falkTX
/
FFmpeg
mirror of https://github.com/falkTX/FFmpeg.git
1
0
Fork 0
Code Issues 0 Releases 338 Wiki Activity
Browse Source

Merge commit 'ac4b32df71bd932838043a4838b86d11e169707f' 

* commit 'ac4b32df71bd932838043a4838b86d11e169707f':
  On2 VP7 decoder

Conflicts:
	Changelog
	libavcodec/arm/h264pred_init_arm.c
	libavcodec/arm/vp8dsp.h
	libavcodec/arm/vp8dsp_init_arm.c
	libavcodec/arm/vp8dsp_init_armv6.c
	libavcodec/arm/vp8dsp_init_neon.c
	libavcodec/avcodec.h
	libavcodec/h264pred.c
	libavcodec/version.h
	libavcodec/vp8.c
	libavcodec/vp8.h
	libavcodec/vp8data.h
	libavcodec/vp8dsp.c
	libavcodec/vp8dsp.h
	libavcodec/x86/h264_intrapred_init.c
	libavcodec/x86/vp8dsp_init.c

See: 89f2f5dbd7 and others
Merged-by: Michael Niedermayer <michaelni@gmx.at>
tags/n2.3
Michael Niedermayer 12 years ago
parent
16509d3a28 ac4b32df71
commit
fb61ed1e9f
18 changed files with 798 additions and 562 deletions
Split View
Diff Options
Show Stats Download Patch File Download Diff File
  1. +6
    -0
      libavcodec/arm/Makefile
  2. +4
    -2
      libavcodec/arm/h264pred_init_arm.c
  3. +4
    -2
      libavcodec/arm/vp8dsp.h
  4. +13
    -3
      libavcodec/arm/vp8dsp_init_arm.c
  5. +25
    -24
      libavcodec/arm/vp8dsp_init_armv6.c
  6. +24
    -23
      libavcodec/arm/vp8dsp_init_neon.c
  7. +1
    -0
      libavcodec/avcodec.h
  8. +4
    -3
      libavcodec/h264pred.c
  9. +1
    -0
      libavcodec/ppc/Makefile
  10. +2
    -1
      libavcodec/ppc/vp8dsp_altivec.c
  11. +1
    -0
      libavcodec/utils.c
  12. +1
    -1
      libavcodec/version.h
  13. +360
    -223
      libavcodec/vp8.c
  14. +54
    -45
      libavcodec/vp8data.h
  15. +191
    -143
      libavcodec/vp8dsp.c
  16. +13
    -4
      libavcodec/vp8dsp.h
  17. +2
    -1
      libavcodec/x86/h264_intrapred_init.c
  18. +92
    -87
      libavcodec/x86/vp8dsp_init.c

+ 6
- 0
libavcodec/arm/Makefile View File

@@ -32,6 +32,7 @@ OBJS-$(CONFIG_MLP_DECODER) += arm/mlpdsp_init_arm.o
OBJS-$(CONFIG_VC1_DECODER) += arm/vc1dsp_init_arm.o
OBJS-$(CONFIG_VORBIS_DECODER) += arm/vorbisdsp_init_arm.o
OBJS-$(CONFIG_VP6_DECODER) += arm/vp6dsp_init_arm.o
OBJS-$(CONFIG_VP7_DECODER) += arm/vp8dsp_init_arm.o
OBJS-$(CONFIG_VP8_DECODER) += arm/vp8dsp_init_arm.o
OBJS-$(CONFIG_RV30_DECODER) += arm/rv34dsp_init_arm.o
OBJS-$(CONFIG_RV40_DECODER) += arm/rv34dsp_init_arm.o \
@@ -56,6 +57,9 @@ ARMV6-OBJS-$(CONFIG_HPELDSP) += arm/hpeldsp_init_armv6.o \
ARMV6-OBJS-$(CONFIG_MPEGAUDIODSP) += arm/mpegaudiodsp_fixed_armv6.o

ARMV6-OBJS-$(CONFIG_MLP_DECODER) += arm/mlpdsp_armv6.o
ARMV6-OBJS-$(CONFIG_VP7_DECODER) += arm/vp8_armv6.o \
arm/vp8dsp_init_armv6.o \
arm/vp8dsp_armv6.o
ARMV6-OBJS-$(CONFIG_VP8_DECODER) += arm/vp8_armv6.o \
arm/vp8dsp_init_armv6.o \
arm/vp8dsp_armv6.o
@@ -103,5 +107,7 @@ NEON-OBJS-$(CONFIG_VC1_DECODER) += arm/vc1dsp_init_neon.o \
arm/vc1dsp_neon.o
NEON-OBJS-$(CONFIG_VORBIS_DECODER) += arm/vorbisdsp_neon.o
NEON-OBJS-$(CONFIG_VP6_DECODER) += arm/vp6dsp_neon.o
NEON-OBJS-$(CONFIG_VP7_DECODER) += arm/vp8dsp_init_neon.o \
arm/vp8dsp_neon.o
NEON-OBJS-$(CONFIG_VP8_DECODER) += arm/vp8dsp_init_neon.o \
arm/vp8dsp_neon.o

+ 4
- 2
libavcodec/arm/h264pred_init_arm.c View File

@@ -60,7 +60,8 @@ static av_cold void h264_pred_init_neon(H264PredContext *h, int codec_id,
if (codec_id != AV_CODEC_ID_VP7 && codec_id != AV_CODEC_ID_VP8)
h->pred8x8[PLANE_PRED8x8] = ff_pred8x8_plane_neon;
h->pred8x8[DC_128_PRED8x8 ] = ff_pred8x8_128_dc_neon;
if (codec_id != AV_CODEC_ID_RV40 && codec_id != AV_CODEC_ID_VP7 && codec_id != AV_CODEC_ID_VP8) {
if (codec_id != AV_CODEC_ID_RV40 && codec_id != AV_CODEC_ID_VP7 &&
codec_id != AV_CODEC_ID_VP8) {
h->pred8x8[DC_PRED8x8 ] = ff_pred8x8_dc_neon;
h->pred8x8[LEFT_DC_PRED8x8] = ff_pred8x8_left_dc_neon;
h->pred8x8[TOP_DC_PRED8x8 ] = ff_pred8x8_top_dc_neon;
@@ -77,7 +78,8 @@ static av_cold void h264_pred_init_neon(H264PredContext *h, int codec_id,
h->pred16x16[LEFT_DC_PRED8x8] = ff_pred16x16_left_dc_neon;
h->pred16x16[TOP_DC_PRED8x8 ] = ff_pred16x16_top_dc_neon;
h->pred16x16[DC_128_PRED8x8 ] = ff_pred16x16_128_dc_neon;
if (codec_id != AV_CODEC_ID_SVQ3 && codec_id != AV_CODEC_ID_RV40 && codec_id != AV_CODEC_ID_VP7 && codec_id != AV_CODEC_ID_VP8)
if (codec_id != AV_CODEC_ID_SVQ3 && codec_id != AV_CODEC_ID_RV40 &&
codec_id != AV_CODEC_ID_VP7 && codec_id != AV_CODEC_ID_VP8)
h->pred16x16[PLANE_PRED8x8 ] = ff_pred16x16_plane_neon;
#endif // HAVE_NEON
}


+ 4
- 2
libavcodec/arm/vp8dsp.h View File

@@ -21,8 +21,10 @@

#include "libavcodec/vp8dsp.h"

void ff_vp8dsp_init_armv6(VP8DSPContext *dsp, int vp7);
void ff_vp8dsp_init_neon(VP8DSPContext *dsp, int vp7);
void ff_vp78dsp_init_armv6(VP8DSPContext *dsp);
void ff_vp78dsp_init_neon(VP8DSPContext *dsp);
void ff_vp8dsp_init_armv6(VP8DSPContext *dsp);
void ff_vp8dsp_init_neon(VP8DSPContext *dsp);

#define VP8_LF_Y(hv, inner, opt) \
void ff_vp8_##hv##_loop_filter16##inner##_##opt(uint8_t *dst, \


+ 13
- 3
libavcodec/arm/vp8dsp_init_arm.c View File

@@ -23,12 +23,22 @@
#include "libavcodec/vp8dsp.h"
#include "vp8dsp.h"

av_cold void ff_vp8dsp_init_arm(VP8DSPContext *dsp, int vp7)
av_cold void ff_vp78dsp_init_arm(VP8DSPContext *dsp)
{
int cpu_flags = av_get_cpu_flags();

if (have_armv6(cpu_flags))
ff_vp8dsp_init_armv6(dsp, vp7);
ff_vp78dsp_init_armv6(dsp);
if (have_neon(cpu_flags))
ff_vp8dsp_init_neon(dsp, vp7);
ff_vp78dsp_init_neon(dsp);
}

av_cold void ff_vp8dsp_init_arm(VP8DSPContext *dsp)
{
int cpu_flags = av_get_cpu_flags();

if (have_armv6(cpu_flags))
ff_vp8dsp_init_armv6(dsp);
if (have_neon(cpu_flags))
ff_vp8dsp_init_neon(dsp);
}

