JNT_COMP: add SIMD and interface for high bit-depth
Add high bit-depth macro definitions:
highbd_jnt_sad
highbd_8(10/12)_jnt_sub_pixel_avg.
Add SIMD functions:
aom_highbd_jnt_comp_avg_pred_sse2
aom_highbd_jnt_comp_avg_upsampled_pred_sse2
This patch also solves the seg fault caused by low bit-depth and
high bit-depth paths
BUG=aomedia:967
BUG=aomedia:944
Change-Id: Iea69f114e81ca226a30d84a540ad846f1b94b8d6
diff --git a/aom_dsp/aom_dsp_rtcd_defs.pl b/aom_dsp/aom_dsp_rtcd_defs.pl
index 66a6f25..cf334b8 100755
--- a/aom_dsp/aom_dsp_rtcd_defs.pl
+++ b/aom_dsp/aom_dsp_rtcd_defs.pl
@@ -926,6 +926,9 @@
specialize "aom_highbd_sad${w}x${h}", qw/sse2/;
specialize "aom_highbd_sad${w}x${h}_avg", qw/sse2/;
}
+ if (aom_config("CONFIG_JNT_COMP") eq "yes") {
+ add_proto qw/unsigned int/, "aom_highbd_jnt_sad${w}x${h}_avg", "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred, const JNT_COMP_PARAMS* jcp_param";
+ }
}
specialize qw/aom_highbd_sad128x128 avx2/;
specialize qw/aom_highbd_sad128x64 avx2/;
@@ -1166,6 +1169,11 @@
specialize qw/aom_highbd_upsampled_pred sse2/;
add_proto qw/void aom_highbd_comp_avg_upsampled_pred/, "uint16_t *comp_pred, const uint8_t *pred8, int width, int height, int subsample_x_q3, int subsample_y_q3, const uint8_t *ref8, int ref_stride, int bd";
specialize qw/aom_highbd_comp_avg_upsampled_pred sse2/;
+
+ if (aom_config("CONFIG_JNT_COMP") eq "yes") {
+ add_proto qw/void aom_highbd_jnt_comp_avg_upsampled_pred/, "uint16_t *comp_pred, const uint8_t *pred8, int width, int height, int subsample_x_q3, int subsample_y_q3, const uint8_t *ref8, int ref_stride, int bd, const JNT_COMP_PARAMS *jcp_param";
+ specialize qw/aom_highbd_jnt_comp_avg_upsampled_pred sse2/;
+ }
}
#
@@ -1324,6 +1332,10 @@
specialize "aom_highbd_${bd}_sub_pixel_variance${w}x${h}", "sse4_1";
specialize "aom_highbd_${bd}_sub_pixel_avg_variance${w}x${h}", "sse4_1";
}
+
+ if (aom_config("CONFIG_JNT_COMP") eq "yes") {
+ add_proto qw/uint32_t/, "aom_highbd_${bd}_jnt_sub_pixel_avg_variance${w}x${h}", "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse, const uint8_t *second_pred, const JNT_COMP_PARAMS* jcp_param";
+ }
}
}
} # CONFIG_HIGHBITDEPTH
@@ -1564,6 +1576,11 @@
add_proto qw/void aom_highbd_comp_avg_pred/, "uint16_t *comp_pred, const uint8_t *pred8, int width, int height, const uint8_t *ref8, int ref_stride";
+ if (aom_config("CONFIG_JNT_COMP") eq "yes") {
