg_rv40_qpel16_mc20_c; c->avg_pixels_tab[0][ 1] = avg_rv40_qpel16_mc10_c;
c->avg_rv40_qpel_pixels_tab[0][ 3] = avg_rv40_qpel16_mc30_c; c->avg_pixels_tab[0][ 2] = avg_rv40_qpel16_mc20_c;
c->avg_rv40_qpel_pixels_tab[0][ 4] = avg_rv40_qpel16_mc01_c; c->avg_pixels_tab[0][ 3] = avg_rv40_qpel16_mc30_c;
c->avg_rv40_qpel_pixels_tab[0][ 5] = avg_rv40_qpel16_mc11_c; c->avg_pixels_tab[0][ 4] = avg_rv40_qpel16_mc01_c;
c->avg_rv40_qpel_pixels_tab[0][ 6] = avg_rv40_qpel16_mc21_c; c->avg_pixels_tab[0][ 5] = avg_rv40_qpel16_mc11_c;
c->avg_rv40_qpel_pixels_tab[0][ 7] = avg_rv40_qpel16_mc31_c; c->avg_pixels_tab[0][ 6] = avg_rv40_qpel16_mc21_c;
c->avg_rv40_qpel_pixels_tab[0][ 8] = avg_rv40_qpel16_mc02_c; c->avg_pixels_tab[0][ 7] = avg_rv40_qpel16_mc31_c;
c->avg_rv40_qpel_pixels_tab[0][ 9] = avg_rv40_qpel16_mc12_c; c->avg_pixels_tab[0][ 8] = avg_rv40_qpel16_mc02_c;
c->avg_rv40_qpel_pixels_tab[0][10] = avg_rv40_qpel16_mc22_c; c->avg_pixels_tab[0][ 9] = avg_rv40_qpel16_mc12_c;
c->avg_rv40_qpel_pixels_tab[0][11] = avg_rv40_qpel16_mc32_c; c->avg_pixels_tab[0][10] = avg_rv40_qpel16_mc22_c;
c->avg_rv40_qpel_pixels_tab[0][12] = avg_rv40_qpel16_mc03_c; c->avg_pixels_tab[0][11] = avg_rv40_qpel16_mc32_c;
c->avg_rv40_qpel_pixels_tab[0][13] = avg_rv40_qpel16_mc13_c; c->avg_pixels_tab[0][12] = avg_rv40_qpel16_mc03_c;
c->avg_rv40_qpel_pixels_tab[0][14] = avg_rv40_qpel16_mc23_c; c->avg_pixels_tab[0][13] = avg_rv40_qpel16_mc13_c;
c->put_rv40_qpel_pixels_tab[1][ 0] = c->put_h264_qpel_pixels_tab[1][0]; c->avg_pixels_tab[0][14] = avg_rv40_qpel16_mc23_c;
c->put_rv40_qpel_pixels_tab[1][ 1] = put_rv40_qpel8_mc10_c; c->avg_pixels_tab[0][15] = ff_avg_rv40_qpel16_mc33_c;
c->put_rv40_qpel_pixels_tab[1][ 2] = put_rv40_qpel8_mc20_c; c->put_pixels_tab[1][ 0] = dsp->put_h264_qpel_pixels_tab[1][0];
c->put_rv40_qpel_pixels_tab[1][ 3] = put_rv40_qpel8_mc30_c; c->put_pixels_tab[1][ 1] = put_rv40_qpel8_mc10_c;
c->put_rv40_qpel_pixels_tab[1][ 4] = put_rv40_qpel8_mc01_c; c->put_pixels_tab[1][ 2] = put_rv40_qpel8_mc20_c;
c->put_rv40_qpel_pixels_tab[1][ 5] = put_rv40_qpel8_mc11_c; c->put_pixels_tab[1][ 3] = put_rv40_qpel8_mc30_c;
c->put_rv40_qpel_pixels_tab[1][ 6] = put_rv40_qpel8_mc21_c; c->put_pixels_tab[1][ 4] = put_rv40_qpel8_mc01_c;
c->put_rv40_qpel_pixels_tab[1][ 7] = put_rv40_qpel8_mc31_c; c->put_pixels_tab[1][ 5] = put_rv40_qpel8_mc11_c;
c->put_rv40_qpel_pixels_tab[1][ 8] = put_rv40_qpel8_mc02_c; c->put_pixels_tab[1][ 6] = put_rv40_qpel8_mc21_c;
c->put_rv40_qpel_pixels_tab[1][ 9] = put_rv40_qpel8_mc12_c; c->put_pixels_tab[1][ 7] = put_rv40_qpel8_mc31_c;
c->put_rv40_qpel_pixels_tab[1][10] = put_rv40_qpel8_mc22_c; c->put_pixels_tab[1][ 8] = put_rv40_qpel8_mc02_c;
c->put_rv40_qpel_pixels_tab[1][11] = put_rv40_qpel8_mc32_c; c->put_pixels_tab[1][ 9] = put_rv40_qpel8_mc12_c;