+ 25
- 24
libavcodec/arm/vp8dsp_init_armv6.c View File

@@ -40,31 +40,8 @@ VP8_BILIN(16, armv6);
VP8_BILIN(8, armv6);
VP8_BILIN(4, armv6);

av_cold void ff_vp8dsp_init_armv6(VP8DSPContext *dsp, int vp7)
av_cold void ff_vp78dsp_init_armv6(VP8DSPContext *dsp)
{
if (!vp7) {
dsp->vp8_luma_dc_wht = ff_vp8_luma_dc_wht_armv6;
dsp->vp8_luma_dc_wht_dc = ff_vp8_luma_dc_wht_dc_armv6;

dsp->vp8_idct_add = ff_vp8_idct_add_armv6;
dsp->vp8_idct_dc_add = ff_vp8_idct_dc_add_armv6;
dsp->vp8_idct_dc_add4y = ff_vp8_idct_dc_add4y_armv6;
dsp->vp8_idct_dc_add4uv = ff_vp8_idct_dc_add4uv_armv6;

dsp->vp8_v_loop_filter16y = ff_vp8_v_loop_filter16_armv6;
dsp->vp8_h_loop_filter16y = ff_vp8_h_loop_filter16_armv6;
dsp->vp8_v_loop_filter8uv = ff_vp8_v_loop_filter8uv_armv6;
dsp->vp8_h_loop_filter8uv = ff_vp8_h_loop_filter8uv_armv6;

dsp->vp8_v_loop_filter16y_inner = ff_vp8_v_loop_filter16_inner_armv6;
dsp->vp8_h_loop_filter16y_inner = ff_vp8_h_loop_filter16_inner_armv6;
dsp->vp8_v_loop_filter8uv_inner = ff_vp8_v_loop_filter8uv_inner_armv6;
dsp->vp8_h_loop_filter8uv_inner = ff_vp8_h_loop_filter8uv_inner_armv6;

dsp->vp8_v_loop_filter_simple = ff_vp8_v_loop_filter16_simple_armv6;
dsp->vp8_h_loop_filter_simple = ff_vp8_h_loop_filter16_simple_armv6;
}

dsp->put_vp8_epel_pixels_tab[0][0][0] = ff_put_vp8_pixels16_armv6;
dsp->put_vp8_epel_pixels_tab[0][0][2] = ff_put_vp8_epel16_h6_armv6;
dsp->put_vp8_epel_pixels_tab[0][2][0] = ff_put_vp8_epel16_v6_armv6;
@@ -120,3 +97,27 @@ av_cold void ff_vp8dsp_init_armv6(VP8DSPContext *dsp, int vp7)
dsp->put_vp8_bilinear_pixels_tab[2][2][1] = ff_put_vp8_bilin4_hv_armv6;
dsp->put_vp8_bilinear_pixels_tab[2][2][2] = ff_put_vp8_bilin4_hv_armv6;
}

av_cold void ff_vp8dsp_init_armv6(VP8DSPContext *dsp)
{
dsp->vp8_luma_dc_wht = ff_vp8_luma_dc_wht_armv6;
dsp->vp8_luma_dc_wht_dc = ff_vp8_luma_dc_wht_dc_armv6;

dsp->vp8_idct_add = ff_vp8_idct_add_armv6;
dsp->vp8_idct_dc_add = ff_vp8_idct_dc_add_armv6;
dsp->vp8_idct_dc_add4y = ff_vp8_idct_dc_add4y_armv6;
dsp->vp8_idct_dc_add4uv = ff_vp8_idct_dc_add4uv_armv6;

dsp->vp8_v_loop_filter16y = ff_vp8_v_loop_filter16_armv6;
dsp->vp8_h_loop_filter16y = ff_vp8_h_loop_filter16_armv6;
dsp->vp8_v_loop_filter8uv = ff_vp8_v_loop_filter8uv_armv6;
dsp->vp8_h_loop_filter8uv = ff_vp8_h_loop_filter8uv_armv6;

dsp->vp8_v_loop_filter16y_inner = ff_vp8_v_loop_filter16_inner_armv6;
dsp->vp8_h_loop_filter16y_inner = ff_vp8_h_loop_filter16_inner_armv6;
dsp->vp8_v_loop_filter8uv_inner = ff_vp8_v_loop_filter8uv_inner_armv6;
dsp->vp8_h_loop_filter8uv_inner = ff_vp8_h_loop_filter8uv_inner_armv6;

dsp->vp8_v_loop_filter_simple = ff_vp8_v_loop_filter16_simple_armv6;
dsp->vp8_h_loop_filter_simple = ff_vp8_h_loop_filter16_simple_armv6;
}

+ 24
- 23
libavcodec/arm/vp8dsp_init_neon.c View File

@@ -39,30 +39,8 @@ VP8_BILIN(16, neon);
VP8_BILIN(8, neon);
VP8_BILIN(4, neon);

av_cold void ff_vp8dsp_init_neon(VP8DSPContext *dsp, int vp7)
av_cold void ff_vp78dsp_init_neon(VP8DSPContext *dsp)
{
if (!vp7) {
dsp->vp8_luma_dc_wht = ff_vp8_luma_dc_wht_neon;

dsp->vp8_idct_add = ff_vp8_idct_add_neon;
dsp->vp8_idct_dc_add = ff_vp8_idct_dc_add_neon;
dsp->vp8_idct_dc_add4y = ff_vp8_idct_dc_add4y_neon;
dsp->vp8_idct_dc_add4uv = ff_vp8_idct_dc_add4uv_neon;

dsp->vp8_v_loop_filter16y = ff_vp8_v_loop_filter16_neon;
dsp->vp8_h_loop_filter16y = ff_vp8_h_loop_filter16_neon;
dsp->vp8_v_loop_filter8uv = ff_vp8_v_loop_filter8uv_neon;
dsp->vp8_h_loop_filter8uv = ff_vp8_h_loop_filter8uv_neon;

dsp->vp8_v_loop_filter16y_inner = ff_vp8_v_loop_filter16_inner_neon;
dsp->vp8_h_loop_filter16y_inner = ff_vp8_h_loop_filter16_inner_neon;
dsp->vp8_v_loop_filter8uv_inner = ff_vp8_v_loop_filter8uv_inner_neon;
dsp->vp8_h_loop_filter8uv_inner = ff_vp8_h_loop_filter8uv_inner_neon;

dsp->vp8_v_loop_filter_simple = ff_vp8_v_loop_filter16_simple_neon;
dsp->vp8_h_loop_filter_simple = ff_vp8_h_loop_filter16_simple_neon;
}

dsp->put_vp8_epel_pixels_tab[0][0][0] = ff_put_vp8_pixels16_neon;
dsp->put_vp8_epel_pixels_tab[0][0][2] = ff_put_vp8_epel16_h6_neon;
dsp->put_vp8_epel_pixels_tab[0][2][0] = ff_put_vp8_epel16_v6_neon;
@@ -116,3 +94,26 @@ av_cold void ff_vp8dsp_init_neon(VP8DSPContext *dsp, int vp7)
dsp->put_vp8_bilinear_pixels_tab[2][2][1] = ff_put_vp8_bilin4_hv_neon;
dsp->put_vp8_bilinear_pixels_tab[2][2][2] = ff_put_vp8_bilin4_hv_neon;
}

av_cold void ff_vp8dsp_init_neon(VP8DSPContext *dsp)
{
dsp->vp8_luma_dc_wht = ff_vp8_luma_dc_wht_neon;

dsp->vp8_idct_add = ff_vp8_idct_add_neon;
dsp->vp8_idct_dc_add = ff_vp8_idct_dc_add_neon;
dsp->vp8_idct_dc_add4y = ff_vp8_idct_dc_add4y_neon;
dsp->vp8_idct_dc_add4uv = ff_vp8_idct_dc_add4uv_neon;

dsp->vp8_v_loop_filter16y = ff_vp8_v_loop_filter16_neon;
dsp->vp8_h_loop_filter16y = ff_vp8_h_loop_filter16_neon;
dsp->vp8_v_loop_filter8uv = ff_vp8_v_loop_filter8uv_neon;
dsp->vp8_h_loop_filter8uv = ff_vp8_h_loop_filter8uv_neon;

dsp->vp8_v_loop_filter16y_inner = ff_vp8_v_loop_filter16_inner_neon;
dsp->vp8_h_loop_filter16y_inner = ff_vp8_h_loop_filter16_inner_neon;
dsp->vp8_v_loop_filter8uv_inner = ff_vp8_v_loop_filter8uv_inner_neon;
dsp->vp8_h_loop_filter8uv_inner = ff_vp8_h_loop_filter8uv_inner_neon;

dsp->vp8_v_loop_filter_simple = ff_vp8_v_loop_filter16_simple_neon;
dsp->vp8_h_loop_filter_simple = ff_vp8_h_loop_filter16_simple_neon;
}

+ 1
- 0
libavcodec/avcodec.h View File

@@ -290,6 +290,7 @@ enum AVCodecID {
AV_CODEC_ID_BRENDER_PIX_DEPRECATED,
AV_CODEC_ID_PAF_VIDEO_DEPRECATED,
AV_CODEC_ID_EXR_DEPRECATED,
AV_CODEC_ID_VP7_DEPRECATED,

AV_CODEC_ID_BRENDER_PIX= MKBETAG('B','P','I','X'),
AV_CODEC_ID_Y41P = MKBETAG('Y','4','1','P'),


+ 4
- 3
libavcodec/h264pred.c View File

@@ -420,7 +420,7 @@ av_cold void ff_h264_pred_init(H264PredContext *h, int codec_id,

