generic-library/vpx
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Merge "High bit-depth support for wedge partition expt" into nextgen

Browse Source
This commit is contained in:
Debargha Mukherjee 2015-04-13 10:09:51 -07:00 committed by Gerrit Code Review
commit febb434356
5 changed files with 818 additions and 12 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

82
vp9/common/vp9_reconinter.c
View File

@ -551,6 +551,28 @@ static void build_masked_compound(uint8_t *dst, int dst_stride,
}
}
#if CONFIG_VP9_HIGHBITDEPTH
static void build_masked_compound_highbd(uint8_t *dst_8, int dst_stride,
uint8_t *dst2_8, int dst2_stride,
int wedge_index, BLOCK_SIZE sb_type,
int h, int w) {
int i, j;
uint8_t mask[4096];
uint16_t *dst = CONVERT_TO_SHORTPTR(dst_8);
uint16_t *dst2 = CONVERT_TO_SHORTPTR(dst2_8);
vp9_generate_masked_weight(wedge_index, sb_type, h, w, mask, 64);
for (i = 0; i < h; ++i)
for (j = 0; j < w; ++j) {
int m = mask[i * 64 + j];
dst[i * dst_stride + j] = (dst[i * dst_stride + j] * m +
dst2[i * dst2_stride + j] *
((1 << WEDGE_WEIGHT_BITS) - m) +
(1 << (WEDGE_WEIGHT_BITS - 1))) >>
WEDGE_WEIGHT_BITS;
}
}
#endif // CONFIG_VP9_HIGHBITDEPTH
#if CONFIG_SUPERTX
void generate_masked_weight_extend(int wedge_index, int plane,
BLOCK_SIZE sb_type, int h, int w,
@ -686,7 +708,14 @@ static void build_inter_predictors(MACROBLOCKD *xd, int plane, int block,
#if CONFIG_WEDGE_PARTITION
if (ref && get_wedge_bits(mi->mbmi.sb_type)
&& mi->mbmi.use_wedge_interinter) {
#if CONFIG_VP9_HIGHBITDEPTH
uint8_t tmp_dst_[8192];
uint8_t *tmp_dst =
(xd->cur_buf->flags & YV12_FLAG_HIGHBITDEPTH) ?
CONVERT_TO_BYTEPTR(tmp_dst_) : tmp_dst_;
#else
uint8_t tmp_dst[4096];
#endif
#if CONFIG_VP9_HIGHBITDEPTH
if (xd->cur_buf->flags & YV12_FLAG_HIGHBITDEPTH) {
highbd_inter_predictor(pre, pre_buf->stride, tmp_dst, 64,
@ -722,13 +751,27 @@ static void build_inter_predictors(MACROBLOCKD *xd, int plane, int block,
wedge_offset_x, wedge_offset_y, h, w);
#endif // CONFIG_VP9_HIGHBITDEPTH
#else // CONFIG_SUPERTX
build_masked_compound(dst, dst_buf->stride, tmp_dst, 64,
mi->mbmi.interinter_wedge_index, mi->mbmi.sb_type,
h, w);
#if CONFIG_VP9_HIGHBITDEPTH
if (xd->cur_buf->flags & YV12_FLAG_HIGHBITDEPTH)
build_masked_compound_highbd(dst, dst_buf->stride, tmp_dst, 64,
mi->mbmi.interinter_wedge_index,
mi->mbmi.sb_type, h, w);
else
#endif // CONFIG_VP9_HIGHBITDEPTH
build_masked_compound(dst, dst_buf->stride, tmp_dst, 64,
mi->mbmi.interinter_wedge_index, mi->mbmi.sb_type,
h, w);
#endif // CONFIG_SUPERTX
} else {
inter_predictor(pre, pre_buf->stride, dst, dst_buf->stride,
subpel_x, subpel_y, sf, w, h, ref, kernel, xs, ys);
#if CONFIG_VP9_HIGHBITDEPTH
if (xd->cur_buf->flags & YV12_FLAG_HIGHBITDEPTH)
highbd_inter_predictor(pre, pre_buf->stride, dst, dst_buf->stride,
subpel_x, subpel_y, sf, w, h, ref, kernel,
xs, ys, xd->bd);
else
#endif // CONFIG_VP9_HIGHBITDEPTH
inter_predictor(pre, pre_buf->stride, dst, dst_buf->stride,
subpel_x, subpel_y, sf, w, h, ref, kernel, xs, ys);
}
#else // CONFIG_WEDGE_PARTITION
@ -1410,7 +1453,14 @@ static void dec_build_inter_predictors(MACROBLOCKD *xd, int plane, int block,
#if CONFIG_WEDGE_PARTITION
if (ref && get_wedge_bits(mi->mbmi.sb_type)
&& mi->mbmi.use_wedge_interinter) {
#if CONFIG_VP9_HIGHBITDEPTH
uint8_t tmp_dst_[8192];
uint8_t *tmp_dst =
(xd->cur_buf->flags & YV12_FLAG_HIGHBITDEPTH) ?