+ add_proto qw/void aom_highbd_jnt_comp_avg_pred/, "uint16_t *comp_pred, const uint8_t *pred8, int width, int height, const uint8_t *ref8, int ref_stride, const JNT_COMP_PARAMS *jcp_param";
+ specialize qw/aom_highbd_jnt_comp_avg_pred sse2/;
+ }
+
#
# Subpixel Variance
#
diff --git a/aom_dsp/sad.c b/aom_dsp/sad.c
index e42ec1b..809437a 100644
--- a/aom_dsp/sad.c
+++ b/aom_dsp/sad.c
@@ -247,6 +247,29 @@
return sad;
}
+#if CONFIG_JNT_COMP
+#define highbd_sadMxN(m, n) \
+ unsigned int aom_highbd_sad##m##x##n##_c(const uint8_t *src, int src_stride, \
+ const uint8_t *ref, \
+ int ref_stride) { \
+ return highbd_sad(src, src_stride, ref, ref_stride, m, n); \
+ } \
+ unsigned int aom_highbd_sad##m##x##n##_avg_c( \
+ const uint8_t *src, int src_stride, const uint8_t *ref, int ref_stride, \
+ const uint8_t *second_pred) { \
+ uint16_t comp_pred[m * n]; \
+ aom_highbd_comp_avg_pred(comp_pred, second_pred, m, n, ref, ref_stride); \
+ return highbd_sadb(src, src_stride, comp_pred, m, m, n); \
+ } \
+ unsigned int aom_highbd_jnt_sad##m##x##n##_avg_c( \
+ const uint8_t *src, int src_stride, const uint8_t *ref, int ref_stride, \
+ const uint8_t *second_pred, const JNT_COMP_PARAMS *jcp_param) { \
+ uint16_t comp_pred[m * n]; \
+ aom_highbd_jnt_comp_avg_pred(comp_pred, second_pred, m, n, ref, \
+ ref_stride, jcp_param); \
+ return highbd_sadb(src, src_stride, comp_pred, m, m, n); \
+ }
+#else
#define highbd_sadMxN(m, n) \
unsigned int aom_highbd_sad##m##x##n##_c(const uint8_t *src, int src_stride, \
const uint8_t *ref, \
@@ -260,6 +283,7 @@
aom_highbd_comp_avg_pred_c(comp_pred, second_pred, m, n, ref, ref_stride); \
return highbd_sadb(src, src_stride, comp_pred, m, m, n); \
}
+#endif // CONFIG_JNT_COMP
#define highbd_sadMxNxK(m, n, k) \
void aom_highbd_sad##m##x##n##x##k##_c( \
diff --git a/aom_dsp/variance.c b/aom_dsp/variance.c
index 71a00f4..b47eca3 100644
--- a/aom_dsp/variance.c
+++ b/aom_dsp/variance.c
@@ -655,6 +655,128 @@
dst, dst_stride, sse); \
}
+#if CONFIG_JNT_COMP
+#define HIGHBD_SUBPIX_AVG_VAR(W, H) \
+ uint32_t aom_highbd_8_sub_pixel_avg_variance##W##x##H##_c( \
+ const uint8_t *src, int src_stride, int xoffset, int yoffset, \
+ const uint8_t *dst, int dst_stride, uint32_t *sse, \
+ const uint8_t *second_pred) { \
+ uint16_t fdata3[(H + 1) * W]; \
+ uint16_t temp2[H * W]; \
+ DECLARE_ALIGNED(16, uint16_t, temp3[H * W]); \
+ \
+ aom_highbd_var_filter_block2d_bil_first_pass( \
+ src, fdata3, src_stride, 1, H + 1, W, bilinear_filters_2t[xoffset]); \
+ aom_highbd_var_filter_block2d_bil_second_pass( \
