Optimize and cleanup obmc predictor and rd search.
Use vpx_blend_a64_hmask and vpx_blend_a64_vmask to speed up
computing the obmc predictor. Clean up calc_target_weighted_pred.
Encoder speedup: 1.3%
Decoder speedup: 6.5%
Change-Id: I0c774fe53d22399e92a10d1daf3af0010d88d2c5
diff --git a/vp10/common/reconinter.c b/vp10/common/reconinter.c
index 53fd1a6..d2fc980 100644
--- a/vp10/common/reconinter.c
+++ b/vp10/common/reconinter.c
@@ -1298,97 +1298,63 @@
#endif // CONFIG_SUPERTX
#if CONFIG_OBMC
-// obmc_mask_N[is_neighbor_predictor][overlap_position]
-static const uint8_t obmc_mask_1[2][1] = {
- { 55},
- { 9}
+// obmc_mask_N[overlap_position]
+static const uint8_t obmc_mask_1[1] = {
+ 55
};
-static const uint8_t obmc_mask_2[2][2] = {
- { 45, 62},
- { 19, 2}
+static const uint8_t obmc_mask_2[2] = {
+ 45, 62
};
-static const uint8_t obmc_mask_4[2][4] = {
- { 39, 50, 59, 64},
- { 25, 14, 5, 0}
+static const uint8_t obmc_mask_4[4] = {
+ 39, 50, 59, 64
};
-static const uint8_t obmc_mask_8[2][8] = {
- { 36, 42, 48, 53, 57, 61, 63, 64},
- { 28, 22, 16, 11, 7, 3, 1, 0}
+static const uint8_t obmc_mask_8[8] = {
+ 36, 42, 48, 53, 57, 61, 63, 64
};
-static const uint8_t obmc_mask_16[2][16] = {
- { 34, 37, 40, 43, 46, 49, 52, 54, 56, 58, 60, 61, 63, 64, 64, 64},
- { 30, 27, 24, 21, 18, 15, 12, 10, 8, 6, 4, 3, 1, 0, 0, 0}
+static const uint8_t obmc_mask_16[16] = {
+ 34, 37, 40, 43, 46, 49, 52, 54, 56, 58, 60, 61, 63, 64, 64, 64
};
-static const uint8_t obmc_mask_32[2][32] = {
- { 33, 35, 36, 38, 40, 41, 43, 44,
- 45, 47, 48, 50, 51, 52, 53, 55,
- 56, 57, 58, 59, 60, 60, 61, 62,
- 62, 63, 63, 64, 64, 64, 64, 64 },
- { 31, 29, 28, 26, 24, 23, 21, 20,
- 19, 17, 16, 14, 13, 12, 11, 9,
- 8, 7, 6, 5, 4, 4, 3, 2,
- 2, 1, 1, 0, 0, 0, 0, 0 }
+static const uint8_t obmc_mask_32[32] = {
+ 33, 35, 36, 38, 40, 41, 43, 44, 45, 47, 48, 50, 51, 52, 53, 55,
+ 56, 57, 58, 59, 60, 60, 61, 62, 62, 63, 63, 64, 64, 64, 64, 64
};
#if CONFIG_EXT_PARTITION
-static const uint8_t obmc_mask_64[2][64] = {
- {
- 33, 34, 35, 35, 36, 37, 38, 39, 40, 40, 41, 42, 43, 44, 44, 44,
- 45, 46, 47, 47, 48, 49, 50, 51, 51, 51, 52, 52, 53, 54, 55, 56,
- 56, 56, 57, 57, 58, 58, 59, 60, 60, 60, 60, 60, 61, 62, 62, 62,
- 62, 62, 63, 63, 63, 63, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64,
- }, {
- 31, 30, 29, 29, 28, 27, 26, 25, 24, 24, 23, 22, 21, 20, 20, 20,
- 19, 18, 17, 17, 16, 15, 14, 13, 13, 13, 12, 12, 11, 10, 9, 8,
- 8, 8, 7, 7, 6, 6, 5, 4, 4, 4, 4, 4, 3, 2, 2, 2,
- 2, 2, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
- }
+static const uint8_t obmc_mask_64[64] = {
+ 33, 34, 35, 35, 36, 37, 38, 39, 40, 40, 41, 42, 43, 44, 44, 44,
+ 45, 46, 47, 47, 48, 49, 50, 51, 51, 51, 52, 52, 53, 54, 55, 56,
+ 56, 56, 57, 57, 58, 58, 59, 60, 60, 60, 60, 60, 61, 62, 62, 62,
+ 62, 62, 63, 63, 63, 63, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64,
};
#endif // CONFIG_EXT_PARTITION
-void setup_obmc_mask(int length, const uint8_t *mask[2]) {
+const uint8_t* vp10_get_obmc_mask(int length) {
switch (length) {
case 1:
- mask[0] = obmc_mask_1[0];
- mask[1] = obmc_mask_1[1];
- break;
+ return obmc_mask_1;
case 2:
- mask[0] = obmc_mask_2[0];
- mask[1] = obmc_mask_2[1];
- break;
+ return obmc_mask_2;
case 4:
- mask[0] = obmc_mask_4[0];
- mask[1] = obmc_mask_4[1];
- break;
+ return obmc_mask_4;
case 8:
- mask[0] = obmc_mask_8[0];
- mask[1] = obmc_mask_8[1];
- break;
+ return obmc_mask_8;
case 16:
- mask[0] = obmc_mask_16[0];