CONVERT_TO_BYTEPTR(tmp_dst_) : tmp_dst_;
#else
uint8_t tmp_dst[4096];
#endif
#if CONFIG_VP9_HIGHBITDEPTH
if (xd->cur_buf->flags & YV12_FLAG_HIGHBITDEPTH) {
highbd_inter_predictor(buf_ptr, buf_stride, tmp_dst, 64,
@ -1445,24 +1495,32 @@ static void dec_build_inter_predictors(MACROBLOCKD *xd, int plane, int block,
wedge_offset_x, wedge_offset_y, h, w);
#endif // CONFIG_VP9_HIGHBITDEPTH
#else // CONFIG_SUPERTX
#if CONFIG_VP9_HIGHBITDEPTH
if (xd->cur_buf->flags & YV12_FLAG_HIGHBITDEPTH) {
build_masked_compound_highbd(dst, dst_buf->stride, tmp_dst, 64,
mi->mbmi.interinter_wedge_index,
mi->mbmi.sb_type, h, w);
} else {
build_masked_compound(dst, dst_buf->stride, tmp_dst, 64,
mi->mbmi.interinter_wedge_index, mi->mbmi.sb_type,
h, w);
}
#else
build_masked_compound(dst, dst_buf->stride, tmp_dst, 64,
mi->mbmi.interinter_wedge_index, mi->mbmi.sb_type,
h, w);
#endif // CONFIG_VP9_HIGHBITDEPTH
#endif // CONFIG_SUPERTX
} else {
#if CONFIG_VP9_HIGHBITDEPTH
if (xd->cur_buf->flags & YV12_FLAG_HIGHBITDEPTH) {
if (xd->cur_buf->flags & YV12_FLAG_HIGHBITDEPTH)
highbd_inter_predictor(buf_ptr, buf_stride, dst, dst_buf->stride,
subpel_x, subpel_y, sf, w, h, ref, kernel,
xs, ys, xd->bd);
} else {
else
#endif // CONFIG_VP9_HIGHBITDEPTH
inter_predictor(buf_ptr, buf_stride, dst, dst_buf->stride, subpel_x,
subpel_y, sf, w, h, ref, kernel, xs, ys);
}
#else
inter_predictor(buf_ptr, buf_stride, dst, dst_buf->stride, subpel_x,
subpel_y, sf, w, h, ref, kernel, xs, ys);
#endif // CONFIG_VP9_HIGHBITDEPTH
}
#else // CONFIG_WEDGE_PARTITION

275
vp9/common/vp9_rtcd_defs.pl
View File

@ -2127,6 +2127,281 @@ if (vpx_config("CONFIG_VP9_HIGHBITDEPTH") eq "yes") {
add_proto qw/unsigned int vp9_highbd_12_mse8x8/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int recon_stride, unsigned int *sse";
specialize qw/vp9_highbd_12_mse8x8/;
if (vpx_config("CONFIG_WEDGE_PARTITION") eq "yes") {
add_proto qw/unsigned int vp9_highbd_masked_variance32x16/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride, unsigned int *sse";
specialize qw/vp9_highbd_masked_variance32x16/;
add_proto qw/unsigned int vp9_highbd_masked_variance16x32/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride, unsigned int *sse";
specialize qw/vp9_highbd_masked_variance16x32/;
add_proto qw/unsigned int vp9_highbd_masked_variance64x32/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride, unsigned int *sse";
specialize qw/vp9_highbd_masked_variance64x32/;
add_proto qw/unsigned int vp9_highbd_masked_variance32x64/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride, unsigned int *sse";
specialize qw/vp9_highbd_masked_variance32x64/;
add_proto qw/unsigned int vp9_highbd_masked_variance32x32/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride, unsigned int *sse";
specialize qw/vp9_highbd_masked_variance32x32/;
add_proto qw/unsigned int vp9_highbd_masked_variance64x64/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride, unsigned int *sse";
specialize qw/vp9_highbd_masked_variance64x64/;
add_proto qw/unsigned int vp9_highbd_masked_variance16x16/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride, unsigned int *sse";
specialize qw/vp9_highbd_masked_variance16x16/;
add_proto qw/unsigned int vp9_highbd_masked_variance16x8/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride, unsigned int *sse";
specialize qw/vp9_highbd_masked_variance16x8/;
add_proto qw/unsigned int vp9_highbd_masked_variance8x16/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride, unsigned int *sse";
specialize qw/vp9_highbd_masked_variance8x16/;
add_proto qw/unsigned int vp9_highbd_masked_variance8x8/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride, unsigned int *sse";
specialize qw/vp9_highbd_masked_variance8x8/;
add_proto qw/unsigned int vp9_highbd_masked_variance8x4/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride, unsigned int *sse";
specialize qw/vp9_highbd_masked_variance8x4/;
add_proto qw/unsigned int vp9_highbd_masked_variance4x8/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride, unsigned int *sse";
specialize qw/vp9_highbd_masked_variance4x8/;
add_proto qw/unsigned int vp9_highbd_masked_variance4x4/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride, unsigned int *sse";
specialize qw/vp9_highbd_masked_variance4x4/;
add_proto qw/unsigned int vp9_highbd_10_masked_variance32x16/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride, unsigned int *sse";
specialize qw/vp9_highbd_10_masked_variance32x16/;
add_proto qw/unsigned int vp9_highbd_10_masked_variance16x32/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride, unsigned int *sse";
specialize qw/vp9_highbd_10_masked_variance16x32/;
add_proto qw/unsigned int vp9_highbd_10_masked_variance64x32/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride, unsigned int *sse";
specialize qw/vp9_highbd_10_masked_variance64x32/;
add_proto qw/unsigned int vp9_highbd_10_masked_variance32x64/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride, unsigned int *sse";
specialize qw/vp9_highbd_10_masked_variance32x64/;