+ fdata3, temp2, W, W, H, W, bilinear_filters_2t[yoffset]); \
+ \
+ aom_highbd_comp_avg_pred_c(temp3, second_pred, W, H, \
+ CONVERT_TO_BYTEPTR(temp2), W); \
+ \
+ return aom_highbd_8_variance##W##x##H##_c(CONVERT_TO_BYTEPTR(temp3), W, \
+ dst, dst_stride, sse); \
+ } \
+ \
+ uint32_t aom_highbd_10_sub_pixel_avg_variance##W##x##H##_c( \
+ const uint8_t *src, int src_stride, int xoffset, int yoffset, \
+ const uint8_t *dst, int dst_stride, uint32_t *sse, \
+ const uint8_t *second_pred) { \
+ uint16_t fdata3[(H + 1) * W]; \
+ uint16_t temp2[H * W]; \
+ DECLARE_ALIGNED(16, uint16_t, temp3[H * W]); \
+ \
+ aom_highbd_var_filter_block2d_bil_first_pass( \
+ src, fdata3, src_stride, 1, H + 1, W, bilinear_filters_2t[xoffset]); \
+ aom_highbd_var_filter_block2d_bil_second_pass( \
+ fdata3, temp2, W, W, H, W, bilinear_filters_2t[yoffset]); \
+ \
+ aom_highbd_comp_avg_pred_c(temp3, second_pred, W, H, \
+ CONVERT_TO_BYTEPTR(temp2), W); \
+ \
+ return aom_highbd_10_variance##W##x##H##_c(CONVERT_TO_BYTEPTR(temp3), W, \
+ dst, dst_stride, sse); \
+ } \
+ \
+ uint32_t aom_highbd_12_sub_pixel_avg_variance##W##x##H##_c( \
+ const uint8_t *src, int src_stride, int xoffset, int yoffset, \
+ const uint8_t *dst, int dst_stride, uint32_t *sse, \
+ const uint8_t *second_pred) { \
+ uint16_t fdata3[(H + 1) * W]; \
+ uint16_t temp2[H * W]; \
+ DECLARE_ALIGNED(16, uint16_t, temp3[H * W]); \
+ \
+ aom_highbd_var_filter_block2d_bil_first_pass( \
+ src, fdata3, src_stride, 1, H + 1, W, bilinear_filters_2t[xoffset]); \
+ aom_highbd_var_filter_block2d_bil_second_pass( \
+ fdata3, temp2, W, W, H, W, bilinear_filters_2t[yoffset]); \
+ \
+ aom_highbd_comp_avg_pred_c(temp3, second_pred, W, H, \
+ CONVERT_TO_BYTEPTR(temp2), W); \
+ \
+ return aom_highbd_12_variance##W##x##H##_c(CONVERT_TO_BYTEPTR(temp3), W, \
+ dst, dst_stride, sse); \
+ } \
+ \
+ uint32_t aom_highbd_8_jnt_sub_pixel_avg_variance##W##x##H##_c( \
+ const uint8_t *src, int src_stride, int xoffset, int yoffset, \
+ const uint8_t *dst, int dst_stride, uint32_t *sse, \
+ const uint8_t *second_pred, const JNT_COMP_PARAMS *jcp_param) { \
+ uint16_t fdata3[(H + 1) * W]; \
+ uint16_t temp2[H * W]; \
+ DECLARE_ALIGNED(16, uint16_t, temp3[H * W]); \
+ \
+ aom_highbd_var_filter_block2d_bil_first_pass( \
+ src, fdata3, src_stride, 1, H + 1, W, bilinear_filters_2t[xoffset]); \
+ aom_highbd_var_filter_block2d_bil_second_pass( \
+ fdata3, temp2, W, W, H, W, bilinear_filters_2t[yoffset]); \
+ \