- mask[1] = obmc_mask_16[1];
- break;
+ return obmc_mask_16;
case 32:
- mask[0] = obmc_mask_32[0];
- mask[1] = obmc_mask_32[1];
- break;
+ return obmc_mask_32;
#if CONFIG_EXT_PARTITION
case 64:
- mask[0] = obmc_mask_64[0];
- mask[1] = obmc_mask_64[1];
- break;
+ return obmc_mask_64;
#endif // CONFIG_EXT_PARTITION
default:
- mask[0] = NULL;
- mask[1] = NULL;
assert(0);
- break;
+ return NULL;
}
}
@@ -1398,168 +1364,101 @@
// xd->plane[].dst.buf
void vp10_build_obmc_inter_prediction(VP10_COMMON *cm,
MACROBLOCKD *xd, int mi_row, int mi_col,
- int use_tmp_dst_buf,
- uint8_t *final_buf[MAX_MB_PLANE],
- int final_stride[MAX_MB_PLANE],
- uint8_t *tmp_buf1[MAX_MB_PLANE],
- int tmp_stride1[MAX_MB_PLANE],
- uint8_t *tmp_buf2[MAX_MB_PLANE],
- int tmp_stride2[MAX_MB_PLANE]) {
- const TileInfo *const tile = &xd->tile;
- BLOCK_SIZE bsize = xd->mi[0]->mbmi.sb_type;
- int plane, i, mi_step;
- int above_available = mi_row > tile->mi_row_start;
+ uint8_t *above[MAX_MB_PLANE],
+ int above_stride[MAX_MB_PLANE],
+ uint8_t *left[MAX_MB_PLANE],
+ int left_stride[MAX_MB_PLANE]) {
+ const BLOCK_SIZE bsize = xd->mi[0]->mbmi.sb_type;
+ int plane, i;
#if CONFIG_VP9_HIGHBITDEPTH
- int is_hbd = (xd->cur_buf->flags & YV12_FLAG_HIGHBITDEPTH) ? 1 : 0;
+ const int is_hbd = (xd->cur_buf->flags & YV12_FLAG_HIGHBITDEPTH) ? 1 : 0;
#endif // CONFIG_VP9_HIGHBITDEPTH
- if (use_tmp_dst_buf) {
- for (plane = 0; plane < MAX_MB_PLANE; ++plane) {
- const struct macroblockd_plane *pd = &xd->plane[plane];
- int bw = (xd->n8_w * 8) >> pd->subsampling_x;
- int bh = (xd->n8_h * 8) >> pd->subsampling_y;
- int row;
-#if CONFIG_VP9_HIGHBITDEPTH
- if (is_hbd) {
- uint16_t *final_buf16 = CONVERT_TO_SHORTPTR(final_buf[plane]);
- uint16_t *bmc_buf16 = CONVERT_TO_SHORTPTR(pd->dst.buf);
- for (row = 0; row < bh; ++row)
- memcpy(final_buf16 + row * final_stride[plane],
- bmc_buf16 + row * pd->dst.stride, bw * sizeof(uint16_t));
- } else {
-#endif
- for (row = 0; row < bh; ++row)
- memcpy(final_buf[plane] + row * final_stride[plane],
- pd->dst.buf + row * pd->dst.stride, bw);
-#if CONFIG_VP9_HIGHBITDEPTH
- }
-#endif // CONFIG_VP9_HIGHBITDEPTH
- }
- }
-
// handle above row
- for (i = 0; above_available && i < VPXMIN(xd->n8_w, cm->mi_cols - mi_col);
- i += mi_step) {
- int mi_row_offset = -1;
- int mi_col_offset = i;
- int overlap;
- MODE_INFO *above_mi = xd->mi[mi_col_offset +
- mi_row_offset * xd->mi_stride];
- MB_MODE_INFO *above_mbmi = &above_mi->mbmi;
+ if (xd->up_available) {
+ const int overlap = num_4x4_blocks_high_lookup[bsize] * 2;
+ const int miw = VPXMIN(xd->n8_w, cm->mi_cols - mi_col);
+ const int mi_row_offset = -1;
- mi_step = VPXMIN(xd->n8_w,
- num_8x8_blocks_wide_lookup[above_mbmi->sb_type]);
+ assert(miw > 0);
- if (!is_neighbor_overlappable(above_mbmi))
- continue;
+ i = 0;
+ do { // for each mi in the above row
+ const int mi_col_offset = i;
+ const MB_MODE_INFO *const above_mbmi =
+ &xd->mi[mi_col_offset + mi_row_offset * xd->mi_stride]->mbmi;
+ const int mi_step =
+ VPXMIN(xd->n8_w, num_8x8_blocks_wide_lookup[above_mbmi->sb_type]);
- overlap = num_4x4_blocks_high_lookup[bsize] << 1;
-
- for (plane = 0; plane < MAX_MB_PLANE; ++plane) {
- const struct macroblockd_plane *pd = &xd->plane[plane];
- int bw = (mi_step * MI_SIZE) >> pd->subsampling_x;
- int bh = overlap >> pd->subsampling_y;
- int row, col;
- int dst_stride = use_tmp_dst_buf ? final_stride[plane] : pd->dst.stride;
- uint8_t *dst = use_tmp_dst_buf ?