#define H264_PRED(depth) \
if(codec_id != AV_CODEC_ID_RV40){\
if(codec_id == AV_CODEC_ID_VP7 || codec_id == AV_CODEC_ID_VP8) {\
if (codec_id == AV_CODEC_ID_VP7 || codec_id == AV_CODEC_ID_VP8) {\
h->pred4x4[VERT_PRED ]= FUNCD(pred4x4_vertical_vp8);\
h->pred4x4[HOR_PRED ]= FUNCD(pred4x4_horizontal_vp8);\
} else {\
@@ -440,7 +440,7 @@ av_cold void ff_h264_pred_init(H264PredContext *h, int codec_id,
} else\
h->pred4x4[VERT_LEFT_PRED ]= FUNCC(pred4x4_vertical_left , depth);\
h->pred4x4[HOR_UP_PRED ]= FUNCC(pred4x4_horizontal_up , depth);\
if(codec_id != AV_CODEC_ID_VP7 && codec_id != AV_CODEC_ID_VP8) {\
if (codec_id != AV_CODEC_ID_VP7 && codec_id != AV_CODEC_ID_VP8) {\
h->pred4x4[LEFT_DC_PRED ]= FUNCC(pred4x4_left_dc , depth);\
h->pred4x4[TOP_DC_PRED ]= FUNCC(pred4x4_top_dc , depth);\
} else {\
@@ -498,7 +498,8 @@ av_cold void ff_h264_pred_init(H264PredContext *h, int codec_id,
}\
} else\
h->pred8x8[PLANE_PRED8x8]= FUNCD(pred8x8_tm_vp8);\
if(codec_id != AV_CODEC_ID_RV40 && codec_id != AV_CODEC_ID_VP7 && codec_id != AV_CODEC_ID_VP8){\
if (codec_id != AV_CODEC_ID_RV40 && codec_id != AV_CODEC_ID_VP7 && \
codec_id != AV_CODEC_ID_VP8) {\
if (chroma_format_idc <= 1) {\
h->pred8x8[DC_PRED8x8 ]= FUNCC(pred8x8_dc , depth);\
h->pred8x8[LEFT_DC_PRED8x8]= FUNCC(pred8x8_left_dc , depth);\


+ 1
- 0
libavcodec/ppc/Makefile View File

@@ -13,6 +13,7 @@ OBJS-$(CONFIG_VP3DSP) += ppc/vp3dsp_altivec.o

OBJS-$(CONFIG_VC1_DECODER) += ppc/vc1dsp_altivec.o
OBJS-$(CONFIG_VORBIS_DECODER) += ppc/vorbisdsp_altivec.o
OBJS-$(CONFIG_VP7_DECODER) += ppc/vp8dsp_altivec.o
OBJS-$(CONFIG_VP8_DECODER) += ppc/vp8dsp_altivec.o

ALTIVEC-OBJS-$(CONFIG_DSPUTIL) += ppc/dsputil_altivec.o \


+ 2
- 1
libavcodec/ppc/vp8dsp_altivec.c View File

@@ -311,7 +311,8 @@ static void put_vp8_pixels16_altivec(uint8_t *dst, ptrdiff_t dstride, uint8_t *s

#endif /* HAVE_ALTIVEC */

av_cold void ff_vp8dsp_init_ppc(VP8DSPContext *c)

av_cold void ff_vp78dsp_init_ppc(VP8DSPContext *c)
{
#if HAVE_ALTIVEC
if (!(av_get_cpu_flags() & AV_CPU_FLAG_ALTIVEC))


+ 1
- 0
libavcodec/utils.c View File

@@ -2687,6 +2687,7 @@ static enum AVCodecID remap_deprecated_codec_id(enum AVCodecID id)
case AV_CODEC_ID_PAF_VIDEO_DEPRECATED : return AV_CODEC_ID_PAF_VIDEO;
case AV_CODEC_ID_WEBP_DEPRECATED: return AV_CODEC_ID_WEBP;
case AV_CODEC_ID_HEVC_DEPRECATED: return AV_CODEC_ID_HEVC;
case AV_CODEC_ID_VP7_DEPRECATED : return AV_CODEC_ID_VP7;
default : return id;
}
}


+ 1
- 1
libavcodec/version.h View File

@@ -30,7 +30,7 @@

#define LIBAVCODEC_VERSION_MAJOR 55
#define LIBAVCODEC_VERSION_MINOR 56
#define LIBAVCODEC_VERSION_MICRO 107
#define LIBAVCODEC_VERSION_MICRO 108

#define LIBAVCODEC_VERSION_INT AV_VERSION_INT(LIBAVCODEC_VERSION_MAJOR, \
LIBAVCODEC_VERSION_MINOR, \


+ 360
- 223
libavcodec/vp8.c
File diff suppressed because it is too large
View File


+ 54
- 45
libavcodec/vp8data.h View File

@@ -30,8 +30,7 @@
#include "vp8.h"
#include "h264pred.h"

static const uint8_t vp7_pred4x4_mode[] =
{
static const uint8_t vp7_pred4x4_mode[] = {
[DC_PRED8x8] = DC_PRED,
[VERT_PRED8x8] = TM_VP8_PRED,
[HOR_PRED8x8] = TM_VP8_PRED,
@@ -59,9 +58,9 @@ static const int8_t vp8_pred16x16_tree_inter[4][2] = {
{ -PLANE_PRED8x8, -MODE_I4x4 }, // '110', '111'
};

typedef struct {
int8_t yoffset;
int8_t xoffset;
typedef struct VP7MVPred {
int8_t yoffset;
int8_t xoffset;
uint8_t subblock;
uint8_t score;
} VP7MVPred;
@@ -147,7 +146,9 @@ static const uint8_t vp8_mbsplit_prob[3] = {
110, 111, 150
};

static const uint8_t vp7_submv_prob[3] = { 180, 162, 25 };
static const uint8_t vp7_submv_prob[3] = {
180, 162, 25
};

static const uint8_t vp8_submv_prob[5][3] = {
{ 147, 136, 18 },
@@ -161,7 +162,7 @@ static const uint8_t vp8_pred16x16_prob_intra[4] = {
145, 156, 163, 128
};
static const uint8_t vp8_pred16x16_prob_inter[4] = {
112, 86, 140, 37
112, 86, 140, 37
};

static const int8_t vp8_pred4x4_tree[9][2] = {
@@ -708,7 +709,7 @@ static const uint8_t vp8_token_update_probs[4][8][3][NUM_DCT_TOKENS - 1] = {
};

// fixme: copied from h264data.h
static const uint8_t zigzag_scan[16] = {
static const uint8_t zigzag_scan[16]={
0 + 0 * 4, 1 + 0 * 4, 0 + 1 * 4, 0 + 2 * 4,
1 + 1 * 4, 2 + 0 * 4, 3 + 0 * 4, 2 + 1 * 4,
1 + 2 * 4, 0 + 3 * 4, 1 + 3 * 4, 2 + 2 * 4,
@@ -771,59 +772,67 @@ static const uint8_t vp8_mv_default_prob[2][19] = {
};

static const uint8_t vp7_feature_value_size[2][4] = {
{7, 6, 0, 8},
{7, 6, 0, 5},
{ 7, 6, 0, 8 },
{ 7, 6, 0, 5 },
};

static const int8_t vp7_feature_index_tree[4][2] =
{
{ 1, 2 },
{ -0, -1 }, // '00', '01'
{ -2, -3 }, // '10', '11'
{ 1, 2 },
{ -0, -1 }, // '00', '01'
{ -2, -3 }, // '10', '11'
};

static const uint16_t vp7_ydc_qlookup[] = {
4, 4, 5, 6, 6, 7, 8, 8, 9, 10, 11, 12, 13, 14, 15, 16,
17, 18, 19, 20, 21, 22, 23, 23, 24, 25, 26, 27, 28, 29, 30, 31,
32, 33, 33, 34, 35, 36, 36, 37, 38, 39, 39, 40, 41, 41, 42, 43,
43, 44, 45, 45, 46, 47, 48, 48, 49, 50, 51, 52, 53, 53, 54, 56,
57, 58, 59, 60, 62, 63, 65, 66, 68, 70, 72, 74, 76, 79, 81, 84,
87, 90, 93, 96, 100, 104, 108, 112, 116, 121, 126, 131, 136, 142, 148, 154,
160, 167, 174, 182, 189, 198, 206, 215, 224, 234, 244, 254, 265, 277, 288, 301,
313, 327, 340, 355, 370, 385, 401, 417, 434, 452, 470, 489, 509, 529, 550, 572,
4, 4, 5, 6, 6, 7, 8, 8, 9, 10, 11, 12, 13, 14, 15,
16, 17, 18, 19, 20, 21, 22, 23, 23, 24, 25, 26, 27, 28, 29,
30, 31, 32, 33, 33, 34, 35, 36, 36, 37, 38, 39, 39, 40, 41,
41, 42, 43, 43, 44, 45, 45, 46, 47, 48, 48, 49, 50, 51, 52,
53, 53, 54, 56, 57, 58, 59, 60, 62, 63, 65, 66, 68, 70, 72,
74, 76, 79, 81, 84, 87, 90, 93, 96, 100, 104, 108, 112, 116, 121,
126, 131, 136, 142, 148, 154, 160, 167, 174, 182, 189, 198, 206, 215, 224,
234, 244, 254, 265, 277, 288, 301, 313, 327, 340, 355, 370, 385, 401, 417,
434, 452, 470, 489, 509, 529, 550, 572,
};