add_proto qw/unsigned int vp9_highbd_10_masked_variance32x32/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride, unsigned int *sse";
specialize qw/vp9_highbd_10_masked_variance32x32/;
add_proto qw/unsigned int vp9_highbd_10_masked_variance64x64/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride, unsigned int *sse";
specialize qw/vp9_highbd_10_masked_variance64x64/;
add_proto qw/unsigned int vp9_highbd_10_masked_variance16x16/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride, unsigned int *sse";
specialize qw/vp9_highbd_10_masked_variance16x16/;
add_proto qw/unsigned int vp9_highbd_10_masked_variance16x8/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride, unsigned int *sse";
specialize qw/vp9_highbd_10_masked_variance16x8/;
add_proto qw/unsigned int vp9_highbd_10_masked_variance8x16/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride, unsigned int *sse";
specialize qw/vp9_highbd_10_masked_variance8x16/;
add_proto qw/unsigned int vp9_highbd_10_masked_variance8x8/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride, unsigned int *sse";
specialize qw/vp9_highbd_10_masked_variance8x8/;
add_proto qw/unsigned int vp9_highbd_10_masked_variance8x4/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride, unsigned int *sse";
specialize qw/vp9_highbd_10_masked_variance8x4/;
add_proto qw/unsigned int vp9_highbd_10_masked_variance4x8/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride, unsigned int *sse";
specialize qw/vp9_highbd_10_masked_variance4x8/;
add_proto qw/unsigned int vp9_highbd_10_masked_variance4x4/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride, unsigned int *sse";
specialize qw/vp9_highbd_10_masked_variance4x4/;
add_proto qw/unsigned int vp9_highbd_12_masked_variance32x16/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride, unsigned int *sse";
specialize qw/vp9_highbd_12_masked_variance32x16/;
add_proto qw/unsigned int vp9_highbd_12_masked_variance16x32/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride, unsigned int *sse";
specialize qw/vp9_highbd_12_masked_variance16x32/;
add_proto qw/unsigned int vp9_highbd_12_masked_variance64x32/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride, unsigned int *sse";
specialize qw/vp9_highbd_12_masked_variance64x32/;
add_proto qw/unsigned int vp9_highbd_12_masked_variance32x64/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride, unsigned int *sse";
specialize qw/vp9_highbd_12_masked_variance32x64/;
add_proto qw/unsigned int vp9_highbd_12_masked_variance32x32/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride, unsigned int *sse";
specialize qw/vp9_highbd_12_masked_variance32x32/;
add_proto qw/unsigned int vp9_highbd_12_masked_variance64x64/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride, unsigned int *sse";
specialize qw/vp9_highbd_12_masked_variance64x64/;
add_proto qw/unsigned int vp9_highbd_12_masked_variance16x16/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride, unsigned int *sse";
specialize qw/vp9_highbd_12_masked_variance16x16/;
add_proto qw/unsigned int vp9_highbd_12_masked_variance16x8/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride, unsigned int *sse";
specialize qw/vp9_highbd_12_masked_variance16x8/;
add_proto qw/unsigned int vp9_highbd_12_masked_variance8x16/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride, unsigned int *sse";
specialize qw/vp9_highbd_12_masked_variance8x16/;
add_proto qw/unsigned int vp9_highbd_12_masked_variance8x8/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride, unsigned int *sse";
specialize qw/vp9_highbd_12_masked_variance8x8/;
add_proto qw/unsigned int vp9_highbd_12_masked_variance8x4/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride, unsigned int *sse";
specialize qw/vp9_highbd_12_masked_variance8x4/;
add_proto qw/unsigned int vp9_highbd_12_masked_variance4x8/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride, unsigned int *sse";
specialize qw/vp9_highbd_12_masked_variance4x8/;
add_proto qw/unsigned int vp9_highbd_12_masked_variance4x4/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride, unsigned int *sse";
specialize qw/vp9_highbd_12_masked_variance4x4/;
add_proto qw/unsigned int vp9_highbd_masked_sub_pixel_variance64x64/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, const uint8_t *m, int m_stride, unsigned int *sse";
specialize qw/vp9_highbd_masked_sub_pixel_variance64x64/;
add_proto qw/unsigned int vp9_highbd_masked_sub_pixel_variance64x32/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, const uint8_t *m, int m_stride, unsigned int *sse";
specialize qw/vp9_highbd_masked_sub_pixel_variance64x32/;
add_proto qw/unsigned int vp9_highbd_masked_sub_pixel_variance32x64/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, const uint8_t *m, int m_stride, unsigned int *sse";
specialize qw/vp9_highbd_masked_sub_pixel_variance32x64/;
add_proto qw/unsigned int vp9_highbd_masked_sub_pixel_variance32x32/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, const uint8_t *m, int m_stride, unsigned int *sse";
specialize qw/vp9_highbd_masked_sub_pixel_variance32x32/;