+ aom_highbd_jnt_comp_avg_pred(temp3, second_pred, W, H, \
+ CONVERT_TO_BYTEPTR(temp2), W, jcp_param); \
+ \
+ return aom_highbd_8_variance##W##x##H(CONVERT_TO_BYTEPTR(temp3), W, dst, \
+ dst_stride, sse); \
+ } \
+ \
+ uint32_t aom_highbd_10_jnt_sub_pixel_avg_variance##W##x##H##_c( \
+ const uint8_t *src, int src_stride, int xoffset, int yoffset, \
+ const uint8_t *dst, int dst_stride, uint32_t *sse, \
+ const uint8_t *second_pred, const JNT_COMP_PARAMS *jcp_param) { \
+ uint16_t fdata3[(H + 1) * W]; \
+ uint16_t temp2[H * W]; \
+ DECLARE_ALIGNED(16, uint16_t, temp3[H * W]); \
+ \
+ aom_highbd_var_filter_block2d_bil_first_pass( \
+ src, fdata3, src_stride, 1, H + 1, W, bilinear_filters_2t[xoffset]); \
+ aom_highbd_var_filter_block2d_bil_second_pass( \
+ fdata3, temp2, W, W, H, W, bilinear_filters_2t[yoffset]); \
+ \
+ aom_highbd_jnt_comp_avg_pred(temp3, second_pred, W, H, \
+ CONVERT_TO_BYTEPTR(temp2), W, jcp_param); \
+ \
+ return aom_highbd_10_variance##W##x##H(CONVERT_TO_BYTEPTR(temp3), W, dst, \
+ dst_stride, sse); \
+ } \
+ \
+ uint32_t aom_highbd_12_jnt_sub_pixel_avg_variance##W##x##H##_c( \
+ const uint8_t *src, int src_stride, int xoffset, int yoffset, \
+ const uint8_t *dst, int dst_stride, uint32_t *sse, \
+ const uint8_t *second_pred, const JNT_COMP_PARAMS *jcp_param) { \
+ uint16_t fdata3[(H + 1) * W]; \
+ uint16_t temp2[H * W]; \
+ DECLARE_ALIGNED(16, uint16_t, temp3[H * W]); \
+ \
+ aom_highbd_var_filter_block2d_bil_first_pass( \
+ src, fdata3, src_stride, 1, H + 1, W, bilinear_filters_2t[xoffset]); \
+ aom_highbd_var_filter_block2d_bil_second_pass( \
+ fdata3, temp2, W, W, H, W, bilinear_filters_2t[yoffset]); \
+ \
+ aom_highbd_jnt_comp_avg_pred(temp3, second_pred, W, H, \
+ CONVERT_TO_BYTEPTR(temp2), W, jcp_param); \
+ \
+ return aom_highbd_12_variance##W##x##H(CONVERT_TO_BYTEPTR(temp3), W, dst, \
+ dst_stride, sse); \
+ }
+#else // CONFIG_JNT_COMP
#define HIGHBD_SUBPIX_AVG_VAR(W, H) \
uint32_t aom_highbd_8_sub_pixel_avg_variance##W##x##H##_c( \
const uint8_t *src, int src_stride, int xoffset, int yoffset, \
@@ -715,6 +837,7 @@
return aom_highbd_12_variance##W##x##H##_c(CONVERT_TO_BYTEPTR(temp3), W, \
dst, dst_stride, sse); \
}
+#endif // CONFIG_JNT_COMP
/* All three forms of the variance are available in the same sizes. */
#define HIGHBD_VARIANCES(W, H) \
@@ -853,6 +976,53 @@
pred += width;
}
}
+
+#if CONFIG_JNT_COMP
+void aom_highbd_jnt_comp_avg_pred_c(uint16_t *comp_pred, const uint8_t *pred8,
+ int width, int height, const uint8_t *ref8,