static const uint16_t vp7_yac_qlookup[] = {
4, 4, 5, 5, 6, 6, 7, 8, 9, 10, 11, 12, 13, 15, 16, 17,
19, 20, 22, 23, 25, 26, 28, 29, 31, 32, 34, 35, 37, 38, 40, 41,
42, 44, 45, 46, 48, 49, 50, 51, 53, 54, 55, 56, 57, 58, 59, 61,
62, 63, 64, 65, 67, 68, 69, 70, 72, 73, 75, 76, 78, 80, 82, 84,
86, 88, 91, 93, 96, 99, 102, 105, 109, 112, 116, 121, 125, 130, 135, 140,
146, 152, 158, 165, 172, 180, 188, 196, 205, 214, 224, 234, 245, 256, 268, 281,
294, 308, 322, 337, 353, 369, 386, 404, 423, 443, 463, 484, 506, 529, 553, 578,
604, 631, 659, 688, 718, 749, 781, 814, 849, 885, 922, 960, 1000, 1041, 1083, 1127,
4, 4, 5, 5, 6, 6, 7, 8, 9, 10, 11, 12, 13, 15,
16, 17, 19, 20, 22, 23, 25, 26, 28, 29, 31, 32, 34, 35,
37, 38, 40, 41, 42, 44, 45, 46, 48, 49, 50, 51, 53, 54,
55, 56, 57, 58, 59, 61, 62, 63, 64, 65, 67, 68, 69, 70,
72, 73, 75, 76, 78, 80, 82, 84, 86, 88, 91, 93, 96, 99,
102, 105, 109, 112, 116, 121, 125, 130, 135, 140, 146, 152, 158, 165,
172, 180, 188, 196, 205, 214, 224, 234, 245, 256, 268, 281, 294, 308,
322, 337, 353, 369, 386, 404, 423, 443, 463, 484, 506, 529, 553, 578,
604, 631, 659, 688, 718, 749, 781, 814, 849, 885, 922, 960, 1000, 1041,
1083, 1127,
};

static const uint16_t vp7_y2dc_qlookup[] = {
7, 9, 11, 13, 15, 17, 19, 21, 23, 26, 28, 30, 33, 35, 37, 39,
42, 44, 46, 48, 51, 53, 55, 57, 59, 61, 63, 65, 67, 69, 70, 72,
74, 75, 77, 78, 80, 81, 83, 84, 85, 87, 88, 89, 90, 92, 93, 94,
95, 96, 97, 99, 100, 101, 102, 104, 105, 106, 108, 109, 111, 113, 114, 116,
118, 120, 123, 125, 128, 131, 134, 137, 140, 144, 148, 152, 156, 161, 166, 171,
176, 182, 188, 195, 202, 209, 217, 225, 234, 243, 253, 263, 274, 285, 297, 309,
322, 336, 350, 365, 381, 397, 414, 432, 450, 470, 490, 511, 533, 556, 579, 604,
630, 656, 684, 713, 742, 773, 805, 838, 873, 908, 945, 983, 1022, 1063, 1105, 1148,
7, 9, 11, 13, 15, 17, 19, 21, 23, 26, 28, 30, 33, 35,
37, 39, 42, 44, 46, 48, 51, 53, 55, 57, 59, 61, 63, 65,
67, 69, 70, 72, 74, 75, 77, 78, 80, 81, 83, 84, 85, 87,
88, 89, 90, 92, 93, 94, 95, 96, 97, 99, 100, 101, 102, 104,
105, 106, 108, 109, 111, 113, 114, 116, 118, 120, 123, 125, 128, 131,
134, 137, 140, 144, 148, 152, 156, 161, 166, 171, 176, 182, 188, 195,
202, 209, 217, 225, 234, 243, 253, 263, 274, 285, 297, 309, 322, 336,
350, 365, 381, 397, 414, 432, 450, 470, 490, 511, 533, 556, 579, 604,
630, 656, 684, 713, 742, 773, 805, 838, 873, 908, 945, 983, 1022, 1063,
1105, 1148,
};

static const uint16_t vp7_y2ac_qlookup[] = {
7, 9, 11, 13, 16, 18, 21, 24, 26, 29, 32, 35, 38, 41, 43, 46,
49, 52, 55, 58, 61, 64, 66, 69, 72, 74, 77, 79, 82, 84, 86, 88,
91, 93, 95, 97, 98, 100, 102, 104, 105, 107, 109, 110, 112, 113, 115, 116,
117, 119, 120, 122, 123, 125, 127, 128, 130, 132, 134, 136, 138, 141, 143, 146,
149, 152, 155, 158, 162, 166, 171, 175, 180, 185, 191, 197, 204, 210, 218, 226,
234, 243, 252, 262, 273, 284, 295, 308, 321, 335, 350, 365, 381, 398, 416, 435,
455, 476, 497, 520, 544, 569, 595, 622, 650, 680, 711, 743, 776, 811, 848, 885,
925, 965, 1008, 1052, 1097, 1144, 1193, 1244, 1297, 1351, 1407, 1466, 1526, 1588, 1652, 1719,
7, 9, 11, 13, 16, 18, 21, 24, 26, 29, 32, 35,
38, 41, 43, 46, 49, 52, 55, 58, 61, 64, 66, 69,
72, 74, 77, 79, 82, 84, 86, 88, 91, 93, 95, 97,
98, 100, 102, 104, 105, 107, 109, 110, 112, 113, 115, 116,
117, 119, 120, 122, 123, 125, 127, 128, 130, 132, 134, 136,
138, 141, 143, 146, 149, 152, 155, 158, 162, 166, 171, 175,
180, 185, 191, 197, 204, 210, 218, 226, 234, 243, 252, 262,
273, 284, 295, 308, 321, 335, 350, 365, 381, 398, 416, 435,
455, 476, 497, 520, 544, 569, 595, 622, 650, 680, 711, 743,
776, 811, 848, 885, 925, 965, 1008, 1052, 1097, 1144, 1193, 1244,
1297, 1351, 1407, 1466, 1526, 1588, 1652, 1719,
};

#endif /* AVCODEC_VP8DATA_H */

+ 191
- 143
libavcodec/vp8dsp.c View File

@@ -31,21 +31,23 @@
#include "mathops.h"
#include "vp8dsp.h"

#define MK_IDCT_DC_ADD4_C(name) \
static void name ## _idct_dc_add4uv_c(uint8_t *dst, int16_t block[4][16], ptrdiff_t stride)\
{\
name ## _idct_dc_add_c(dst + stride * 0 + 0, block[0], stride);\
name ## _idct_dc_add_c(dst + stride * 0 + 4, block[1], stride);\
name ## _idct_dc_add_c(dst + stride * 4 + 0, block[2], stride);\
name ## _idct_dc_add_c(dst + stride * 4 + 4, block[3], stride);\
}\
\
static void name ## _idct_dc_add4y_c(uint8_t *dst, int16_t block[4][16], ptrdiff_t stride)\
{\
name ## _idct_dc_add_c(dst + 0, block[0], stride);\
name ## _idct_dc_add_c(dst + 4, block[1], stride);\
name ## _idct_dc_add_c(dst + 8, block[2], stride);\
name ## _idct_dc_add_c(dst + 12, block[3], stride);\
#define MK_IDCT_DC_ADD4_C(name) \
static void name ## _idct_dc_add4uv_c(uint8_t *dst, int16_t block[4][16], \
ptrdiff_t stride) \
{ \
name ## _idct_dc_add_c(dst + stride * 0 + 0, block[0], stride); \
name ## _idct_dc_add_c(dst + stride * 0 + 4, block[1], stride); \
name ## _idct_dc_add_c(dst + stride * 4 + 0, block[2], stride); \
name ## _idct_dc_add_c(dst + stride * 4 + 4, block[3], stride); \
} \
\
static void name ## _idct_dc_add4y_c(uint8_t *dst, int16_t block[4][16], \
ptrdiff_t stride) \
{ \
name ## _idct_dc_add_c(dst + 0, block[0], stride); \
name ## _idct_dc_add_c(dst + 4, block[1], stride); \
name ## _idct_dc_add_c(dst + 8, block[2], stride); \
name ## _idct_dc_add_c(dst + 12, block[3], stride); \
}

#if CONFIG_VP7_DECODER
@@ -113,10 +115,14 @@ static void vp7_idct_add_c(uint8_t *dst, int16_t block[16], ptrdiff_t stride)
b1 = (tmp[i + 0] - tmp[i + 8]) * 23170;
c1 = tmp[i + 4] * 12540 - tmp[i + 12] * 30274;
d1 = tmp[i + 4] * 30274 + tmp[i + 12] * 12540;
dst[0 * stride + i] = av_clip_uint8(dst[0 * stride + i] + ((a1 + d1 + 0x20000) >> 18));
dst[3 * stride + i] = av_clip_uint8(dst[3 * stride + i] + ((a1 - d1 + 0x20000) >> 18));
dst[1 * stride + i] = av_clip_uint8(dst[1 * stride + i] + ((b1 + c1 + 0x20000) >> 18));
dst[2 * stride + i] = av_clip_uint8(dst[2 * stride + i] + ((b1 - c1 + 0x20000) >> 18));
dst[0 * stride + i] = av_clip_uint8(dst[0 * stride + i] +
((a1 + d1 + 0x20000) >> 18));
dst[3 * stride + i] = av_clip_uint8(dst[3 * stride + i] +
((a1 - d1 + 0x20000) >> 18));
dst[1 * stride + i] = av_clip_uint8(dst[1 * stride + i] +
((b1 + c1 + 0x20000) >> 18));
dst[2 * stride + i] = av_clip_uint8(dst[2 * stride + i] +
((b1 - c1 + 0x20000) >> 18));
}
}