- &final_buf[plane][(i * MI_SIZE) >> pd->subsampling_x] :
- &pd->dst.buf[(i * MI_SIZE) >> pd->subsampling_x];
- int tmp_stride = tmp_stride1[plane];
- uint8_t *tmp = &tmp_buf1[plane][(i * MI_SIZE) >> pd->subsampling_x];
- const uint8_t *mask[2];
-
- setup_obmc_mask(bh, mask);
+ if (is_neighbor_overlappable(above_mbmi)) {
+ for (plane = 0; plane < MAX_MB_PLANE; ++plane) {
+ const struct macroblockd_plane *pd = &xd->plane[plane];
+ const int bw = (mi_step * MI_SIZE) >> pd->subsampling_x;
+ const int bh = overlap >> pd->subsampling_y;
+ const int dst_stride = pd->dst.stride;
+ uint8_t *const dst =
+ &pd->dst.buf[(i * MI_SIZE) >> pd->subsampling_x];
+ const int tmp_stride = above_stride[plane];
+ const uint8_t *const tmp =
+ &above[plane][(i * MI_SIZE) >> pd->subsampling_x];
+ const uint8_t *const mask = vp10_get_obmc_mask(bh);
#if CONFIG_VP9_HIGHBITDEPTH
- if (is_hbd) {
- uint16_t *dst16 = CONVERT_TO_SHORTPTR(dst);
- uint16_t *tmp16 = CONVERT_TO_SHORTPTR(tmp);
-
- for (row = 0; row < bh; ++row) {
- for (col = 0; col < bw; ++col)
- dst16[col] = ROUND_POWER_OF_TWO(mask[0][row] * dst16[col] +
- mask[1][row] * tmp16[col], 6);
-
- dst16 += dst_stride;
- tmp16 += tmp_stride;
+ if (is_hbd)
+ vpx_highbd_blend_a64_vmask(dst, dst_stride, dst, dst_stride,
+ tmp, tmp_stride, mask, bh, bw, xd->bd);
+ else
+#endif // CONFIG_VP9_HIGHBITDEPTH
+ vpx_blend_a64_vmask(dst, dst_stride, dst, dst_stride,
+ tmp, tmp_stride, mask, bh, bw);
}
- } else {
-#endif // CONFIG_VP9_HIGHBITDEPTH
- for (row = 0; row < bh; ++row) {
- for (col = 0; col < bw; ++col)
- dst[col] = ROUND_POWER_OF_TWO(mask[0][row] * dst[col] +
- mask[1][row] * tmp[col], 6);
- dst += dst_stride;
- tmp += tmp_stride;
}
-#if CONFIG_VP9_HIGHBITDEPTH
- }
-#endif // CONFIG_VP9_HIGHBITDEPTH
- }
- } // each mi in the above row
+ i += mi_step;
+ } while (i < miw);
+ }
- if (mi_col == 0 || (mi_col - 1 < tile->mi_col_start))
- return;
// handle left column
- for (i = 0; i < VPXMIN(xd->n8_h, cm->mi_rows - mi_row);
- i += mi_step) {
- int mi_row_offset = i;
- int mi_col_offset = -1;
- int overlap;
- MODE_INFO *left_mi = xd->mi[mi_col_offset +
- mi_row_offset * xd->mi_stride];
- MB_MODE_INFO *left_mbmi = &left_mi->mbmi;
+ if (xd->left_available) {
+ const int overlap = num_4x4_blocks_wide_lookup[bsize] * 2;
+ const int mih = VPXMIN(xd->n8_h, cm->mi_rows - mi_row);
+ const int mi_col_offset = -1;
- mi_step = VPXMIN(xd->n8_h,
- num_8x8_blocks_high_lookup[left_mbmi->sb_type]);
+ assert(mih > 0);
- if (!is_neighbor_overlappable(left_mbmi))
- continue;
+ i = 0;
+ do { // for each mi in the left column
+ const int mi_row_offset = i;
+ const MB_MODE_INFO *const left_mbmi =
+ &xd->mi[mi_col_offset + mi_row_offset * xd->mi_stride]->mbmi;
+ const int mi_step =
+ VPXMIN(xd->n8_h, num_8x8_blocks_high_lookup[left_mbmi->sb_type]);
- overlap = num_4x4_blocks_wide_lookup[bsize] << 1;
-
- for (plane = 0; plane < MAX_MB_PLANE; ++plane) {
- const struct macroblockd_plane *pd = &xd->plane[plane];
- int bw = overlap >> pd->subsampling_x;
- int bh = (mi_step * MI_SIZE) >> pd->subsampling_y;
- int row, col;
- int dst_stride = use_tmp_dst_buf ? final_stride[plane] : pd->dst.stride;
- uint8_t *dst = use_tmp_dst_buf ?