c->put_rv40_qpel_pixels_tab[1][12] = put_rv40_qpel8_mc03_c; c->put_pixels_tab[1][10] = put_rv40_qpel8_mc22_c;
c->put_rv40_qpel_pixels_tab[1][13] = put_rv40_qpel8_mc13_c; c->put_pixels_tab[1][11] = put_rv40_qpel8_mc32_c;
c->put_rv40_qpel_pixels_tab[1][14] = put_rv40_qpel8_mc23_c; c->put_pixels_tab[1][12] = put_rv40_qpel8_mc03_c;
c->avg_rv40_qpel_pixels_tab[1][ 0] = c->avg_h264_qpel_pixels_tab[1][0]; c->put_pixels_tab[1][13] = put_rv40_qpel8_mc13_c;
c->avg_rv40_qpel_pixels_tab[1][ 1] = avg_rv40_qpel8_mc10_c; c->put_pixels_tab[1][14] = put_rv40_qpel8_mc23_c;
c->avg_rv40_qpel_pixels_tab[1][ 2] = avg_rv40_qpel8_mc20_c; c->put_pixels_tab[1][15] = ff_put_rv40_qpel8_mc33_c;
c->avg_rv40_qpel_pixels_tab[1][ 3] = avg_rv40_qpel8_mc30_c; c->avg_pixels_tab[1][ 0] = dsp->avg_h264_qpel_pixels_tab[1][0];
c->avg_rv40_qpel_pixels_tab[1][ 4] = avg_rv40_qpel8_mc01_c; c->avg_pixels_tab[1][ 1] = avg_rv40_qpel8_mc10_c;
c->avg_rv40_qpel_pixels_tab[1][ 5] = avg_rv40_qpel8_mc11_c; c->avg_pixels_tab[1][ 2] = avg_rv40_qpel8_mc20_c;
c->avg_rv40_qpel_pixels_tab[1][ 6] = avg_rv40_qpel8_mc21_c; c->avg_pixels_tab[1][ 3] = avg_rv40_qpel8_mc30_c;
c->avg_rv40_qpel_pixels_tab[1][ 7] = avg_rv40_qpel8_mc31_c; c->avg_pixels_tab[1][ 4] = avg_rv40_qpel8_mc01_c;
c->avg_rv40_qpel_pixels_tab[1][ 8] = avg_rv40_qpel8_mc02_c; c->avg_pixels_tab[1][ 5] = avg_rv40_qpel8_mc11_c;
c->avg_rv40_qpel_pixels_tab[1][ 9] = avg_rv40_qpel8_mc12_c; c->avg_pixels_tab[1][ 6] = avg_rv40_qpel8_mc21_c;
c->avg_rv40_qpel_pixels_tab[1][10] = avg_rv40_qpel8_mc22_c; c->avg_pixels_tab[1][ 7] = avg_rv40_qpel8_mc31_c;
c->avg_rv40_qpel_pixels_tab[1][11] = avg_rv40_qpel8_mc32_c; c->avg_pixels_tab[1][ 8] = avg_rv40_qpel8_mc02_c;
c->avg_rv40_qpel_pixels_tab[1][12] = avg_rv40_qpel8_mc03_c; c->avg_pixels_tab[1][ 9] = avg_rv40_qpel8_mc12_c;
c->avg_rv40_qpel_pixels_tab[1][13] = avg_rv40_qpel8_mc13_c; c->avg_pixels_tab[1][10] = avg_rv40_qpel8_mc22_c;
c->avg_rv40_qpel_pixels_tab[1][14] = avg_rv40_qpel8_mc23_c; c->avg_pixels_tab[1][11] = avg_rv40_qpel8_mc32_c;
c->avg_pixels_tab[1][12] = avg_rv40_qpel8_mc03_c;
c->avg_pixels_tab[1][13] = avg_rv40_qpel8_mc13_c;
c->avg_pixels_tab[1][14] = avg_rv40_qpel8_mc23_c;
c->avg_pixels_tab[1][15] = ff_avg_rv40_qpel8_mc33_c;
c->put_rv40_chroma_pixels_tab[0] = put_rv40_chroma_mc8_c; c->put_chroma_pixels_tab[0] = put_rv40_chroma_mc8_c;
c->put_rv40_chroma_pixels_tab[1] = put_rv40_chroma_mc4_c; c->put_chroma_pixels_tab[1] = put_rv40_chroma_mc4_c;
c->avg_rv40_chroma_pixels_tab[0] = avg_rv40_chroma_mc8_c; c->avg_chroma_pixels_tab[0] = avg_rv40_chroma_mc8_c;
c->avg_rv40_chroma_pixels_tab[1] = avg_rv40_chroma_mc4_c; c->avg_chroma_pixels_tab[1] = avg_rv40_chroma_mc4_c;
if (HAVE_MMX)
ff_rv40dsp_init_x86(c, dsp);
} }

2
libavcodec/x86/Makefile
Unescape View File