@@ -130,12 +136,12 @@ static void vp7_idct_dc_add_c(uint8_t *dst, int16_t block[16], ptrdiff_t stride)
dst[1] = av_clip_uint8(dst[1] + dc);
dst[2] = av_clip_uint8(dst[2] + dc);
dst[3] = av_clip_uint8(dst[3] + dc);
dst += stride;
dst += stride;
}
}

MK_IDCT_DC_ADD4_C(vp7)
#endif
#endif /* CONFIG_VP7_DECODER */

// TODO: Maybe add dequant
#if CONFIG_VP8_DECODER
@@ -235,7 +241,7 @@ static void vp8_idct_dc_add_c(uint8_t *dst, int16_t block[16], ptrdiff_t stride)
}

MK_IDCT_DC_ADD4_C(vp8)
#endif
#endif /* CONFIG_VP8_DECODER */

// because I like only having two parameters to pass functions...
#define LOAD_PIXELS \
@@ -251,7 +257,7 @@ MK_IDCT_DC_ADD4_C(vp8)
#define clip_int8(n) (cm[n + 0x80] - 0x80)

static av_always_inline void filter_common(uint8_t *p, ptrdiff_t stride,
int is4tap, int vpn)
int is4tap, int is_vp7)
{
LOAD_PIXELS
int a, f1, f2;
@@ -268,7 +274,7 @@ static av_always_inline void filter_common(uint8_t *p, ptrdiff_t stride,
// since that's what libvpx does.
f1 = FFMIN(a + 4, 127) >> 3;

if (vpn == 7)
if (is_vp7)
f2 = f1 - ((a & 7) == 4);
else
f2 = FFMIN(a + 3, 127) >> 3;
@@ -280,19 +286,33 @@ static av_always_inline void filter_common(uint8_t *p, ptrdiff_t stride,

// only used for _inner on blocks without high edge variance
if (!is4tap) {
a = (f1 + 1) >> 1;
a = (f1 + 1) >> 1;
p[-2 * stride] = cm[p1 + a];
p[ 1 * stride] = cm[q1 - a];
}
}

static av_always_inline int vp7_simple_limit(uint8_t *p, ptrdiff_t stride, int flim)
static av_always_inline void vp7_filter_common(uint8_t *p, ptrdiff_t stride,
int is4tap)
{
filter_common(p, stride, is4tap, IS_VP7);
}

static av_always_inline void vp8_filter_common(uint8_t *p, ptrdiff_t stride,
int is4tap)
{
filter_common(p, stride, is4tap, IS_VP8);
}

static av_always_inline int vp7_simple_limit(uint8_t *p, ptrdiff_t stride,
int flim)
{
LOAD_PIXELS
return FFABS(p0-q0) <= flim;
return FFABS(p0 - q0) <= flim;
}

static av_always_inline int vp8_simple_limit(uint8_t *p, ptrdiff_t stride, int flim)
static av_always_inline int vp8_simple_limit(uint8_t *p, ptrdiff_t stride,
int flim)
{
LOAD_PIXELS
return 2 * FFABS(p0 - q0) + (FFABS(p1 - q1) >> 1) <= flim;
@@ -302,18 +322,16 @@ static av_always_inline int vp8_simple_limit(uint8_t *p, ptrdiff_t stride, int f
* E - limit at the macroblock edge
* I - limit for interior difference
*/
#define NORMAL_LIMIT(vpn) \
static av_always_inline int vp ## vpn ## _normal_limit(uint8_t *p, ptrdiff_t stride, \
int E, int I)\
{ \
LOAD_PIXELS \
return vp ## vpn ## _simple_limit(p, stride, E) && \
FFABS(p3 - p2) <= I && \
FFABS(p2 - p1) <= I && \
FFABS(p1 - p0) <= I && \
FFABS(q3 - q2) <= I && \
FFABS(q2 - q1) <= I && \
FFABS(q1 - q0) <= I; \
#define NORMAL_LIMIT(vpn) \
static av_always_inline int vp ## vpn ## _normal_limit(uint8_t *p, \
ptrdiff_t stride, \
int E, int I) \
{ \
LOAD_PIXELS \
return vp ## vpn ## _simple_limit(p, stride, E) && \
FFABS(p3 - p2) <= I && FFABS(p2 - p1) <= I && \
FFABS(p1 - p0) <= I && FFABS(q3 - q2) <= I && \
FFABS(q2 - q1) <= I && FFABS(q1 - q0) <= I; \
}

NORMAL_LIMIT(7)
@@ -350,90 +368,89 @@ static av_always_inline void filter_mbedge(uint8_t *p, ptrdiff_t stride)

#define LOOP_FILTER(vpn, dir, size, stridea, strideb, maybe_inline) \
static maybe_inline \
void vp ## vpn ## _ ## dir ## _loop_filter ## size ## _c(uint8_t *dst, \
ptrdiff_t stride, \
int flim_E, int flim_I, \
int hev_thresh) \
void vpn ## _ ## dir ## _loop_filter ## size ## _c(uint8_t *dst, \
ptrdiff_t stride, \
int flim_E, int flim_I, \
int hev_thresh) \
{ \
int i; \
for (i = 0; i < size; i++) \
if (vp ## vpn ## _normal_limit(dst + i * stridea, strideb, flim_E, flim_I)) { \
if (vpn ## _normal_limit(dst + i * stridea, strideb, \
flim_E, flim_I)) { \
if (hev(dst + i * stridea, strideb, hev_thresh)) \
filter_common(dst + i * stridea, strideb, 1, vpn); \
vpn ## _filter_common(dst + i * stridea, strideb, 1); \
else \
filter_mbedge(dst + i * stridea, strideb); \
} \
} \
\
static maybe_inline \
void vp ## vpn ## _ ## dir ## _loop_filter ## size ## _inner_c(uint8_t *dst, \
ptrdiff_t stride, \
int flim_E, int flim_I, \
int hev_thresh) \
void vpn ## _ ## dir ## _loop_filter ## size ## _inner_c(uint8_t *dst, \
ptrdiff_t stride, \
int flim_E, \
int flim_I, \
int hev_thresh) \
{ \
int i; \
for (i = 0; i < size; i++) \
if (vp ## vpn ## _normal_limit(dst + i * stridea, strideb, flim_E, flim_I)) { \
if (vpn ## _normal_limit(dst + i * stridea, strideb, \
flim_E, flim_I)) { \
int hv = hev(dst + i * stridea, strideb, hev_thresh); \
if (hv) \
filter_common(dst + i * stridea, strideb, 1, vpn); \
vpn ## _filter_common(dst + i * stridea, strideb, 1); \
else \
filter_common(dst + i * stridea, strideb, 0, vpn); \
vpn ## _filter_common(dst + i * stridea, strideb, 0); \
} \
}

#define UV_LOOP_FILTER(vpn, dir, stridea, strideb) \
LOOP_FILTER(vpn, dir, 8, stridea, strideb, av_always_inline) \
static void vp ## vpn ## _ ## dir ## _loop_filter8uv_c(uint8_t *dstU, uint8_t *dstV, \
ptrdiff_t stride, int fE, \
int fI, int hev_thresh)\
{\
vp ## vpn ## _ ## dir ## _loop_filter8_c(dstU, stride, fE, fI, hev_thresh);\
vp ## vpn ## _ ## dir ## _loop_filter8_c(dstV, stride, fE, fI, hev_thresh);\
}\
static void vp ## vpn ## _ ## dir ## _loop_filter8uv_inner_c(uint8_t *dstU, \
uint8_t *dstV, \
ptrdiff_t stride, int fE, \
int fI, int hev_thresh) \
{\
vp ## vpn ## _ ## dir ## _loop_filter8_inner_c(dstU, stride, fE, fI, hev_thresh);\
vp ## vpn ## _ ## dir ## _loop_filter8_inner_c(dstV, stride, fE, fI, hev_thresh);\
#define UV_LOOP_FILTER(vpn, dir, stridea, strideb) \
LOOP_FILTER(vpn, dir, 8, stridea, strideb, av_always_inline) \
static void vpn ## _ ## dir ## _loop_filter8uv_c(uint8_t *dstU, \
uint8_t *dstV, \
ptrdiff_t stride, int fE, \
int fI, int hev_thresh) \
{ \
vpn ## _ ## dir ## _loop_filter8_c(dstU, stride, fE, fI, hev_thresh); \
vpn ## _ ## dir ## _loop_filter8_c(dstV, stride, fE, fI, hev_thresh); \
} \
\
static void vpn ## _ ## dir ## _loop_filter8uv_inner_c(uint8_t *dstU, \
uint8_t *dstV, \
ptrdiff_t stride, \
int fE, int fI, \
int hev_thresh) \
{ \
vpn ## _ ## dir ## _loop_filter8_inner_c(dstU, stride, fE, fI, \
hev_thresh); \
vpn ## _ ## dir ## _loop_filter8_inner_c(dstV, stride, fE, fI, \
hev_thresh); \
}