- &final_buf[plane][(i * MI_SIZE * dst_stride) >> pd->subsampling_y] :
- &pd->dst.buf[(i * MI_SIZE * dst_stride) >> pd->subsampling_y];
- int tmp_stride = tmp_stride2[plane];
- uint8_t *tmp = &tmp_buf2[plane]
- [(i * MI_SIZE * tmp_stride) >> pd->subsampling_y];
- const uint8_t *mask[2];
-
- setup_obmc_mask(bw, mask);
+ if (is_neighbor_overlappable(left_mbmi)) {
+ for (plane = 0; plane < MAX_MB_PLANE; ++plane) {
+ const struct macroblockd_plane *pd = &xd->plane[plane];
+ const int bw = overlap >> pd->subsampling_x;
+ const int bh = (mi_step * MI_SIZE) >> pd->subsampling_y;
+ const int dst_stride = pd->dst.stride;
+ uint8_t *const dst =
+ &pd->dst.buf[(i * MI_SIZE * dst_stride) >> pd->subsampling_y];
+ const int tmp_stride = left_stride[plane];
+ const uint8_t *const tmp =
+ &left[plane][(i * MI_SIZE * tmp_stride) >> pd->subsampling_y];
+ const uint8_t *const mask = vp10_get_obmc_mask(bw);
#if CONFIG_VP9_HIGHBITDEPTH
- if (is_hbd) {
- uint16_t *dst16 = CONVERT_TO_SHORTPTR(dst);
- uint16_t *tmp16 = CONVERT_TO_SHORTPTR(tmp);
-
- for (row = 0; row < bh; ++row) {
- for (col = 0; col < bw; ++col)
- dst16[col] = ROUND_POWER_OF_TWO(mask[0][col] * dst16[col] +
- mask[1][col] * tmp16[col], 6);
- dst16 += dst_stride;
- tmp16 += tmp_stride;
+ if (is_hbd)
+ vpx_highbd_blend_a64_hmask(dst, dst_stride, dst, dst_stride,
+ tmp, tmp_stride, mask, bh, bw, xd->bd);
+ else
+#endif // CONFIG_VP9_HIGHBITDEPTH
+ vpx_blend_a64_hmask(dst, dst_stride, dst, dst_stride,
+ tmp, tmp_stride, mask, bh, bw);
}
- } else {
-#endif // CONFIG_VP9_HIGHBITDEPTH
- for (row = 0; row < bh; ++row) {
- for (col = 0; col < bw; ++col)
- dst[col] = ROUND_POWER_OF_TWO(mask[0][col] * dst[col] +
- mask[1][col] * tmp[col], 6);
- dst += dst_stride;
- tmp += tmp_stride;
}
-#if CONFIG_VP9_HIGHBITDEPTH
- }
-#endif // CONFIG_VP9_HIGHBITDEPTH
- }
- } // each mi in the left column
+ i += mi_step;
+ } while (i < mih);
+ }
}
#if CONFIG_EXT_INTER
diff --git a/vp10/common/reconinter.h b/vp10/common/reconinter.h
index ac4a004..c32596e 100644
--- a/vp10/common/reconinter.h
+++ b/vp10/common/reconinter.h
@@ -562,16 +562,13 @@
#endif // CONFIG_EXT_INTERP
#if CONFIG_OBMC
-void setup_obmc_mask(int length, const uint8_t *mask[2]);
+const uint8_t* vp10_get_obmc_mask(int length);
void vp10_build_obmc_inter_prediction(VP10_COMMON *cm,
MACROBLOCKD *xd, int mi_row, int mi_col,
- int use_tmp_dst_buf,
- uint8_t *final_buf[MAX_MB_PLANE],
- int final_stride[MAX_MB_PLANE],
- uint8_t *tmp_buf1[MAX_MB_PLANE],
- int tmp_stride1[MAX_MB_PLANE],
- uint8_t *tmp_buf2[MAX_MB_PLANE],
- int tmp_stride2[MAX_MB_PLANE]);
+ uint8_t *above[MAX_MB_PLANE],
+ int above_stride[MAX_MB_PLANE],
+ uint8_t *left[MAX_MB_PLANE],
+ int left_stride[MAX_MB_PLANE]);
void vp10_build_prediction_by_above_preds(VP10_COMMON *cm,
MACROBLOCKD *xd,
int mi_row, int mi_col,
diff --git a/vp10/decoder/decodeframe.c b/vp10/decoder/decodeframe.c
index c4487a9..2f341b5 100644
--- a/vp10/decoder/decodeframe.c
+++ b/vp10/decoder/decodeframe.c
@@ -1385,7 +1385,7 @@
dst_buf2, dst_stride2);
vp10_setup_dst_planes(xd->plane, get_frame_new_buffer(cm),
mi_row, mi_col);
- vp10_build_obmc_inter_prediction(cm, xd, mi_row, mi_col, 0, NULL, NULL,
+ vp10_build_obmc_inter_prediction(cm, xd, mi_row, mi_col,
dst_buf1, dst_stride1,
dst_buf2, dst_stride2);
}
diff --git a/vp10/encoder/encodeframe.c b/vp10/encoder/encodeframe.c
index 69f766a..ff1ee6b 100644