add_proto qw/unsigned int vp9_highbd_masked_sub_pixel_variance32x16/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, const uint8_t *m, int m_stride, unsigned int *sse";
specialize qw/vp9_highbd_masked_sub_pixel_variance32x16/;
add_proto qw/unsigned int vp9_highbd_masked_sub_pixel_variance16x32/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, const uint8_t *m, int m_stride, unsigned int *sse";
specialize qw/vp9_highbd_masked_sub_pixel_variance16x32/;
add_proto qw/unsigned int vp9_highbd_masked_sub_pixel_variance16x16/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, const uint8_t *m, int m_stride, unsigned int *sse";
specialize qw/vp9_highbd_masked_sub_pixel_variance16x16/;
add_proto qw/unsigned int vp9_highbd_masked_sub_pixel_variance16x8/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, const uint8_t *m, int m_stride, unsigned int *sse";
specialize qw/vp9_highbd_masked_sub_pixel_variance16x8/;
add_proto qw/unsigned int vp9_highbd_masked_sub_pixel_variance8x16/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, const uint8_t *m, int m_stride, unsigned int *sse";
specialize qw/vp9_highbd_masked_sub_pixel_variance8x16/;
add_proto qw/unsigned int vp9_highbd_masked_sub_pixel_variance8x8/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, const uint8_t *m, int m_stride, unsigned int *sse";
specialize qw/vp9_highbd_masked_sub_pixel_variance8x8/;
add_proto qw/unsigned int vp9_highbd_masked_sub_pixel_variance8x4/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, const uint8_t *m, int m_stride, unsigned int *sse";
specialize qw/vp9_highbd_masked_sub_pixel_variance8x4/;
add_proto qw/unsigned int vp9_highbd_masked_sub_pixel_variance4x8/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, const uint8_t *m, int m_stride, unsigned int *sse";
specialize qw/vp9_highbd_masked_sub_pixel_variance4x8/;
add_proto qw/unsigned int vp9_highbd_masked_sub_pixel_variance4x4/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, const uint8_t *m, int m_stride, unsigned int *sse";
specialize qw/vp9_highbd_masked_sub_pixel_variance4x4/;
add_proto qw/unsigned int vp9_highbd_10_masked_sub_pixel_variance64x64/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, const uint8_t *m, int m_stride, unsigned int *sse";
specialize qw/vp9_highbd_10_masked_sub_pixel_variance64x64/;
add_proto qw/unsigned int vp9_highbd_10_masked_sub_pixel_variance64x32/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, const uint8_t *m, int m_stride, unsigned int *sse";
specialize qw/vp9_highbd_10_masked_sub_pixel_variance64x32/;
add_proto qw/unsigned int vp9_highbd_10_masked_sub_pixel_variance32x64/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, const uint8_t *m, int m_stride, unsigned int *sse";
specialize qw/vp9_highbd_10_masked_sub_pixel_variance32x64/;
add_proto qw/unsigned int vp9_highbd_10_masked_sub_pixel_variance32x32/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, const uint8_t *m, int m_stride, unsigned int *sse";
specialize qw/vp9_highbd_10_masked_sub_pixel_variance32x32/;
add_proto qw/unsigned int vp9_highbd_10_masked_sub_pixel_variance32x16/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, const uint8_t *m, int m_stride, unsigned int *sse";
specialize qw/vp9_highbd_10_masked_sub_pixel_variance32x16/;
add_proto qw/unsigned int vp9_highbd_10_masked_sub_pixel_variance16x32/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, const uint8_t *m, int m_stride, unsigned int *sse";
specialize qw/vp9_highbd_10_masked_sub_pixel_variance16x32/;
add_proto qw/unsigned int vp9_highbd_10_masked_sub_pixel_variance16x16/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, const uint8_t *m, int m_stride, unsigned int *sse";
specialize qw/vp9_highbd_10_masked_sub_pixel_variance16x16/;
add_proto qw/unsigned int vp9_highbd_10_masked_sub_pixel_variance16x8/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, const uint8_t *m, int m_stride, unsigned int *sse";
specialize qw/vp9_highbd_10_masked_sub_pixel_variance16x8/;
add_proto qw/unsigned int vp9_highbd_10_masked_sub_pixel_variance8x16/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, const uint8_t *m, int m_stride, unsigned int *sse";
specialize qw/vp9_highbd_10_masked_sub_pixel_variance8x16/;
add_proto qw/unsigned int vp9_highbd_10_masked_sub_pixel_variance8x8/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, const uint8_t *m, int m_stride, unsigned int *sse";
specialize qw/vp9_highbd_10_masked_sub_pixel_variance8x8/;
add_proto qw/unsigned int vp9_highbd_10_masked_sub_pixel_variance8x4/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, const uint8_t *m, int m_stride, unsigned int *sse";
specialize qw/vp9_highbd_10_masked_sub_pixel_variance8x4/;
add_proto qw/unsigned int vp9_highbd_10_masked_sub_pixel_variance4x8/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, const uint8_t *m, int m_stride, unsigned int *sse";
specialize qw/vp9_highbd_10_masked_sub_pixel_variance4x8/;
add_proto qw/unsigned int vp9_highbd_10_masked_sub_pixel_variance4x4/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, const uint8_t *m, int m_stride, unsigned int *sse";