@ -21,6 +21,8 @@ YASM-OBJS-$(CONFIG_H264PRED) += x86/h264_intrapred.o \
x86/h264_intrapred_10bit.o x86/h264_intrapred_10bit.o
MMX-OBJS-$(CONFIG_H264PRED) += x86/h264_intrapred_init.o MMX-OBJS-$(CONFIG_H264PRED) += x86/h264_intrapred_init.o
MMX-OBJS-$(CONFIG_RV40_DECODER) += x86/rv40dsp.o \
YASM-OBJS-$(CONFIG_VC1_DECODER) += x86/vc1dsp_yasm.o YASM-OBJS-$(CONFIG_VC1_DECODER) += x86/vc1dsp_yasm.o
MMX-OBJS-$(CONFIG_AC3DSP) += x86/ac3dsp_mmx.o MMX-OBJS-$(CONFIG_AC3DSP) += x86/ac3dsp_mmx.o

20
libavcodec/x86/dsputil_mmx.c
Unescape View File

@ -1895,29 +1895,17 @@ PREFETCH(prefetch_3dnow, prefetch)
void ff_put_h264_chroma_mc8_mmx_rnd (uint8_t *dst, uint8_t *src, void ff_put_h264_chroma_mc8_mmx_rnd (uint8_t *dst, uint8_t *src,
int stride, int h, int x, int y); int stride, int h, int x, int y);
void ff_put_rv40_chroma_mc8_mmx (uint8_t *dst, uint8_t *src,
int stride, int h, int x, int y);
void ff_avg_h264_chroma_mc8_mmx2_rnd (uint8_t *dst, uint8_t *src, void ff_avg_h264_chroma_mc8_mmx2_rnd (uint8_t *dst, uint8_t *src,
int stride, int h, int x, int y); int stride, int h, int x, int y);
void ff_avg_rv40_chroma_mc8_mmx2 (uint8_t *dst, uint8_t *src,
int stride, int h, int x, int y);
void ff_avg_h264_chroma_mc8_3dnow_rnd (uint8_t *dst, uint8_t *src, void ff_avg_h264_chroma_mc8_3dnow_rnd (uint8_t *dst, uint8_t *src,
int stride, int h, int x, int y); int stride, int h, int x, int y);
void ff_avg_rv40_chroma_mc8_3dnow (uint8_t *dst, uint8_t *src,
int stride, int h, int x, int y);
void ff_put_h264_chroma_mc4_mmx (uint8_t *dst, uint8_t *src, void ff_put_h264_chroma_mc4_mmx (uint8_t *dst, uint8_t *src,
int stride, int h, int x, int y); int stride, int h, int x, int y);
void ff_put_rv40_chroma_mc4_mmx (uint8_t *dst, uint8_t *src,
int stride, int h, int x, int y);
void ff_avg_h264_chroma_mc4_mmx2 (uint8_t *dst, uint8_t *src, void ff_avg_h264_chroma_mc4_mmx2 (uint8_t *dst, uint8_t *src,
int stride, int h, int x, int y); int stride, int h, int x, int y);
void ff_avg_rv40_chroma_mc4_mmx2 (uint8_t *dst, uint8_t *src,
int stride, int h, int x, int y);
void ff_avg_h264_chroma_mc4_3dnow (uint8_t *dst, uint8_t *src, void ff_avg_h264_chroma_mc4_3dnow (uint8_t *dst, uint8_t *src,
int stride, int h, int x, int y); int stride, int h, int x, int y);
void ff_avg_rv40_chroma_mc4_3dnow (uint8_t *dst, uint8_t *src,
int stride, int h, int x, int y);
void ff_put_h264_chroma_mc2_mmx2 (uint8_t *dst, uint8_t *src, void ff_put_h264_chroma_mc2_mmx2 (uint8_t *dst, uint8_t *src,
int stride, int h, int x, int y); int stride, int h, int x, int y);
@ -2573,9 +2561,6 @@ void dsputil_init_mmx(DSPContext* c, AVCodecContext *avctx)
c->put_h264_chroma_pixels_tab[1]= ff_put_h264_chroma_mc4_mmx; c->put_h264_chroma_pixels_tab[1]= ff_put_h264_chroma_mc4_mmx;
} }