--- a/vp10/encoder/encodeframe.c
+++ b/vp10/encoder/encodeframe.c
@@ -5091,7 +5091,7 @@
dst_stride2);
vp10_setup_dst_planes(xd->plane, get_frame_new_buffer(cm),
mi_row, mi_col);
- vp10_build_obmc_inter_prediction(cm, xd, mi_row, mi_col, 0, NULL, NULL,
+ vp10_build_obmc_inter_prediction(cm, xd, mi_row, mi_col,
dst_buf1, dst_stride1,
dst_buf2, dst_stride2);
}
diff --git a/vp10/encoder/rdopt.c b/vp10/encoder/rdopt.c
index b96e6e4..0ebdfd7 100644
--- a/vp10/encoder/rdopt.c
+++ b/vp10/encoder/rdopt.c
@@ -15,6 +15,7 @@
#include "./vpx_dsp_rtcd.h"
#include "vpx_dsp/vpx_dsp_common.h"
+#include "vpx_dsp/blend.h"
#include "vpx_mem/vpx_mem.h"
#include "vpx_ports/mem.h"
#include "vpx_ports/system_state.h"
@@ -7780,8 +7781,7 @@
vp10_build_inter_predictors_sb(xd, mi_row, mi_col, bsize);
#endif // CONFIG_EXT_INTER
}
- vp10_build_obmc_inter_prediction(cm, xd, mi_row, mi_col, 0,
- NULL, NULL,
+ vp10_build_obmc_inter_prediction(cm, xd, mi_row, mi_col,
dst_buf1, dst_stride1,
dst_buf2, dst_stride2);
model_rd_for_sb(cpi, bsize, x, xd, 0, MAX_MB_PLANE - 1,
@@ -8399,6 +8399,18 @@
}
#endif // CONFIG_EXT_INTRA
+#if CONFIG_OBMC
+static void calc_target_weighted_pred(
+ const VP10_COMMON *cm,
+ const MACROBLOCK *x,
+ const MACROBLOCKD *xd,
+ int mi_row, int mi_col,
+ const uint8_t *above, int above_stride,
+ const uint8_t *left, int left_stride,
+ int32_t *mask_buf,
+ int32_t *wsrc_buf);
+#endif // CONFIG_OBMC
+
void vp10_rd_pick_inter_mode_sb(VP10_COMP *cpi,
TileDataEnc *tile_data,
MACROBLOCK *x,
@@ -9579,7 +9591,7 @@
vp10_build_inter_predictors_sb(xd, mi_row, mi_col, bsize);
#if CONFIG_OBMC
if (mbmi->motion_variation == OBMC_CAUSAL)
- vp10_build_obmc_inter_prediction(cm, xd, mi_row, mi_col, 0, NULL, NULL,
+ vp10_build_obmc_inter_prediction(cm, xd, mi_row, mi_col,
dst_buf1, dst_stride1,
dst_buf2, dst_stride2);
#endif // CONFIG_OBMC
@@ -10980,189 +10992,225 @@
}
#if CONFIG_OBMC
-void calc_target_weighted_pred(VP10_COMMON *cm,
- MACROBLOCK *x,
- MACROBLOCKD *xd,
- int mi_row, int mi_col,
- uint8_t *above_buf, int above_stride,
- uint8_t *left_buf, int left_stride,
- int32_t *mask_buf,
- int32_t *weighted_src_buf) {
- BLOCK_SIZE bsize = xd->mi[0]->mbmi.sb_type;
- int row, col, i, mi_step;
- int bw = 8 * xd->n8_w;
- int bh = 8 * xd->n8_h;
+// This function has a structure similar to vp10_build_obmc_inter_prediction
+//
+// The OBMC predictor is computed as:
+//
+// PObmc(x,y) =
+// VPX_BLEND_A64(Mh(x),
+// VPX_BLEND_A64(Mv(y), P(x,y), PAbove(x,y)),
+// PLeft(x, y))
+//
+// Scaling up by VPX_BLEND_A64_MAX_ALPHA ** 2 and omitting the intermediate
+// rounding, this can be written as:
+//
+// VPX_BLEND_A64_MAX_ALPHA * VPX_BLEND_A64_MAX_ALPHA * Pobmc(x,y) =
+// Mh(x) * Mv(y) * P(x,y) +
+// Mh(x) * Cv(y) * Pabove(x,y) +
+// VPX_BLEND_A64_MAX_ALPHA * Ch(x) * PLeft(x, y)
+//
+// Where :
+//
+// Cv(y) = VPX_BLEND_A64_MAX_ALPHA - Mv(y)
+// Ch(y) = VPX_BLEND_A64_MAX_ALPHA - Mh(y)
+//
+// This function computes 'wsrc' and 'mask' as:
+//
+// wsrc(x, y) =
+// VPX_BLEND_A64_MAX_ALPHA * VPX_BLEND_A64_MAX_ALPHA * src(x, y) -
+// Mh(x) * Cv(y) * Pabove(x,y) +
+// VPX_BLEND_A64_MAX_ALPHA * Ch(x) * PLeft(x, y)
+//
+// mask(x, y) = Mh(x) * Mv(y)
+//
+// These can then be used to efficiently approximate the error for any
+// predictor P in the context of the provided neighbouring predictors by
+// computing:
+//
+// error(x, y) =