specialize qw/vp9_highbd_10_masked_sub_pixel_variance4x4/;
add_proto qw/unsigned int vp9_highbd_12_masked_sub_pixel_variance64x64/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, const uint8_t *m, int m_stride, unsigned int *sse";
specialize qw/vp9_highbd_12_masked_sub_pixel_variance64x64/;
add_proto qw/unsigned int vp9_highbd_12_masked_sub_pixel_variance64x32/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, const uint8_t *m, int m_stride, unsigned int *sse";
specialize qw/vp9_highbd_12_masked_sub_pixel_variance64x32/;
add_proto qw/unsigned int vp9_highbd_12_masked_sub_pixel_variance32x64/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, const uint8_t *m, int m_stride, unsigned int *sse";
specialize qw/vp9_highbd_12_masked_sub_pixel_variance32x64/;
add_proto qw/unsigned int vp9_highbd_12_masked_sub_pixel_variance32x32/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, const uint8_t *m, int m_stride, unsigned int *sse";
specialize qw/vp9_highbd_12_masked_sub_pixel_variance32x32/;
add_proto qw/unsigned int vp9_highbd_12_masked_sub_pixel_variance32x16/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, const uint8_t *m, int m_stride, unsigned int *sse";
specialize qw/vp9_highbd_12_masked_sub_pixel_variance32x16/;
add_proto qw/unsigned int vp9_highbd_12_masked_sub_pixel_variance16x32/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, const uint8_t *m, int m_stride, unsigned int *sse";
specialize qw/vp9_highbd_12_masked_sub_pixel_variance16x32/;
add_proto qw/unsigned int vp9_highbd_12_masked_sub_pixel_variance16x16/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, const uint8_t *m, int m_stride, unsigned int *sse";
specialize qw/vp9_highbd_12_masked_sub_pixel_variance16x16/;
add_proto qw/unsigned int vp9_highbd_12_masked_sub_pixel_variance16x8/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, const uint8_t *m, int m_stride, unsigned int *sse";
specialize qw/vp9_highbd_12_masked_sub_pixel_variance16x8/;
add_proto qw/unsigned int vp9_highbd_12_masked_sub_pixel_variance8x16/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, const uint8_t *m, int m_stride, unsigned int *sse";
specialize qw/vp9_highbd_12_masked_sub_pixel_variance8x16/;
add_proto qw/unsigned int vp9_highbd_12_masked_sub_pixel_variance8x8/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, const uint8_t *m, int m_stride, unsigned int *sse";
specialize qw/vp9_highbd_12_masked_sub_pixel_variance8x8/;
add_proto qw/unsigned int vp9_highbd_12_masked_sub_pixel_variance8x4/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, const uint8_t *m, int m_stride, unsigned int *sse";
specialize qw/vp9_highbd_12_masked_sub_pixel_variance8x4/;
add_proto qw/unsigned int vp9_highbd_12_masked_sub_pixel_variance4x8/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, const uint8_t *m, int m_stride, unsigned int *sse";
specialize qw/vp9_highbd_12_masked_sub_pixel_variance4x8/;
add_proto qw/unsigned int vp9_highbd_12_masked_sub_pixel_variance4x4/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, const uint8_t *m, int m_stride, unsigned int *sse";
specialize qw/vp9_highbd_12_masked_sub_pixel_variance4x4/;
add_proto qw/unsigned int vp9_highbd_masked_sad64x64/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride";
specialize qw/vp9_highbd_masked_sad64x64/;
add_proto qw/unsigned int vp9_highbd_masked_sad32x64/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride";
specialize qw/vp9_highbd_masked_sad32x64/;
add_proto qw/unsigned int vp9_highbd_masked_sad64x32/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride";
specialize qw/vp9_highbd_masked_sad64x32/;
add_proto qw/unsigned int vp9_highbd_masked_sad32x16/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride";
specialize qw/vp9_highbd_masked_sad32x16/;
add_proto qw/unsigned int vp9_highbd_masked_sad16x32/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride";
specialize qw/vp9_highbd_masked_sad16x32/;
add_proto qw/unsigned int vp9_highbd_masked_sad32x32/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride";
specialize qw/vp9_highbd_masked_sad32x32/;
add_proto qw/unsigned int vp9_highbd_masked_sad16x16/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride";
specialize qw/vp9_highbd_masked_sad16x16/;
add_proto qw/unsigned int vp9_highbd_masked_sad16x8/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride";
specialize qw/vp9_highbd_masked_sad16x8/;
add_proto qw/unsigned int vp9_highbd_masked_sad8x16/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride";
specialize qw/vp9_highbd_masked_sad8x16/;
add_proto qw/unsigned int vp9_highbd_masked_sad8x8/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride";
specialize qw/vp9_highbd_masked_sad8x8/;
add_proto qw/unsigned int vp9_highbd_masked_sad8x4/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride";
specialize qw/vp9_highbd_masked_sad8x4/;
add_proto qw/unsigned int vp9_highbd_masked_sad4x8/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride";
specialize qw/vp9_highbd_masked_sad4x8/;