#define LOOP_FILTER_SIMPLE(vpn) \
static void vp ## vpn ## _v_loop_filter_simple_c(uint8_t *dst, ptrdiff_t stride, int flim)\
{\
int i;\
\
for (i = 0; i < 16; i++)\
if (vp ## vpn ## _simple_limit(dst + i, stride, flim))\
filter_common(dst + i, stride, 1, vpn);\
}\
\
static void vp ## vpn ## _h_loop_filter_simple_c(uint8_t *dst, ptrdiff_t stride, int flim)\
{\
int i;\
\
for (i = 0; i < 16; i++)\
if (vp ## vpn ## _simple_limit(dst + i * stride, 1, flim))\
filter_common(dst + i * stride, 1, 1, vpn);\
#define LOOP_FILTER_SIMPLE(vpn) \
static void vpn ## _v_loop_filter_simple_c(uint8_t *dst, ptrdiff_t stride, \
int flim) \
{ \
int i; \
for (i = 0; i < 16; i++) \
if (vpn ## _simple_limit(dst + i, stride, flim)) \
vpn ## _filter_common(dst + i, stride, 1); \
} \
\
static void vpn ## _h_loop_filter_simple_c(uint8_t *dst, ptrdiff_t stride, \
int flim) \
{ \
int i; \
for (i = 0; i < 16; i++) \
if (vpn ## _simple_limit(dst + i * stride, 1, flim)) \
vpn ## _filter_common(dst + i * stride, 1, 1); \
}

#if CONFIG_VP7_DECODER
LOOP_FILTER(7, v, 16, 1, stride,)
LOOP_FILTER(7, h, 16, stride, 1,)
UV_LOOP_FILTER(7, v, 1, stride)
UV_LOOP_FILTER(7, h, stride, 1)
LOOP_FILTER_SIMPLE(7)
#endif

#if CONFIG_VP8_DECODER
LOOP_FILTER(8, v, 16, 1, stride,)
LOOP_FILTER(8, h, 16, stride, 1,)
UV_LOOP_FILTER(8, v, 1, stride)
UV_LOOP_FILTER(8, h, stride, 1)
LOOP_FILTER_SIMPLE(8)
#endif
#define LOOP_FILTERS(vpn) \
LOOP_FILTER(vpn, v, 16, 1, stride, ) \
LOOP_FILTER(vpn, h, 16, stride, 1, ) \
UV_LOOP_FILTER(vpn, v, 1, stride) \
UV_LOOP_FILTER(vpn, h, stride, 1) \
LOOP_FILTER_SIMPLE(vpn) \

static const uint8_t subpel_filters[7][6] = {
{ 0, 6, 123, 12, 1, 0 },
@@ -622,7 +639,7 @@ VP8_BILINEAR(16)
VP8_BILINEAR(8)
VP8_BILINEAR(4)

#define VP8_MC_FUNC(IDX, SIZE) \
#define VP78_MC_FUNC(IDX, SIZE) \
dsp->put_vp8_epel_pixels_tab[IDX][0][0] = put_vp8_pixels ## SIZE ## _c; \
dsp->put_vp8_epel_pixels_tab[IDX][0][1] = put_vp8_epel ## SIZE ## _h4_c; \
dsp->put_vp8_epel_pixels_tab[IDX][0][2] = put_vp8_epel ## SIZE ## _h6_c; \
@@ -633,8 +650,8 @@ VP8_BILINEAR(4)
dsp->put_vp8_epel_pixels_tab[IDX][2][1] = put_vp8_epel ## SIZE ## _h4v6_c; \
dsp->put_vp8_epel_pixels_tab[IDX][2][2] = put_vp8_epel ## SIZE ## _h6v6_c

#define VP8_BILINEAR_MC_FUNC(IDX, SIZE) \
dsp->put_vp8_bilinear_pixels_tab[IDX][0][0] = put_vp8_pixels ## SIZE ## _c; \
#define VP78_BILINEAR_MC_FUNC(IDX, SIZE) \
dsp->put_vp8_bilinear_pixels_tab[IDX][0][0] = put_vp8_pixels ## SIZE ## _c; \
dsp->put_vp8_bilinear_pixels_tab[IDX][0][1] = put_vp8_bilinear ## SIZE ## _h_c; \
dsp->put_vp8_bilinear_pixels_tab[IDX][0][2] = put_vp8_bilinear ## SIZE ## _h_c; \
dsp->put_vp8_bilinear_pixels_tab[IDX][1][0] = put_vp8_bilinear ## SIZE ## _v_c; \
@@ -644,48 +661,79 @@ VP8_BILINEAR(4)
dsp->put_vp8_bilinear_pixels_tab[IDX][2][1] = put_vp8_bilinear ## SIZE ## _hv_c; \
dsp->put_vp8_bilinear_pixels_tab[IDX][2][2] = put_vp8_bilinear ## SIZE ## _hv_c

av_cold void ff_vp8dsp_init(VP8DSPContext *dsp, int vp7)
av_cold void ff_vp78dsp_init(VP8DSPContext *dsp)
{
#if CONFIG_VP7_DECODER && CONFIG_VP8_DECODER
#define VPX(f) vp7 ? vp7_ ## f : vp8_ ## f
#elif CONFIG_VP7_DECODER
#define VPX(f) vp7_ ## f
#else // CONFIG_VP8_DECODER
#define VPX(f) vp8_ ## f
#endif

dsp->vp8_luma_dc_wht = VPX(luma_dc_wht_c);
dsp->vp8_luma_dc_wht_dc = VPX(luma_dc_wht_dc_c);
dsp->vp8_idct_add = VPX(idct_add_c);
dsp->vp8_idct_dc_add = VPX(idct_dc_add_c);
dsp->vp8_idct_dc_add4y = VPX(idct_dc_add4y_c);
dsp->vp8_idct_dc_add4uv = VPX(idct_dc_add4uv_c);

dsp->vp8_v_loop_filter16y = VPX(v_loop_filter16_c);
dsp->vp8_h_loop_filter16y = VPX(h_loop_filter16_c);
dsp->vp8_v_loop_filter8uv = VPX(v_loop_filter8uv_c);
dsp->vp8_h_loop_filter8uv = VPX(h_loop_filter8uv_c);

dsp->vp8_v_loop_filter16y_inner = VPX(v_loop_filter16_inner_c);
dsp->vp8_h_loop_filter16y_inner = VPX(h_loop_filter16_inner_c);
dsp->vp8_v_loop_filter8uv_inner = VPX(v_loop_filter8uv_inner_c);
dsp->vp8_h_loop_filter8uv_inner = VPX(h_loop_filter8uv_inner_c);

dsp->vp8_v_loop_filter_simple = VPX(v_loop_filter_simple_c);
dsp->vp8_h_loop_filter_simple = VPX(h_loop_filter_simple_c);

VP8_MC_FUNC(0, 16);
VP8_MC_FUNC(1, 8);
VP8_MC_FUNC(2, 4);

VP8_BILINEAR_MC_FUNC(0, 16);
VP8_BILINEAR_MC_FUNC(1, 8);
VP8_BILINEAR_MC_FUNC(2, 4);
VP78_MC_FUNC(0, 16);
VP78_MC_FUNC(1, 8);
VP78_MC_FUNC(2, 4);

VP78_BILINEAR_MC_FUNC(0, 16);
VP78_BILINEAR_MC_FUNC(1, 8);
VP78_BILINEAR_MC_FUNC(2, 4);

if (ARCH_ARM)
ff_vp8dsp_init_arm(dsp, vp7);
ff_vp78dsp_init_arm(dsp);
if (ARCH_PPC)
ff_vp8dsp_init_ppc(dsp);
ff_vp78dsp_init_ppc(dsp);
if (ARCH_X86)
ff_vp78dsp_init_x86(dsp);
}

#if CONFIG_VP7_DECODER
LOOP_FILTERS(vp7)

av_cold void ff_vp7dsp_init(VP8DSPContext *dsp)
{
dsp->vp8_luma_dc_wht = vp7_luma_dc_wht_c;
dsp->vp8_luma_dc_wht_dc = vp7_luma_dc_wht_dc_c;
dsp->vp8_idct_add = vp7_idct_add_c;
dsp->vp8_idct_dc_add = vp7_idct_dc_add_c;
dsp->vp8_idct_dc_add4y = vp7_idct_dc_add4y_c;
dsp->vp8_idct_dc_add4uv = vp7_idct_dc_add4uv_c;

dsp->vp8_v_loop_filter16y = vp7_v_loop_filter16_c;
dsp->vp8_h_loop_filter16y = vp7_h_loop_filter16_c;
dsp->vp8_v_loop_filter8uv = vp7_v_loop_filter8uv_c;
dsp->vp8_h_loop_filter8uv = vp7_h_loop_filter8uv_c;

dsp->vp8_v_loop_filter16y_inner = vp7_v_loop_filter16_inner_c;
dsp->vp8_h_loop_filter16y_inner = vp7_h_loop_filter16_inner_c;
dsp->vp8_v_loop_filter8uv_inner = vp7_v_loop_filter8uv_inner_c;
dsp->vp8_h_loop_filter8uv_inner = vp7_h_loop_filter8uv_inner_c;

dsp->vp8_v_loop_filter_simple = vp7_v_loop_filter_simple_c;
dsp->vp8_h_loop_filter_simple = vp7_h_loop_filter_simple_c;
}
#endif /* CONFIG_VP7_DECODER */