+// wsrc(x, y) - mask(x, y) * P(x, y) / (VPX_BLEND_A64_MAX_ALPHA ** 2)
+//
+static void calc_target_weighted_pred(
+ const VP10_COMMON *cm,
+ const MACROBLOCK *x,
+ const MACROBLOCKD *xd,
+ int mi_row, int mi_col,
+ const uint8_t *above, int above_stride,
+ const uint8_t *left, int left_stride,
+ int32_t *mask_buf,
+ int32_t *wsrc_buf) {
+ const BLOCK_SIZE bsize = xd->mi[0]->mbmi.sb_type;
+ int row, col, i;
+ const int bw = 8 * xd->n8_w;
+ const int bh = 8 * xd->n8_h;
+ const int wsrc_stride = bw;
const int mask_stride = bw;
- const int weighted_src_stride = bw;
- int32_t *dst = weighted_src_buf;
- int32_t *mask2d = mask_buf;
- uint8_t *src;
+ const int src_scale = VPX_BLEND_A64_MAX_ALPHA * VPX_BLEND_A64_MAX_ALPHA;
#if CONFIG_VP9_HIGHBITDEPTH
- int is_hbd = (xd->cur_buf->flags & YV12_FLAG_HIGHBITDEPTH) ? 1 : 0;
+ const int is_hbd = (xd->cur_buf->flags & YV12_FLAG_HIGHBITDEPTH) ? 1 : 0;
+#else
+ const int is_hbd = 0;
#endif // CONFIG_VP9_HIGHBITDEPTH
- for (row = 0; row < bh; ++row) {
- for (col = 0; col < bw; ++col) {
- dst[col] = 0;
- mask2d[col] = 64;
- }
- dst += weighted_src_stride;
- mask2d += mask_stride;
- }
+
+ // plane 0 should not be subsampled
+ assert(xd->plane[0].subsampling_x == 0);
+ assert(xd->plane[0].subsampling_y == 0);
+
+ vp10_zero_array(wsrc_buf, bw * bh);
+ for (i = 0; i < bw * bh; ++i)
+ mask_buf[i] = VPX_BLEND_A64_MAX_ALPHA;
// handle above row
if (xd->up_available) {
- for (i = 0; i < VPXMIN(xd->n8_w, cm->mi_cols - mi_col); i += mi_step) {
- int mi_row_offset = -1;
- int mi_col_offset = i;
- MODE_INFO *above_mi = xd->mi[mi_col_offset +
- mi_row_offset * xd->mi_stride];
- MB_MODE_INFO *above_mbmi = &above_mi->mbmi;
- int overlap = num_4x4_blocks_high_lookup[bsize] << 1;
+ const int overlap = num_4x4_blocks_high_lookup[bsize] * 2;
+ const int miw = VPXMIN(xd->n8_w, cm->mi_cols - mi_col);
+ const int mi_row_offset = -1;
+ const uint8_t *const mask1d = vp10_get_obmc_mask(overlap);
- mi_step = VPXMIN(xd->n8_w,
- num_8x8_blocks_wide_lookup[above_mbmi->sb_type]);
+ assert(miw > 0);
+
+ i = 0;
+ do { // for each mi in the above row
+ const int mi_col_offset = i;
+ const MB_MODE_INFO *const above_mbmi =
+ &xd->mi[mi_col_offset + mi_row_offset * xd->mi_stride]->mbmi;
+ const int mi_step =
+ VPXMIN(xd->n8_w, num_8x8_blocks_wide_lookup[above_mbmi->sb_type]);
+ const int neighbor_bw = mi_step * MI_SIZE;
if (is_neighbor_overlappable(above_mbmi)) {
- const struct macroblockd_plane *pd = &xd->plane[0];
- int bw = (mi_step * MI_SIZE) >> pd->subsampling_x;
- int bh = overlap >> pd->subsampling_y;
- int dst_stride = weighted_src_stride;
- int32_t *dst = weighted_src_buf + (i * MI_SIZE >> pd->subsampling_x);
- int tmp_stride = above_stride;
- uint8_t *tmp = above_buf + (i * MI_SIZE >> pd->subsampling_x);
- int mask2d_stride = mask_stride;
- int32_t *mask2d = mask_buf + (i * MI_SIZE >> pd->subsampling_x);
- const uint8_t *mask1d[2];
+ const int tmp_stride = above_stride;
+ int32_t *wsrc = wsrc_buf + (i * MI_SIZE);
+ int32_t *mask = mask_buf + (i * MI_SIZE);
- setup_obmc_mask(bh, mask1d);
+ if (!is_hbd) {
+ const uint8_t *tmp = above;
-#if CONFIG_VP9_HIGHBITDEPTH
- if (is_hbd) {
- uint16_t *tmp16 = CONVERT_TO_SHORTPTR(tmp);
-
- for (row = 0; row < bh; ++row) {
- for (col = 0; col < bw; ++col) {
- dst[col] = mask1d[1][row] * tmp16[col];
- mask2d[col] = mask1d[0][row];
+ for (row = 0; row < overlap; ++row) {