+ int ref_stride,
+ const JNT_COMP_PARAMS *jcp_param) {
+ int i, j;
+ const int fwd_offset = jcp_param->fwd_offset;
+ const int bck_offset = jcp_param->bck_offset;
+ uint16_t *pred = CONVERT_TO_SHORTPTR(pred8);
+ uint16_t *ref = CONVERT_TO_SHORTPTR(ref8);
+
+ for (i = 0; i < height; ++i) {
+ for (j = 0; j < width; ++j) {
+ int tmp = pred[j] * bck_offset + ref[j] * fwd_offset;
+ tmp = ROUND_POWER_OF_TWO(tmp, DIST_PRECISION_BITS);
+ comp_pred[j] = (uint16_t)tmp;
+ }
+ comp_pred += width;
+ pred += width;
+ ref += ref_stride;
+ }
+}
+
+void aom_highbd_jnt_comp_avg_upsampled_pred_c(
+ uint16_t *comp_pred, const uint8_t *pred8, int width, int height,
+ int subpel_x_q3, int subpel_y_q3, const uint8_t *ref8, int ref_stride,
+ int bd, const JNT_COMP_PARAMS *jcp_param) {
+ int i, j;
+ const int fwd_offset = jcp_param->fwd_offset;
+ const int bck_offset = jcp_param->bck_offset;
+ const uint16_t *pred = CONVERT_TO_SHORTPTR(pred8);
+
+ aom_highbd_upsampled_pred(comp_pred, width, height, subpel_x_q3, subpel_y_q3,
+ ref8, ref_stride, bd);
+
+ for (i = 0; i < height; i++) {
+ for (j = 0; j < width; j++) {
+ int tmp = pred[j] * bck_offset + comp_pred[j] * fwd_offset;
+ tmp = ROUND_POWER_OF_TWO(tmp, DIST_PRECISION_BITS);
+ comp_pred[j] = (uint16_t)tmp;
+ }
+ comp_pred += width;
+ pred += width;
+ }
+}
+#endif // CONFIG_JNT_COMP
#endif // CONFIG_HIGHBITDEPTH
#if CONFIG_AV1
diff --git a/aom_dsp/x86/highbd_variance_sse2.c b/aom_dsp/x86/highbd_variance_sse2.c
index 62acf3e..8448985 100644
--- a/aom_dsp/x86/highbd_variance_sse2.c
+++ b/aom_dsp/x86/highbd_variance_sse2.c
@@ -15,6 +15,8 @@
#include "./aom_config.h"
#include "./aom_dsp_rtcd.h"
+#include "aom_dsp/x86/synonyms.h"
+
#include "aom_ports/mem.h"
#include "./av1_rtcd.h"
@@ -707,3 +709,98 @@
pred += 8;
}
}
+
+#if CONFIG_JNT_COMP
+static void highbd_compute_jnt_comp_avg(__m128i *p0, __m128i *p1,
+ const __m128i *w0, const __m128i *w1,
+ const __m128i *r, void *const result) {
+ __m128i mult0 = _mm_mullo_epi16(*p0, *w0);
+ __m128i mult1 = _mm_mullo_epi16(*p1, *w1);
+ __m128i sum = _mm_add_epi16(mult0, mult1);
+ __m128i round = _mm_add_epi16(sum, *r);
+ __m128i shift = _mm_srai_epi16(round, DIST_PRECISION_BITS);
+
+ xx_storeu_128(result, shift);
+}
+
+void aom_highbd_jnt_comp_avg_pred_sse2(uint16_t *comp_pred,
+ const uint8_t *pred8, int width,
+ int height, const uint8_t *ref8,
+ int ref_stride,
+ const JNT_COMP_PARAMS *jcp_param) {
+ int i;
+ const uint16_t wt0 = (uint16_t)jcp_param->fwd_offset;