#if CONFIG_VP8_DECODER
LOOP_FILTERS(vp8)

av_cold void ff_vp8dsp_init(VP8DSPContext *dsp)
{
dsp->vp8_luma_dc_wht = vp8_luma_dc_wht_c;
dsp->vp8_luma_dc_wht_dc = vp8_luma_dc_wht_dc_c;
dsp->vp8_idct_add = vp8_idct_add_c;
dsp->vp8_idct_dc_add = vp8_idct_dc_add_c;
dsp->vp8_idct_dc_add4y = vp8_idct_dc_add4y_c;
dsp->vp8_idct_dc_add4uv = vp8_idct_dc_add4uv_c;

dsp->vp8_v_loop_filter16y = vp8_v_loop_filter16_c;
dsp->vp8_h_loop_filter16y = vp8_h_loop_filter16_c;
dsp->vp8_v_loop_filter8uv = vp8_v_loop_filter8uv_c;
dsp->vp8_h_loop_filter8uv = vp8_h_loop_filter8uv_c;

dsp->vp8_v_loop_filter16y_inner = vp8_v_loop_filter16_inner_c;
dsp->vp8_h_loop_filter16y_inner = vp8_h_loop_filter16_inner_c;
dsp->vp8_v_loop_filter8uv_inner = vp8_v_loop_filter8uv_inner_c;
dsp->vp8_h_loop_filter8uv_inner = vp8_h_loop_filter8uv_inner_c;

dsp->vp8_v_loop_filter_simple = vp8_v_loop_filter_simple_c;
dsp->vp8_h_loop_filter_simple = vp8_h_loop_filter_simple_c;

if (ARCH_ARM)
ff_vp8dsp_init_arm(dsp);
if (ARCH_X86)
ff_vp8dsp_init_x86(dsp, vp7);
ff_vp8dsp_init_x86(dsp);
}
#endif /* CONFIG_VP8_DECODER */

+ 13
- 4
libavcodec/vp8dsp.h View File

@@ -88,9 +88,18 @@ void ff_put_vp8_pixels8_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride,
void ff_put_vp8_pixels4_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride,
int h, int x, int y);

void ff_vp8dsp_init(VP8DSPContext *c, int vp7);
void ff_vp8dsp_init_x86(VP8DSPContext *c, int vp7);
void ff_vp8dsp_init_arm(VP8DSPContext *c, int vp7);
void ff_vp8dsp_init_ppc(VP8DSPContext *c);
void ff_vp7dsp_init(VP8DSPContext *c);

void ff_vp78dsp_init(VP8DSPContext *c);
void ff_vp78dsp_init_arm(VP8DSPContext *c);
void ff_vp78dsp_init_ppc(VP8DSPContext *c);
void ff_vp78dsp_init_x86(VP8DSPContext *c);

void ff_vp8dsp_init(VP8DSPContext *c);
void ff_vp8dsp_init_arm(VP8DSPContext *c);
void ff_vp8dsp_init_x86(VP8DSPContext *c);

#define IS_VP7 1
#define IS_VP8 0

#endif /* AVCODEC_VP8DSP_H */

+ 2
- 1
libavcodec/x86/h264_intrapred_init.c View File

@@ -231,7 +231,8 @@ av_cold void ff_h264_pred_init_x86(H264PredContext *h, int codec_id,
h->pred4x4 [VERT_RIGHT_PRED ] = ff_pred4x4_vertical_right_8_mmxext;
h->pred4x4 [HOR_DOWN_PRED ] = ff_pred4x4_horizontal_down_8_mmxext;
h->pred4x4 [DC_PRED ] = ff_pred4x4_dc_8_mmxext;
if (codec_id == AV_CODEC_ID_VP7 || codec_id == AV_CODEC_ID_VP8 || codec_id == AV_CODEC_ID_H264) {
if (codec_id == AV_CODEC_ID_VP7 || codec_id == AV_CODEC_ID_VP8 ||
codec_id == AV_CODEC_ID_H264) {
h->pred4x4 [DIAG_DOWN_LEFT_PRED] = ff_pred4x4_down_left_8_mmxext;
}
if (codec_id == AV_CODEC_ID_SVQ3 || codec_id == AV_CODEC_ID_H264) {


+ 92
- 87
libavcodec/x86/vp8dsp_init.c View File

@@ -315,44 +315,18 @@ DECLARE_LOOP_FILTER(sse4)
c->put_vp8_bilinear_pixels_tab[IDX][2][2] = ff_put_vp8_bilinear ## SIZE ## _hv_ ## OPT


av_cold void ff_vp8dsp_init_x86(VP8DSPContext* c, int vp7)
av_cold void ff_vp78dsp_init_x86(VP8DSPContext *c)
{
#if HAVE_YASM
int cpu_flags = av_get_cpu_flags();

if (EXTERNAL_MMX(cpu_flags)) {
if (!vp7) {
c->vp8_idct_dc_add = ff_vp8_idct_dc_add_mmx;
c->vp8_idct_dc_add4uv = ff_vp8_idct_dc_add4uv_mmx;
}
#if ARCH_X86_32
if (!vp7) {
c->vp8_idct_dc_add4y = ff_vp8_idct_dc_add4y_mmx;
c->vp8_idct_add = ff_vp8_idct_add_mmx;
c->vp8_luma_dc_wht = ff_vp8_luma_dc_wht_mmx;
}
c->put_vp8_epel_pixels_tab[0][0][0] =
c->put_vp8_bilinear_pixels_tab[0][0][0] = ff_put_vp8_pixels16_mmx;
#endif
c->put_vp8_epel_pixels_tab[1][0][0] =
c->put_vp8_bilinear_pixels_tab[1][0][0] = ff_put_vp8_pixels8_mmx;

#if ARCH_X86_32
if (!vp7) {
c->vp8_v_loop_filter_simple = ff_vp8_v_loop_filter_simple_mmx;
c->vp8_h_loop_filter_simple = ff_vp8_h_loop_filter_simple_mmx;

c->vp8_v_loop_filter16y_inner = ff_vp8_v_loop_filter16y_inner_mmx;
c->vp8_h_loop_filter16y_inner = ff_vp8_h_loop_filter16y_inner_mmx;
c->vp8_v_loop_filter8uv_inner = ff_vp8_v_loop_filter8uv_inner_mmx;
c->vp8_h_loop_filter8uv_inner = ff_vp8_h_loop_filter8uv_inner_mmx;

c->vp8_v_loop_filter16y = ff_vp8_v_loop_filter16y_mbedge_mmx;
c->vp8_h_loop_filter16y = ff_vp8_h_loop_filter16y_mbedge_mmx;
c->vp8_v_loop_filter8uv = ff_vp8_v_loop_filter8uv_mbedge_mmx;
c->vp8_h_loop_filter8uv = ff_vp8_h_loop_filter8uv_mbedge_mmx;
}
#endif
}

/* note that 4-tap width=16 functions are missing because w=16
@@ -365,29 +339,10 @@ av_cold void ff_vp8dsp_init_x86(VP8DSPContext* c, int vp7)
VP8_MC_FUNC(1, 8, mmxext);
VP8_BILINEAR_MC_FUNC(0, 16, mmxext);
VP8_BILINEAR_MC_FUNC(1, 8, mmxext);

if (!vp7) {
c->vp8_v_loop_filter_simple = ff_vp8_v_loop_filter_simple_mmxext;
c->vp8_h_loop_filter_simple = ff_vp8_h_loop_filter_simple_mmxext;

c->vp8_v_loop_filter16y_inner = ff_vp8_v_loop_filter16y_inner_mmxext;
c->vp8_h_loop_filter16y_inner = ff_vp8_h_loop_filter16y_inner_mmxext;
c->vp8_v_loop_filter8uv_inner = ff_vp8_v_loop_filter8uv_inner_mmxext;
c->vp8_h_loop_filter8uv_inner = ff_vp8_h_loop_filter8uv_inner_mmxext;

c->vp8_v_loop_filter16y = ff_vp8_v_loop_filter16y_mbedge_mmxext;
c->vp8_h_loop_filter16y = ff_vp8_h_loop_filter16y_mbedge_mmxext;
c->vp8_v_loop_filter8uv = ff_vp8_v_loop_filter8uv_mbedge_mmxext;
c->vp8_h_loop_filter8uv = ff_vp8_h_loop_filter8uv_mbedge_mmxext;
}
#endif
}

if (EXTERNAL_SSE(cpu_flags)) {
if (!vp7) {
c->vp8_idct_add = ff_vp8_idct_add_sse;
c->vp8_luma_dc_wht = ff_vp8_luma_dc_wht_sse;
}
c->put_vp8_epel_pixels_tab[0][0][0] =
c->put_vp8_bilinear_pixels_tab[0][0][0] = ff_put_vp8_pixels16_sse;
}
@@ -397,64 +352,114 @@ av_cold void ff_vp8dsp_init_x86(VP8DSPContext* c, int vp7)
VP8_MC_FUNC(1, 8, sse2);
VP8_BILINEAR_MC_FUNC(0, 16, sse2);
VP8_BILINEAR_MC_FUNC(1, 8, sse2);
}

if (!vp7) {
c->vp8_v_loop_filter_simple = ff_vp8_v_loop_filter_simple_sse2;
if (EXTERNAL_SSSE3(cpu_flags)) {
VP8_LUMA_MC_FUNC(0, 16, ssse3);
VP8_MC_FUNC(1, 8, ssse3);
VP8_MC_FUNC(2, 4, ssse3);
VP8_BILINEAR_MC_FUNC(0, 16, ssse3);
VP8_BILINEAR_MC_FUNC(1, 8, ssse3);
VP8_BILINEAR_MC_FUNC(2, 4, ssse3);
}
#endif /* HAVE_YASM */
}

c->vp8_v_loop_filter16y_inner = ff_vp8_v_loop_filter16y_inner_sse2;
c->vp8_v_loop_filter8uv_inner = ff_vp8_v_loop_filter8uv_inner_sse2;
av_cold void ff_vp8dsp_init_x86(VP8DSPContext *c)
{
#if HAVE_YASM
int cpu_flags = av_get_cpu_flags();