+ const uint8_t m0 = mask1d[row];
+ const uint8_t m1 = VPX_BLEND_A64_MAX_ALPHA - m0;
+ for (col = 0; col < neighbor_bw; ++col) {
+ wsrc[col] = m1 * tmp[col];
+ mask[col] = m0;
}
- dst += dst_stride;
- tmp16 += tmp_stride;
- mask2d += mask2d_stride;
+ wsrc += wsrc_stride;
+ mask += mask_stride;
+ tmp += tmp_stride;
}
- } else {
-#endif // CONFIG_VP9_HIGHBITDEPTH
- for (row = 0; row < bh; ++row) {
- for (col = 0; col < bw; ++col) {
- dst[col] = mask1d[1][row] * tmp[col];
- mask2d[col] = mask1d[0][row];
- }
- dst += dst_stride;
- tmp += tmp_stride;
- mask2d += mask2d_stride;
- }
#if CONFIG_VP9_HIGHBITDEPTH
- }
+ } else {
+ const uint16_t *tmp = CONVERT_TO_SHORTPTR(above);
+
+ for (row = 0; row < overlap; ++row) {
+ const uint8_t m0 = mask1d[row];
+ const uint8_t m1 = VPX_BLEND_A64_MAX_ALPHA - m0;
+ for (col = 0; col < neighbor_bw; ++col) {
+ wsrc[col] = m1 * tmp[col];
+ mask[col] = m0;
+ }
+ wsrc += wsrc_stride;
+ mask += mask_stride;
+ tmp += tmp_stride;
+ }
#endif // CONFIG_VP9_HIGHBITDEPTH
+ }
}
- } // each mi in the above row
+
+ above += neighbor_bw;
+ i += mi_step;
+ } while (i < miw);
+ }
+
+ for (i = 0; i < bw * bh; ++i) {
+ wsrc_buf[i] *= VPX_BLEND_A64_MAX_ALPHA;
+ mask_buf[i] *= VPX_BLEND_A64_MAX_ALPHA;
}
// handle left column
- dst = weighted_src_buf;
- mask2d = mask_buf;
- for (row = 0; row < bh; ++row) {
- for (col = 0; col < bw; ++col) {
- dst[col] = dst[col] << 6;
- mask2d[col] = mask2d[col] << 6;
- }
- dst += weighted_src_stride;
- mask2d += mask_stride;
- }
-
if (xd->left_available) {
- for (i = 0; i < VPXMIN(xd->n8_h, cm->mi_rows - mi_row); i += mi_step) {
- int mi_row_offset = i;
- int mi_col_offset = -1;
- int overlap = num_4x4_blocks_wide_lookup[bsize] << 1;
- MODE_INFO *left_mi = xd->mi[mi_col_offset +
- mi_row_offset * xd->mi_stride];
- MB_MODE_INFO *left_mbmi = &left_mi->mbmi;
+ const int overlap = num_4x4_blocks_wide_lookup[bsize] * 2;
+ const int mih = VPXMIN(xd->n8_h, cm->mi_rows - mi_row);
+ const int mi_col_offset = -1;
+ const uint8_t *const mask1d = vp10_get_obmc_mask(overlap);
- mi_step = VPXMIN(xd->n8_h,
- num_8x8_blocks_high_lookup[left_mbmi->sb_type]);
+ assert(mih > 0);
+
+ i = 0;
+ do { // for each mi in the left column
+ const int mi_row_offset = i;
+ const MB_MODE_INFO *const left_mbmi =
+ &xd->mi[mi_col_offset + mi_row_offset * xd->mi_stride]->mbmi;
+ const int mi_step =
+ VPXMIN(xd->n8_h, num_8x8_blocks_high_lookup[left_mbmi->sb_type]);
+ const int neighbor_bh = mi_step * MI_SIZE;
if (is_neighbor_overlappable(left_mbmi)) {
- const struct macroblockd_plane *pd = &xd->plane[0];
- int bw = overlap >> pd->subsampling_x;
- int bh = (mi_step * MI_SIZE) >> pd->subsampling_y;
- int dst_stride = weighted_src_stride;
- int32_t *dst = weighted_src_buf +
- (i * MI_SIZE * dst_stride >> pd->subsampling_y);
- int tmp_stride = left_stride;
- uint8_t *tmp = left_buf +
- (i * MI_SIZE * tmp_stride >> pd->subsampling_y);
- int mask2d_stride = mask_stride;
- int32_t *mask2d = mask_buf +
- (i * MI_SIZE * mask2d_stride >> pd->subsampling_y);
- const uint8_t *mask1d[2];
+ const int tmp_stride = left_stride;
+ int32_t *wsrc = wsrc_buf + (i * MI_SIZE * wsrc_stride);
+ int32_t *mask = mask_buf + (i * MI_SIZE * mask_stride);
- setup_obmc_mask(bw, mask1d);
+ if (!is_hbd) {
+ const uint8_t *tmp = left;
-#if CONFIG_VP9_HIGHBITDEPTH
- if (is_hbd) {
- uint16_t *tmp16 = CONVERT_TO_SHORTPTR(tmp);
-