add_proto qw/unsigned int vp9_highbd_masked_sad4x4/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *mask, int mask_stride";
specialize qw/vp9_highbd_masked_sad4x4/;
}
# ENCODEMB INVOKE
add_proto qw/int64_t vp9_highbd_block_error/, "const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz, int bd";

215
vp9/encoder/vp9_encoder.c
View File

@ -821,6 +821,56 @@ MAKE_BFP_SAD3_WRAPPER(vp9_highbd_sad4x4x3)
MAKE_BFP_SAD8_WRAPPER(vp9_highbd_sad4x4x8)
MAKE_BFP_SAD4D_WRAPPER(vp9_highbd_sad4x4x4d)
#if CONFIG_WEDGE_PARTITION
#define HIGHBD_MBFP(BT, MSDF, MVF, MSVF) \
cpi->fn_ptr[BT].msdf = MSDF; \
cpi->fn_ptr[BT].mvf = MVF; \
cpi->fn_ptr[BT].msvf = MSVF;
#define MAKE_MBFP_SAD_WRAPPER(fnname) \
static unsigned int fnname##_bits8(const uint8_t *src_ptr, \
int source_stride, \
const uint8_t *ref_ptr, \
int ref_stride, \
const uint8_t *m, \
int m_stride) { \
return fnname(src_ptr, source_stride, ref_ptr, ref_stride, \
m, m_stride); \
} \
static unsigned int fnname##_bits10(const uint8_t *src_ptr, \
int source_stride, \
const uint8_t *ref_ptr, \
int ref_stride, \
const uint8_t *m, \
int m_stride) { \
return fnname(src_ptr, source_stride, ref_ptr, ref_stride, \
m, m_stride) >> 2; \
} \
static unsigned int fnname##_bits12(const uint8_t *src_ptr, \
int source_stride, \
const uint8_t *ref_ptr, \
int ref_stride, \
const uint8_t *m, \
int m_stride) { \
return fnname(src_ptr, source_stride, ref_ptr, ref_stride, \
m, m_stride) >> 4; \
}
MAKE_MBFP_SAD_WRAPPER(vp9_highbd_masked_sad64x64)
MAKE_MBFP_SAD_WRAPPER(vp9_highbd_masked_sad64x32)
MAKE_MBFP_SAD_WRAPPER(vp9_highbd_masked_sad32x64)
MAKE_MBFP_SAD_WRAPPER(vp9_highbd_masked_sad32x32)
MAKE_MBFP_SAD_WRAPPER(vp9_highbd_masked_sad32x16)
MAKE_MBFP_SAD_WRAPPER(vp9_highbd_masked_sad16x32)
MAKE_MBFP_SAD_WRAPPER(vp9_highbd_masked_sad16x16)
MAKE_MBFP_SAD_WRAPPER(vp9_highbd_masked_sad16x8)
MAKE_MBFP_SAD_WRAPPER(vp9_highbd_masked_sad8x16)
MAKE_MBFP_SAD_WRAPPER(vp9_highbd_masked_sad8x8)
MAKE_MBFP_SAD_WRAPPER(vp9_highbd_masked_sad8x4)
MAKE_MBFP_SAD_WRAPPER(vp9_highbd_masked_sad4x8)
MAKE_MBFP_SAD_WRAPPER(vp9_highbd_masked_sad4x4)
#endif // CONFIG_WEDGE_PARTITION
static void highbd_set_var_fns(VP9_COMP *const cpi) {
VP9_COMMON *const cm = &cpi->common;
if (cm->use_highbitdepth) {
@ -955,6 +1005,61 @@ static void highbd_set_var_fns(VP9_COMP *const cpi) {
vp9_highbd_sad4x4x3_bits8,
vp9_highbd_sad4x4x8_bits8,
vp9_highbd_sad4x4x4d_bits8)
#if CONFIG_WEDGE_PARTITION
HIGHBD_MBFP(BLOCK_64X64,
vp9_highbd_masked_sad64x64_bits8,
vp9_highbd_masked_variance64x64,
vp9_highbd_masked_sub_pixel_variance64x64)
HIGHBD_MBFP(BLOCK_64X32,
vp9_highbd_masked_sad64x32_bits8,
vp9_highbd_masked_variance64x32,
vp9_highbd_masked_sub_pixel_variance64x32)
HIGHBD_MBFP(BLOCK_32X64,
vp9_highbd_masked_sad32x64_bits8,
vp9_highbd_masked_variance32x64,
vp9_highbd_masked_sub_pixel_variance32x64)
HIGHBD_MBFP(BLOCK_32X32,
vp9_highbd_masked_sad32x32_bits8,
vp9_highbd_masked_variance32x32,
vp9_highbd_masked_sub_pixel_variance32x32)
HIGHBD_MBFP(BLOCK_32X16,
vp9_highbd_masked_sad32x16_bits8,
vp9_highbd_masked_variance32x16,
vp9_highbd_masked_sub_pixel_variance32x16)
HIGHBD_MBFP(BLOCK_16X32,
vp9_highbd_masked_sad16x32_bits8,
vp9_highbd_masked_variance16x32,
vp9_highbd_masked_sub_pixel_variance16x32)
HIGHBD_MBFP(BLOCK_16X16,
vp9_highbd_masked_sad16x16_bits8,
vp9_highbd_masked_variance16x16,
vp9_highbd_masked_sub_pixel_variance16x16)
HIGHBD_MBFP(BLOCK_8X16,
vp9_highbd_masked_sad8x16_bits8,
vp9_highbd_masked_variance8x16,
vp9_highbd_masked_sub_pixel_variance8x16)
HIGHBD_MBFP(BLOCK_16X8,
vp9_highbd_masked_sad16x8_bits8,
vp9_highbd_masked_variance16x8,
vp9_highbd_masked_sub_pixel_variance16x8)
HIGHBD_MBFP(BLOCK_8X8,
vp9_highbd_masked_sad8x8_bits8,
vp9_highbd_masked_variance8x8,
vp9_highbd_masked_sub_pixel_variance8x8)
HIGHBD_MBFP(BLOCK_4X8,
vp9_highbd_masked_sad4x8_bits8,
vp9_highbd_masked_variance4x8,
vp9_highbd_masked_sub_pixel_variance4x8)
HIGHBD_MBFP(BLOCK_8X4,
vp9_highbd_masked_sad8x4_bits8,
vp9_highbd_masked_variance8x4,
vp9_highbd_masked_sub_pixel_variance8x4)
HIGHBD_MBFP(BLOCK_4X4,
vp9_highbd_masked_sad4x4_bits8,
vp9_highbd_masked_variance4x4,
vp9_highbd_masked_sub_pixel_variance4x4)
#endif // CONFIG_WEDGE_PARTITION
break;
case VPX_BITS_10:
@ -1087,6 +1192,61 @@ static void highbd_set_var_fns(VP9_COMP *const cpi) {
vp9_highbd_sad4x4x3_bits10,
vp9_highbd_sad4x4x8_bits10,
vp9_highbd_sad4x4x4d_bits10)
#if CONFIG_WEDGE_PARTITION
HIGHBD_MBFP(BLOCK_64X64,
vp9_highbd_masked_sad64x64_bits10,
vp9_highbd_10_masked_variance64x64,
vp9_highbd_10_masked_sub_pixel_variance64x64)
HIGHBD_MBFP(BLOCK_64X32,
vp9_highbd_masked_sad64x32_bits10,
vp9_highbd_10_masked_variance64x32,
vp9_highbd_10_masked_sub_pixel_variance64x32)
HIGHBD_MBFP(BLOCK_32X64,
vp9_highbd_masked_sad32x64_bits10,
vp9_highbd_10_masked_variance32x64,
vp9_highbd_10_masked_sub_pixel_variance32x64)
HIGHBD_MBFP(BLOCK_32X32,
vp9_highbd_masked_sad32x32_bits10,
vp9_highbd_10_masked_variance32x32,
vp9_highbd_10_masked_sub_pixel_variance32x32)
HIGHBD_MBFP(BLOCK_32X16,
vp9_highbd_masked_sad32x16_bits10,
vp9_highbd_10_masked_variance32x16,
vp9_highbd_10_masked_sub_pixel_variance32x16)