c->put_rv40_chroma_pixels_tab[0]= ff_put_rv40_chroma_mc8_mmx;
c->put_rv40_chroma_pixels_tab[1]= ff_put_rv40_chroma_mc4_mmx;
c->vector_clip_int32 = ff_vector_clip_int32_mmx; c->vector_clip_int32 = ff_vector_clip_int32_mmx;
#endif #endif
@ -2675,9 +2660,6 @@ void dsputil_init_mmx(DSPContext* c, AVCodecContext *avctx)
SET_QPEL_FUNCS(avg_2tap_qpel, 1, 8, mmx2, ); SET_QPEL_FUNCS(avg_2tap_qpel, 1, 8, mmx2, );
#if HAVE_YASM #if HAVE_YASM
c->avg_rv40_chroma_pixels_tab[0]= ff_avg_rv40_chroma_mc8_mmx2;
c->avg_rv40_chroma_pixels_tab[1]= ff_avg_rv40_chroma_mc4_mmx2;
if (!high_bit_depth) { if (!high_bit_depth) {
c->avg_h264_chroma_pixels_tab[0]= ff_avg_h264_chroma_mc8_mmx2_rnd; c->avg_h264_chroma_pixels_tab[0]= ff_avg_h264_chroma_mc8_mmx2_rnd;
c->avg_h264_chroma_pixels_tab[1]= ff_avg_h264_chroma_mc4_mmx2; c->avg_h264_chroma_pixels_tab[1]= ff_avg_h264_chroma_mc4_mmx2;
@ -2760,8 +2742,6 @@ void dsputil_init_mmx(DSPContext* c, AVCodecContext *avctx)
c->avg_h264_chroma_pixels_tab[1]= ff_avg_h264_chroma_mc4_3dnow; c->avg_h264_chroma_pixels_tab[1]= ff_avg_h264_chroma_mc4_3dnow;
} }
c->avg_rv40_chroma_pixels_tab[0]= ff_avg_rv40_chroma_mc8_3dnow;
c->avg_rv40_chroma_pixels_tab[1]= ff_avg_rv40_chroma_mc4_3dnow;
#endif #endif
} }

60
libavcodec/x86/rv40dsp.c
Unescape View File

@ -0,0 +1,60 @@
/*
* RV40 decoder motion compensation functions x86-optimised
* Copyright (c) 2008 Konstantin Shishkov
*
* 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
*/
/**
* @file
* RV40 decoder motion compensation functions x86-optimised
*/
#include "libavcodec/rv34dsp.h"
void ff_put_rv40_chroma_mc8_mmx (uint8_t *dst, uint8_t *src,
int stride, int h, int x, int y);
void ff_avg_rv40_chroma_mc8_mmx2 (uint8_t *dst, uint8_t *src,
int stride, int h, int x, int y);
void ff_avg_rv40_chroma_mc8_3dnow(uint8_t *dst, uint8_t *src,
int stride, int h, int x, int y);
void ff_put_rv40_chroma_mc4_mmx (uint8_t *dst, uint8_t *src,
int stride, int h, int x, int y);
void ff_avg_rv40_chroma_mc4_mmx2 (uint8_t *dst, uint8_t *src,
int stride, int h, int x, int y);
void ff_avg_rv40_chroma_mc4_3dnow(uint8_t *dst, uint8_t *src,
int stride, int h, int x, int y);
void ff_rv40dsp_init_x86(RV34DSPContext *c, DSPContext *dsp)
{
av_unused int mm_flags = av_get_cpu_flags();
#if HAVE_YASM
if (mm_flags & AV_CPU_FLAG_MMX) {
c->put_chroma_pixels_tab[0] = ff_put_rv40_chroma_mc8_mmx;
c->put_chroma_pixels_tab[1] = ff_put_rv40_chroma_mc4_mmx;
}
if (mm_flags & AV_CPU_FLAG_MMX2) {
c->avg_chroma_pixels_tab[0] = ff_avg_rv40_chroma_mc8_mmx2;
c->avg_chroma_pixels_tab[1] = ff_avg_rv40_chroma_mc4_mmx2;
} else if (mm_flags & AV_CPU_FLAG_3DNOW) {
c->avg_chroma_pixels_tab[0] = ff_avg_rv40_chroma_mc8_3dnow;
c->avg_chroma_pixels_tab[1] = ff_avg_rv40_chroma_mc4_3dnow;
}
#endif
}
Write Preview
Loading…
Cancel
Save