+ const uint16_t wt1 = (uint16_t)jcp_param->bck_offset;
+ const __m128i w0 = _mm_set_epi16(wt0, wt0, wt0, wt0, wt0, wt0, wt0, wt0);
+ const __m128i w1 = _mm_set_epi16(wt1, wt1, wt1, wt1, wt1, wt1, wt1, wt1);
+ const uint16_t round = ((1 << DIST_PRECISION_BITS) >> 1);
+ const __m128i r =
+ _mm_set_epi16(round, round, round, round, round, round, round, round);
+ uint16_t *pred = CONVERT_TO_SHORTPTR(pred8);
+ uint16_t *ref = CONVERT_TO_SHORTPTR(ref8);
+
+ if (width >= 8) {
+ // Read 8 pixels one row at a time
+ assert(!(width & 7));
+ for (i = 0; i < height; ++i) {
+ int j;
+ for (j = 0; j < width; j += 8) {
+ __m128i p0 = xx_loadu_128(ref);
+ __m128i p1 = xx_loadu_128(pred);
+
+ highbd_compute_jnt_comp_avg(&p0, &p1, &w0, &w1, &r, comp_pred);
+
+ comp_pred += 8;
+ pred += 8;
+ }
+ ref += ref_stride - width;
+ }
+ } else {
+ // Read 4 pixels two rows at a time
+ assert(!(width & 3));
+ for (i = 0; i < height; i += 2) {
+ __m128i p0_0 = xx_loadl_64(ref + 0 * ref_stride);
+ __m128i p0_1 = xx_loadl_64(ref + 1 * ref_stride);
+ __m128i p0 = _mm_unpacklo_epi64(p0_0, p0_1);
+ __m128i p1 = xx_loadu_128(pred);
+
+ highbd_compute_jnt_comp_avg(&p0, &p1, &w0, &w1, &r, comp_pred);
+
+ comp_pred += 8;
+ pred += 8;
+ ref += 2 * ref_stride;
+ }
+ }
+}
+
+void aom_highbd_jnt_comp_avg_upsampled_pred_sse2(
+ uint16_t *comp_pred, const uint8_t *pred8, int width, int height,
+ int subpel_x_q3, int subpel_y_q3, const uint8_t *ref8, int ref_stride,
+ int bd, const JNT_COMP_PARAMS *jcp_param) {
+ uint16_t *pred = CONVERT_TO_SHORTPTR(pred8);
+ int n;
+ int i;
+ aom_highbd_upsampled_pred(comp_pred, width, height, subpel_x_q3, subpel_y_q3,
+ ref8, ref_stride, bd);
+ assert(!(width * height & 7));
+ n = width * height >> 3;
+
+ const uint16_t wt0 = (uint16_t)jcp_param->fwd_offset;
+ const uint16_t wt1 = (uint16_t)jcp_param->bck_offset;
+ const __m128i w0 = _mm_set_epi16(wt0, wt0, wt0, wt0, wt0, wt0, wt0, wt0);
+ const __m128i w1 = _mm_set_epi16(wt1, wt1, wt1, wt1, wt1, wt1, wt1, wt1);
+ const uint16_t round = ((1 << DIST_PRECISION_BITS) >> 1);
+ const __m128i r =
+ _mm_set_epi16(round, round, round, round, round, round, round, round);
+
+ for (i = 0; i < n; i++) {
+ __m128i p0 = xx_loadu_128(comp_pred);
+ __m128i p1 = xx_loadu_128(pred);
+
+ highbd_compute_jnt_comp_avg(&p0, &p1, &w0, &w1, &r, comp_pred);
+
+ comp_pred += 8;
+ pred += 8;
+ }
+}
+#endif // CONFIG_JNT_COMP
diff --git a/av1/encoder/encoder.c b/av1/encoder/encoder.c