HIGHBD_MBFP(BLOCK_16X32,
vp9_highbd_masked_sad16x32_bits10,
vp9_highbd_10_masked_variance16x32,
vp9_highbd_10_masked_sub_pixel_variance16x32)
HIGHBD_MBFP(BLOCK_16X16,
vp9_highbd_masked_sad16x16_bits10,
vp9_highbd_10_masked_variance16x16,
vp9_highbd_10_masked_sub_pixel_variance16x16)
HIGHBD_MBFP(BLOCK_8X16,
vp9_highbd_masked_sad8x16_bits10,
vp9_highbd_10_masked_variance8x16,
vp9_highbd_10_masked_sub_pixel_variance8x16)
HIGHBD_MBFP(BLOCK_16X8,
vp9_highbd_masked_sad16x8_bits10,
vp9_highbd_10_masked_variance16x8,
vp9_highbd_10_masked_sub_pixel_variance16x8)
HIGHBD_MBFP(BLOCK_8X8,
vp9_highbd_masked_sad8x8_bits10,
vp9_highbd_10_masked_variance8x8,
vp9_highbd_10_masked_sub_pixel_variance8x8)
HIGHBD_MBFP(BLOCK_4X8,
vp9_highbd_masked_sad4x8_bits10,
vp9_highbd_10_masked_variance4x8,
vp9_highbd_10_masked_sub_pixel_variance4x8)
HIGHBD_MBFP(BLOCK_8X4,
vp9_highbd_masked_sad8x4_bits10,
vp9_highbd_10_masked_variance8x4,
vp9_highbd_10_masked_sub_pixel_variance8x4)
HIGHBD_MBFP(BLOCK_4X4,
vp9_highbd_masked_sad4x4_bits10,
vp9_highbd_10_masked_variance4x4,
vp9_highbd_10_masked_sub_pixel_variance4x4)
#endif // CONFIG_WEDGE_PARTITION
break;
case VPX_BITS_12:
@ -1219,6 +1379,61 @@ static void highbd_set_var_fns(VP9_COMP *const cpi) {
vp9_highbd_sad4x4x3_bits12,
vp9_highbd_sad4x4x8_bits12,
vp9_highbd_sad4x4x4d_bits12)
#if CONFIG_WEDGE_PARTITION
HIGHBD_MBFP(BLOCK_64X64,
vp9_highbd_masked_sad64x64_bits12,
vp9_highbd_12_masked_variance64x64,
vp9_highbd_12_masked_sub_pixel_variance64x64)
HIGHBD_MBFP(BLOCK_64X32,
vp9_highbd_masked_sad64x32_bits12,
vp9_highbd_12_masked_variance64x32,
vp9_highbd_12_masked_sub_pixel_variance64x32)
HIGHBD_MBFP(BLOCK_32X64,
vp9_highbd_masked_sad32x64_bits12,
vp9_highbd_12_masked_variance32x64,
vp9_highbd_12_masked_sub_pixel_variance32x64)
HIGHBD_MBFP(BLOCK_32X32,
vp9_highbd_masked_sad32x32_bits12,
vp9_highbd_12_masked_variance32x32,
vp9_highbd_12_masked_sub_pixel_variance32x32)
HIGHBD_MBFP(BLOCK_32X16,
vp9_highbd_masked_sad32x16_bits12,
vp9_highbd_12_masked_variance32x16,
vp9_highbd_12_masked_sub_pixel_variance32x16)
HIGHBD_MBFP(BLOCK_16X32,
vp9_highbd_masked_sad16x32_bits12,
vp9_highbd_12_masked_variance16x32,
vp9_highbd_12_masked_sub_pixel_variance16x32)
HIGHBD_MBFP(BLOCK_16X16,
vp9_highbd_masked_sad16x16_bits12,
vp9_highbd_12_masked_variance16x16,
vp9_highbd_12_masked_sub_pixel_variance16x16)
HIGHBD_MBFP(BLOCK_8X16,
vp9_highbd_masked_sad8x16_bits12,
vp9_highbd_12_masked_variance8x16,
vp9_highbd_12_masked_sub_pixel_variance8x16)
HIGHBD_MBFP(BLOCK_16X8,
vp9_highbd_masked_sad16x8_bits12,
vp9_highbd_12_masked_variance16x8,
vp9_highbd_12_masked_sub_pixel_variance16x8)
HIGHBD_MBFP(BLOCK_8X8,
vp9_highbd_masked_sad8x8_bits12,
vp9_highbd_12_masked_variance8x8,
vp9_highbd_12_masked_sub_pixel_variance8x8)
HIGHBD_MBFP(BLOCK_4X8,
vp9_highbd_masked_sad4x8_bits12,
vp9_highbd_12_masked_variance4x8,
vp9_highbd_12_masked_sub_pixel_variance4x8)
HIGHBD_MBFP(BLOCK_8X4,
vp9_highbd_masked_sad8x4_bits12,
vp9_highbd_12_masked_variance8x4,
vp9_highbd_12_masked_sub_pixel_variance8x4)
HIGHBD_MBFP(BLOCK_4X4,
vp9_highbd_masked_sad4x4_bits12,
vp9_highbd_12_masked_variance4x4,
vp9_highbd_12_masked_sub_pixel_variance4x4)
#endif // CONFIG_WEDGE_PARTITION
break;
default:

49
vp9/encoder/vp9_sad.c
View File

@ -317,4 +317,53 @@ MASKSADMxN(8, 8)
MASKSADMxN(8, 4)
MASKSADMxN(4, 8)
MASKSADMxN(4, 4)
#if CONFIG_VP9_HIGHBITDEPTH
static INLINE unsigned int highbd_masked_sad(const uint8_t *a8, int a_stride,
const uint8_t *b8, int b_stride,
const uint8_t *m, int m_stride,
int width, int height) {
int y, x;
unsigned int sad = 0;
const uint16_t *a = CONVERT_TO_SHORTPTR(a8);
const uint16_t *b = CONVERT_TO_SHORTPTR(b8);
for (y = 0; y < height; y++) {
for (x = 0; x < width; x++)
sad += m[x] * abs(a[x] - b[x]);
a += a_stride;
b += b_stride;
m += m_stride;
}
sad = (sad + 31) >> 6;
return sad;
}
#define highbd_MASKSADMxN(m, n) \
unsigned int vp9_highbd_masked_sad##m##x##n##_c(const uint8_t *src, \
int src_stride, \
const uint8_t *ref, \
int ref_stride, \
const uint8_t *msk, \
int msk_stride) { \
return highbd_masked_sad(src, src_stride, ref, ref_stride, \
msk, msk_stride, m, n); \
}
highbd_MASKSADMxN(64, 64)
highbd_MASKSADMxN(64, 32)
highbd_MASKSADMxN(32, 64)
highbd_MASKSADMxN(32, 32)
highbd_MASKSADMxN(32, 16)
highbd_MASKSADMxN(16, 32)
highbd_MASKSADMxN(16, 16)
highbd_MASKSADMxN(16, 8)
highbd_MASKSADMxN(8, 16)
highbd_MASKSADMxN(8, 8)
highbd_MASKSADMxN(8, 4)
highbd_MASKSADMxN(4, 8)
highbd_MASKSADMxN(4, 4)
#endif // CONFIG_VP9_HIGHBITDEPTH
#endif // CONFIG_WEDGE_PARTITION

209
vp9/encoder/vp9_variance.c
View File

@ -743,4 +743,213 @@ MASK_SUBPIX_VAR(64, 32)
MASK_VAR(64, 64)
MASK_SUBPIX_VAR(64, 64)
#if CONFIG_VP9_HIGHBITDEPTH
void highbd_masked_variance64(const uint8_t *a8, int a_stride,
const uint8_t *b8, int b_stride,
const uint8_t *m, int m_stride,
int w, int h,
uint64_t *sse, int64_t *sum) {
int i, j;
uint16_t *a = CONVERT_TO_SHORTPTR(a8);
uint16_t *b = CONVERT_TO_SHORTPTR(b8);
*sum = 0;
*sse = 0;
for (i = 0; i < h; i++) {
for (j = 0; j < w; j++) {
const int diff = (a[j] - b[j]) * (m[j]);
*sum += diff;
*sse += diff * diff;
}