- for (row = 0; row < bh; ++row) {
- for (col = 0; col < bw; ++col) {
- dst[col] = (dst[col] >> 6) * mask1d[0][col] +
- (tmp16[col] << 6) * mask1d[1][col];
- mask2d[col] = (mask2d[col] >> 6) * mask1d[0][col];
+ for (row = 0; row < neighbor_bh; ++row) {
+ for (col = 0; col < overlap; ++col) {
+ const uint8_t m0 = mask1d[col];
+ const uint8_t m1 = VPX_BLEND_A64_MAX_ALPHA - m0;
+ wsrc[col] = (wsrc[col] >> VPX_BLEND_A64_ROUND_BITS) * m0 +
+ (tmp[col] << VPX_BLEND_A64_ROUND_BITS) * m1;
+ mask[col] = (mask[col] >> VPX_BLEND_A64_ROUND_BITS) * m0;
}
- dst += dst_stride;
- tmp16 += tmp_stride;
- mask2d += mask2d_stride;
+ wsrc += wsrc_stride;
+ mask += mask_stride;
+ tmp += tmp_stride;
}
- } else {
-#endif // CONFIG_VP9_HIGHBITDEPTH
- for (row = 0; row < bh; ++row) {
- for (col = 0; col < bw; ++col) {
- dst[col] = (dst[col] >> 6) * mask1d[0][col] +
- (tmp[col] << 6) * mask1d[1][col];
- mask2d[col] = (mask2d[col] >> 6) * mask1d[0][col];
- }
- dst += dst_stride;
- tmp += tmp_stride;
- mask2d += mask2d_stride;
- }
#if CONFIG_VP9_HIGHBITDEPTH
+ } else {
+ const uint16_t *tmp = CONVERT_TO_SHORTPTR(left);
+
+ for (row = 0; row < neighbor_bh; ++row) {
+ for (col = 0; col < overlap; ++col) {
+ const uint8_t m0 = mask1d[col];
+ const uint8_t m1 = VPX_BLEND_A64_MAX_ALPHA - m0;
+ wsrc[col] = (wsrc[col] >> VPX_BLEND_A64_ROUND_BITS) * m0 +
+ (tmp[col] << VPX_BLEND_A64_ROUND_BITS) * m1;
+ mask[col] = (mask[col] >> VPX_BLEND_A64_ROUND_BITS) * m0;
+ }
+ wsrc += wsrc_stride;
+ mask += mask_stride;
+ tmp += tmp_stride;
+ }
+#endif // CONFIG_VP9_HIGHBITDEPTH
}
-#endif // CONFIG_VP9_HIGHBITDEPTH
}
- } // each mi in the left column
+
+ left += neighbor_bh * left_stride;
+ i += mi_step;
+ } while (i < mih);
}
- dst = weighted_src_buf;
- src = x->plane[0].src.buf;
-#if CONFIG_VP9_HIGHBITDEPTH
- if (is_hbd) {
- uint16_t *src16 = CONVERT_TO_SHORTPTR(src);
+ if (!is_hbd) {
+ const uint8_t *src = x->plane[0].src.buf;
for (row = 0; row < bh; ++row) {
- for (col = 0; col < bw; ++col)
- dst[col] = (src16[col] << 12) - dst[col];
- dst += weighted_src_stride;
- src16 += x->plane[0].src.stride;
+ for (col = 0; col < bw; ++col) {
+ wsrc_buf[col] = src[col] * src_scale - wsrc_buf[col];
+ }
+ wsrc_buf += wsrc_stride;
+ src += x->plane[0].src.stride;
}
- } else {
-#endif // CONFIG_VP9_HIGHBITDEPTH
- for (row = 0; row < bh; ++row) {
- for (col = 0; col < bw; ++col)
- dst[col] = (src[col] << 12) - dst[col];
- dst += weighted_src_stride;
- src += x->plane[0].src.stride;
- }
#if CONFIG_VP9_HIGHBITDEPTH
- }
+ } else {
+ const uint16_t *src = CONVERT_TO_SHORTPTR(x->plane[0].src.buf);
+
+ for (row = 0; row < bh; ++row) {
+ for (col = 0; col < bw; ++col) {
+ wsrc_buf[col] = src[col] * src_scale - wsrc_buf[col];
+ }
+ wsrc_buf += wsrc_stride;
+ src += x->plane[0].src.stride;
+ }
#endif // CONFIG_VP9_HIGHBITDEPTH
+ }
}
#endif // CONFIG_OBMC
diff --git a/vp10/encoder/rdopt.h b/vp10/encoder/rdopt.h
index b660e23..da70a22 100644
--- a/vp10/encoder/rdopt.h
+++ b/vp10/encoder/rdopt.h
@@ -90,16 +90,6 @@
int use_fast_coef_casting);
#endif // CONFIG_SUPERTX
-#if CONFIG_OBMC
-void calc_target_weighted_pred(VP10_COMMON *cm,
- MACROBLOCK *x,
- MACROBLOCKD *xd,
- int mi_row, int mi_col,
- uint8_t *above_buf, int above_stride,
- uint8_t *left_buf, int left_stride,
- int32_t *mask_buf, int32_t *weighted_src_buf);
-#endif // CONFIG_OBMC
-
#ifdef __cplusplus
} // extern "C"
#endif