index 86d7093..51c4b5e 100644
--- a/av1/encoder/encoder.c
+++ b/av1/encoder/encoder.c
@@ -1110,6 +1110,20 @@
}
#if CONFIG_HIGHBITDEPTH
+#if CONFIG_JNT_COMP
+#define HIGHBD_BFP(BT, SDF, SDAF, VF, SVF, SVAF, SDX3F, SDX8F, SDX4DF, JSDAF, \
+ JSVAF) \
+ cpi->fn_ptr[BT].sdf = SDF; \
+ cpi->fn_ptr[BT].sdaf = SDAF; \
+ cpi->fn_ptr[BT].vf = VF; \
+ cpi->fn_ptr[BT].svf = SVF; \
+ cpi->fn_ptr[BT].svaf = SVAF; \
+ cpi->fn_ptr[BT].sdx3f = SDX3F; \
+ cpi->fn_ptr[BT].sdx8f = SDX8F; \
+ cpi->fn_ptr[BT].sdx4df = SDX4DF; \
+ cpi->fn_ptr[BT].jsdaf = JSDAF; \
+ cpi->fn_ptr[BT].jsvaf = JSVAF;
+#else // CONFIG_JNT_COMP
#define HIGHBD_BFP(BT, SDF, SDAF, VF, SVF, SVAF, SDX3F, SDX8F, SDX4DF) \
cpi->fn_ptr[BT].sdf = SDF; \
cpi->fn_ptr[BT].sdaf = SDAF; \
@@ -1119,6 +1133,7 @@
cpi->fn_ptr[BT].sdx3f = SDX3F; \
cpi->fn_ptr[BT].sdx8f = SDX8F; \
cpi->fn_ptr[BT].sdx4df = SDX4DF;
+#endif // CONFIG_JNT_COMP
#define MAKE_BFP_SAD_WRAPPER(fnname) \
static unsigned int fnname##_bits8(const uint8_t *src_ptr, \
@@ -1218,6 +1233,33 @@
for (i = 0; i < 4; i++) sad_array[i] >>= 4; \
}
+#if CONFIG_JNT_COMP
+#define MAKE_BFP_JSADAVG_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 *second_pred, \
+ const JNT_COMP_PARAMS *jcp_param) { \
+ return fnname(src_ptr, source_stride, ref_ptr, ref_stride, second_pred, \
+ jcp_param); \
+ } \
+ static unsigned int fnname##_bits10( \
+ const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, \
+ int ref_stride, const uint8_t *second_pred, \
+ const JNT_COMP_PARAMS *jcp_param) { \
+ return fnname(src_ptr, source_stride, ref_ptr, ref_stride, second_pred, \
+ jcp_param) >> \
+ 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 *second_pred, \
+ const JNT_COMP_PARAMS *jcp_param) { \
+ return fnname(src_ptr, source_stride, ref_ptr, ref_stride, second_pred, \
+ jcp_param) >> \
+ 4; \
+ }
+#endif // CONFIG_JNT_COMP
+
#if CONFIG_EXT_PARTITION
MAKE_BFP_SAD_WRAPPER(aom_highbd_sad128x128)
MAKE_BFP_SADAVG_WRAPPER(aom_highbd_sad128x128_avg)
@@ -1316,6 +1358,39 @@
#endif // CONFIG_EXT_PARTITION
#endif // CONFIG_EXT_PARTITION_TYPES
+#if CONFIG_JNT_COMP
+#if CONFIG_EXT_PARTITION
+MAKE_BFP_JSADAVG_WRAPPER(aom_highbd_jnt_sad128x128_avg)
+MAKE_BFP_JSADAVG_WRAPPER(aom_highbd_jnt_sad128x64_avg)
+MAKE_BFP_JSADAVG_WRAPPER(aom_highbd_jnt_sad64x128_avg)
+#endif // CONFIG_EXT_PARTITION
+MAKE_BFP_JSADAVG_WRAPPER(aom_highbd_jnt_sad32x16_avg)
+MAKE_BFP_JSADAVG_WRAPPER(aom_highbd_jnt_sad16x32_avg)
+MAKE_BFP_JSADAVG_WRAPPER(aom_highbd_jnt_sad64x32_avg)