a += a_stride;
b += b_stride;
m += m_stride;
}
*sum = (*sum >= 0) ? ((*sum + 31) >> 6) : -((-*sum + 31) >> 6);
*sse = (*sse + 2047) >> 12;
}
void highbd_masked_variance(const uint8_t *a8, int a_stride,
const uint8_t *b8, int b_stride,
const uint8_t *m, int m_stride,
int w, int h,
unsigned int *sse, int *sum) {
uint64_t sse_long = 0;
int64_t sum_long = 0;
highbd_masked_variance64(a8, a_stride, b8, b_stride, m, m_stride,
w, h, &sse_long, &sum_long);
*sse = sse_long;
*sum = sum_long;
}
void highbd_10_masked_variance(const uint8_t *a8, int a_stride,
const uint8_t *b8, int b_stride,
const uint8_t *m, int m_stride,
int w, int h,
unsigned int *sse, int *sum) {
uint64_t sse_long = 0;
int64_t sum_long = 0;
highbd_masked_variance64(a8, a_stride, b8, b_stride, m, m_stride,
w, h, &sse_long, &sum_long);
*sum = ROUND_POWER_OF_TWO(sum_long, 2);
*sse = ROUND_POWER_OF_TWO(sse_long, 4);
}
void highbd_12_masked_variance(const uint8_t *a8, int a_stride,
const uint8_t *b8, int b_stride,
const uint8_t *m, int m_stride,
int w, int h,
unsigned int *sse, int *sum) {
uint64_t sse_long = 0;
int64_t sum_long = 0;
highbd_masked_variance64(a8, a_stride, b8, b_stride, m, m_stride,
w, h, &sse_long, &sum_long);
*sum = ROUND_POWER_OF_TWO(sum_long, 4);
*sse = ROUND_POWER_OF_TWO(sse_long, 8);
}
#define HIGHBD_MASK_VAR(W, H) \
unsigned int vp9_highbd_masked_variance##W##x##H##_c(const uint8_t *a, \
int a_stride, \
const uint8_t *b, \
int b_stride, \
const uint8_t *m, \
int m_stride, \
unsigned int *sse) { \
int sum; \
highbd_masked_variance(a, a_stride, b, b_stride, m, m_stride, \
W, H, sse, &sum); \
return *sse - (((int64_t)sum * sum) / (W * H)); \
} \
\
unsigned int vp9_highbd_10_masked_variance##W##x##H##_c(const uint8_t *a, \
int a_stride, \
const uint8_t *b, \
int b_stride, \
const uint8_t *m, \
int m_stride, \
unsigned int *sse) { \
int sum; \
highbd_10_masked_variance(a, a_stride, b, b_stride, m, m_stride, \
W, H, sse, &sum); \
return *sse - (((int64_t)sum * sum) / (W * H)); \
} \
\
unsigned int vp9_highbd_12_masked_variance##W##x##H##_c(const uint8_t *a, \
int a_stride, \
const uint8_t *b, \
int b_stride, \
const uint8_t *m, \
int m_stride, \
unsigned int *sse) { \
int sum; \
highbd_12_masked_variance(a, a_stride, b, b_stride, m, m_stride, \
W, H, sse, &sum); \
return *sse - (((int64_t)sum * sum) / (W * H)); \
}
#define HIGHBD_MASK_SUBPIX_VAR(W, H) \
unsigned int vp9_highbd_masked_sub_pixel_variance##W##x##H##_c( \
const uint8_t *src, int src_stride, \
int xoffset, int yoffset, \
const uint8_t *dst, int dst_stride, \
const uint8_t *msk, int msk_stride, \
unsigned int *sse) { \
uint16_t fdata3[(H + 1) * W]; \
uint16_t temp2[H * W]; \
\
highbd_var_filter_block2d_bil_first_pass(src, fdata3, src_stride, 1, \
H + 1, W, \
BILINEAR_FILTERS_2TAP(xoffset)); \
highbd_var_filter_block2d_bil_second_pass(fdata3, temp2, W, W, H, W, \
BILINEAR_FILTERS_2TAP(yoffset)); \
\
return vp9_highbd_masked_variance##W##x##H##_c(CONVERT_TO_BYTEPTR(temp2), \
W, dst, dst_stride, \
msk, msk_stride, sse); \
} \
\
unsigned int vp9_highbd_10_masked_sub_pixel_variance##W##x##H##_c( \
const uint8_t *src, int src_stride, \
int xoffset, int yoffset, \
const uint8_t *dst, int dst_stride, \
const uint8_t *msk, int msk_stride, \
unsigned int *sse) { \
uint16_t fdata3[(H + 1) * W]; \
uint16_t temp2[H * W]; \
\
highbd_var_filter_block2d_bil_first_pass(src, fdata3, src_stride, 1, \
H + 1, W, \
BILINEAR_FILTERS_2TAP(xoffset)); \
highbd_var_filter_block2d_bil_second_pass(fdata3, temp2, W, W, H, W, \
BILINEAR_FILTERS_2TAP(yoffset)); \
\
return vp9_highbd_10_masked_variance##W##x##H##_c(CONVERT_TO_BYTEPTR(temp2), \
W, dst, dst_stride, \
msk, msk_stride, sse); \
} \
\
unsigned int vp9_highbd_12_masked_sub_pixel_variance##W##x##H##_c( \
const uint8_t *src, int src_stride, \
int xoffset, int yoffset, \
const uint8_t *dst, int dst_stride, \
const uint8_t *msk, int msk_stride, \
unsigned int *sse) { \
uint16_t fdata3[(H + 1) * W]; \
uint16_t temp2[H * W]; \
\
highbd_var_filter_block2d_bil_first_pass(src, fdata3, src_stride, 1, \
H + 1, W, \
BILINEAR_FILTERS_2TAP(xoffset)); \
highbd_var_filter_block2d_bil_second_pass(fdata3, temp2, W, W, H, W, \
BILINEAR_FILTERS_2TAP(yoffset)); \
\
return vp9_highbd_12_masked_variance##W##x##H##_c(CONVERT_TO_BYTEPTR(temp2), \
W, dst, dst_stride, \
msk, msk_stride, sse); \
}
HIGHBD_MASK_VAR(4, 4)
HIGHBD_MASK_SUBPIX_VAR(4, 4)
HIGHBD_MASK_VAR(4, 8)
HIGHBD_MASK_SUBPIX_VAR(4, 8)
HIGHBD_MASK_VAR(8, 4)
HIGHBD_MASK_SUBPIX_VAR(8, 4)
HIGHBD_MASK_VAR(8, 8)
HIGHBD_MASK_SUBPIX_VAR(8, 8)
HIGHBD_MASK_VAR(8, 16)
HIGHBD_MASK_SUBPIX_VAR(8, 16)
HIGHBD_MASK_VAR(16, 8)
HIGHBD_MASK_SUBPIX_VAR(16, 8)
HIGHBD_MASK_VAR(16, 16)
HIGHBD_MASK_SUBPIX_VAR(16, 16)
HIGHBD_MASK_VAR(16, 32)
HIGHBD_MASK_SUBPIX_VAR(16, 32)
HIGHBD_MASK_VAR(32, 16)
HIGHBD_MASK_SUBPIX_VAR(32, 16)
HIGHBD_MASK_VAR(32, 32)
HIGHBD_MASK_SUBPIX_VAR(32, 32)
HIGHBD_MASK_VAR(32, 64)
HIGHBD_MASK_SUBPIX_VAR(32, 64)
HIGHBD_MASK_VAR(64, 32)
HIGHBD_MASK_SUBPIX_VAR(64, 32)
HIGHBD_MASK_VAR(64, 64)
HIGHBD_MASK_SUBPIX_VAR(64, 64)
#endif // CONFIG_VP9_HIGHBITDEPTH
#endif // CONFIG_WEDGE_PARTITION