c->vp8_v_loop_filter16y = ff_vp8_v_loop_filter16y_mbedge_sse2;
c->vp8_v_loop_filter8uv = ff_vp8_v_loop_filter8uv_mbedge_sse2;
}
if (EXTERNAL_MMX(cpu_flags)) {
c->vp8_idct_dc_add = ff_vp8_idct_dc_add_mmx;
c->vp8_idct_dc_add4uv = ff_vp8_idct_dc_add4uv_mmx;
#if ARCH_X86_32
c->vp8_idct_dc_add4y = ff_vp8_idct_dc_add4y_mmx;
c->vp8_idct_add = ff_vp8_idct_add_mmx;
c->vp8_luma_dc_wht = ff_vp8_luma_dc_wht_mmx;

c->vp8_v_loop_filter_simple = ff_vp8_v_loop_filter_simple_mmx;
c->vp8_h_loop_filter_simple = ff_vp8_h_loop_filter_simple_mmx;

c->vp8_v_loop_filter16y_inner = ff_vp8_v_loop_filter16y_inner_mmx;
c->vp8_h_loop_filter16y_inner = ff_vp8_h_loop_filter16y_inner_mmx;
c->vp8_v_loop_filter8uv_inner = ff_vp8_v_loop_filter8uv_inner_mmx;
c->vp8_h_loop_filter8uv_inner = ff_vp8_h_loop_filter8uv_inner_mmx;

c->vp8_v_loop_filter16y = ff_vp8_v_loop_filter16y_mbedge_mmx;
c->vp8_h_loop_filter16y = ff_vp8_h_loop_filter16y_mbedge_mmx;
c->vp8_v_loop_filter8uv = ff_vp8_v_loop_filter8uv_mbedge_mmx;
c->vp8_h_loop_filter8uv = ff_vp8_h_loop_filter8uv_mbedge_mmx;
#endif
}

if (EXTERNAL_SSE2(cpu_flags)) {
if (!vp7) {
c->vp8_idct_dc_add4y = ff_vp8_idct_dc_add4y_sse2;
/* note that 4-tap width=16 functions are missing because w=16
* is only used for luma, and luma is always a copy or sixtap. */
if (EXTERNAL_MMXEXT(cpu_flags)) {
#if ARCH_X86_32
c->vp8_v_loop_filter_simple = ff_vp8_v_loop_filter_simple_mmxext;
c->vp8_h_loop_filter_simple = ff_vp8_h_loop_filter_simple_mmxext;

c->vp8_v_loop_filter16y_inner = ff_vp8_v_loop_filter16y_inner_mmxext;
c->vp8_h_loop_filter16y_inner = ff_vp8_h_loop_filter16y_inner_mmxext;
c->vp8_v_loop_filter8uv_inner = ff_vp8_v_loop_filter8uv_inner_mmxext;
c->vp8_h_loop_filter8uv_inner = ff_vp8_h_loop_filter8uv_inner_mmxext;

c->vp8_v_loop_filter16y = ff_vp8_v_loop_filter16y_mbedge_mmxext;
c->vp8_h_loop_filter16y = ff_vp8_h_loop_filter16y_mbedge_mmxext;
c->vp8_v_loop_filter8uv = ff_vp8_v_loop_filter8uv_mbedge_mmxext;
c->vp8_h_loop_filter8uv = ff_vp8_h_loop_filter8uv_mbedge_mmxext;
#endif
}

c->vp8_h_loop_filter_simple = ff_vp8_h_loop_filter_simple_sse2;
if (EXTERNAL_SSE(cpu_flags)) {
c->vp8_idct_add = ff_vp8_idct_add_sse;
c->vp8_luma_dc_wht = ff_vp8_luma_dc_wht_sse;
}

if (HAVE_SSE2_EXTERNAL && cpu_flags & (AV_CPU_FLAG_SSE2 | AV_CPU_FLAG_SSE2SLOW)) {
c->vp8_v_loop_filter_simple = ff_vp8_v_loop_filter_simple_sse2;

c->vp8_h_loop_filter16y_inner = ff_vp8_h_loop_filter16y_inner_sse2;
c->vp8_h_loop_filter8uv_inner = ff_vp8_h_loop_filter8uv_inner_sse2;
c->vp8_v_loop_filter16y_inner = ff_vp8_v_loop_filter16y_inner_sse2;
c->vp8_v_loop_filter8uv_inner = ff_vp8_v_loop_filter8uv_inner_sse2;

c->vp8_h_loop_filter16y = ff_vp8_h_loop_filter16y_mbedge_sse2;
c->vp8_h_loop_filter8uv = ff_vp8_h_loop_filter8uv_mbedge_sse2;
}
c->vp8_v_loop_filter16y = ff_vp8_v_loop_filter16y_mbedge_sse2;
c->vp8_v_loop_filter8uv = ff_vp8_v_loop_filter8uv_mbedge_sse2;
}

if (EXTERNAL_SSSE3(cpu_flags)) {
VP8_LUMA_MC_FUNC(0, 16, ssse3);
VP8_MC_FUNC(1, 8, ssse3);
VP8_MC_FUNC(2, 4, ssse3);
VP8_BILINEAR_MC_FUNC(0, 16, ssse3);
VP8_BILINEAR_MC_FUNC(1, 8, ssse3);
VP8_BILINEAR_MC_FUNC(2, 4, ssse3);
if (EXTERNAL_SSE2(cpu_flags)) {
c->vp8_idct_dc_add4y = ff_vp8_idct_dc_add4y_sse2;

if (!vp7) {
c->vp8_v_loop_filter_simple = ff_vp8_v_loop_filter_simple_ssse3;
c->vp8_h_loop_filter_simple = ff_vp8_h_loop_filter_simple_ssse3;
c->vp8_h_loop_filter_simple = ff_vp8_h_loop_filter_simple_sse2;

c->vp8_v_loop_filter16y_inner = ff_vp8_v_loop_filter16y_inner_ssse3;
c->vp8_h_loop_filter16y_inner = ff_vp8_h_loop_filter16y_inner_ssse3;
c->vp8_v_loop_filter8uv_inner = ff_vp8_v_loop_filter8uv_inner_ssse3;
c->vp8_h_loop_filter8uv_inner = ff_vp8_h_loop_filter8uv_inner_ssse3;
c->vp8_h_loop_filter16y_inner = ff_vp8_h_loop_filter16y_inner_sse2;
c->vp8_h_loop_filter8uv_inner = ff_vp8_h_loop_filter8uv_inner_sse2;

c->vp8_v_loop_filter16y = ff_vp8_v_loop_filter16y_mbedge_ssse3;
c->vp8_h_loop_filter16y = ff_vp8_h_loop_filter16y_mbedge_ssse3;
c->vp8_v_loop_filter8uv = ff_vp8_v_loop_filter8uv_mbedge_ssse3;
c->vp8_h_loop_filter8uv = ff_vp8_h_loop_filter8uv_mbedge_ssse3;
}
c->vp8_h_loop_filter16y = ff_vp8_h_loop_filter16y_mbedge_sse2;
c->vp8_h_loop_filter8uv = ff_vp8_h_loop_filter8uv_mbedge_sse2;
}

if (EXTERNAL_SSSE3(cpu_flags)) {
c->vp8_v_loop_filter_simple = ff_vp8_v_loop_filter_simple_ssse3;
c->vp8_h_loop_filter_simple = ff_vp8_h_loop_filter_simple_ssse3;

c->vp8_v_loop_filter16y_inner = ff_vp8_v_loop_filter16y_inner_ssse3;
c->vp8_h_loop_filter16y_inner = ff_vp8_h_loop_filter16y_inner_ssse3;
c->vp8_v_loop_filter8uv_inner = ff_vp8_v_loop_filter8uv_inner_ssse3;
c->vp8_h_loop_filter8uv_inner = ff_vp8_h_loop_filter8uv_inner_ssse3;

c->vp8_v_loop_filter16y = ff_vp8_v_loop_filter16y_mbedge_ssse3;
c->vp8_h_loop_filter16y = ff_vp8_h_loop_filter16y_mbedge_ssse3;
c->vp8_v_loop_filter8uv = ff_vp8_v_loop_filter8uv_mbedge_ssse3;
c->vp8_h_loop_filter8uv = ff_vp8_h_loop_filter8uv_mbedge_ssse3;
}

if (EXTERNAL_SSE4(cpu_flags)) {
if (!vp7) {
c->vp8_idct_dc_add = ff_vp8_idct_dc_add_sse4;
c->vp8_idct_dc_add = ff_vp8_idct_dc_add_sse4;

c->vp8_h_loop_filter_simple = ff_vp8_h_loop_filter_simple_sse4;
c->vp8_h_loop_filter16y = ff_vp8_h_loop_filter16y_mbedge_sse4;
c->vp8_h_loop_filter8uv = ff_vp8_h_loop_filter8uv_mbedge_sse4;
}
c->vp8_h_loop_filter_simple = ff_vp8_h_loop_filter_simple_sse4;
c->vp8_h_loop_filter16y = ff_vp8_h_loop_filter16y_mbedge_sse4;
c->vp8_h_loop_filter8uv = ff_vp8_h_loop_filter8uv_mbedge_sse4;
}
#endif /* HAVE_YASM */
}

Write Preview
Loading…
Cancel
Save