+MAKE_BFP_JSADAVG_WRAPPER(aom_highbd_jnt_sad32x64_avg)
+MAKE_BFP_JSADAVG_WRAPPER(aom_highbd_jnt_sad32x32_avg)
+MAKE_BFP_JSADAVG_WRAPPER(aom_highbd_jnt_sad64x64_avg)
+MAKE_BFP_JSADAVG_WRAPPER(aom_highbd_jnt_sad16x16_avg)
+MAKE_BFP_JSADAVG_WRAPPER(aom_highbd_jnt_sad16x8_avg)
+MAKE_BFP_JSADAVG_WRAPPER(aom_highbd_jnt_sad8x16_avg)
+MAKE_BFP_JSADAVG_WRAPPER(aom_highbd_jnt_sad8x8_avg)
+MAKE_BFP_JSADAVG_WRAPPER(aom_highbd_jnt_sad8x4_avg)
+MAKE_BFP_JSADAVG_WRAPPER(aom_highbd_jnt_sad4x8_avg)
+MAKE_BFP_JSADAVG_WRAPPER(aom_highbd_jnt_sad4x4_avg)
+#if CONFIG_EXT_PARTITION_TYPES
+MAKE_BFP_JSADAVG_WRAPPER(aom_highbd_jnt_sad4x16_avg)
+MAKE_BFP_JSADAVG_WRAPPER(aom_highbd_jnt_sad16x4_avg)
+MAKE_BFP_JSADAVG_WRAPPER(aom_highbd_jnt_sad8x32_avg)
+MAKE_BFP_JSADAVG_WRAPPER(aom_highbd_jnt_sad32x8_avg)
+MAKE_BFP_JSADAVG_WRAPPER(aom_highbd_jnt_sad16x64_avg)
+MAKE_BFP_JSADAVG_WRAPPER(aom_highbd_jnt_sad64x16_avg)
+#if CONFIG_EXT_PARTITION
+MAKE_BFP_JSADAVG_WRAPPER(aom_highbd_jnt_sad32x128_avg)
+MAKE_BFP_JSADAVG_WRAPPER(aom_highbd_jnt_sad128x32_avg)
+#endif // CONFIG_EXT_PARTITION
+#endif // CONFIG_EXT_PARTITION_TYPES
+#endif // CONFIG_JNT_COMP
+
#define HIGHBD_MBFP(BT, MCSDF, MCSVF) \
cpi->fn_ptr[BT].msdf = MCSDF; \
cpi->fn_ptr[BT].msvf = MCSVF;
@@ -1436,6 +1511,212 @@
if (cm->use_highbitdepth) {
switch (cm->bit_depth) {
case AOM_BITS_8:
+#if CONFIG_JNT_COMP
+#if CONFIG_EXT_PARTITION_TYPES
+#if CONFIG_EXT_PARTITION
+ HIGHBD_BFP(BLOCK_128X32, aom_highbd_sad128x32_bits8,
+ aom_highbd_sad128x32_avg_bits8, aom_highbd_8_variance128x32,
+ aom_highbd_8_sub_pixel_variance128x32,
+ aom_highbd_8_sub_pixel_avg_variance128x32, NULL, NULL,
+ aom_highbd_sad128x32x4d_bits8,
+ aom_highbd_jnt_sad128x32_avg_bits8,
+ aom_highbd_8_jnt_sub_pixel_avg_variance128x32)
+
+ HIGHBD_BFP(BLOCK_32X128, aom_highbd_sad32x128_bits8,
+ aom_highbd_sad32x128_avg_bits8, aom_highbd_8_variance32x128,
+ aom_highbd_8_sub_pixel_variance32x128,
+ aom_highbd_8_sub_pixel_avg_variance32x128, NULL, NULL,
+ aom_highbd_sad32x128x4d_bits8,
+ aom_highbd_jnt_sad32x128_avg_bits8,
+ aom_highbd_8_jnt_sub_pixel_avg_variance32x128)
+#endif // CONFIG_EXT_PARTITION
+
+ HIGHBD_BFP(BLOCK_64X16, aom_highbd_sad64x16_bits8,
+ aom_highbd_sad64x16_avg_bits8, aom_highbd_8_variance64x16,
+ aom_highbd_8_sub_pixel_variance64x16,
+ aom_highbd_8_sub_pixel_avg_variance64x16, NULL, NULL,
+ aom_highbd_sad64x16x4d_bits8,
+ aom_highbd_jnt_sad64x16_avg_bits8,
+ aom_highbd_8_jnt_sub_pixel_avg_variance64x16)
+
+ HIGHBD_BFP(BLOCK_16X64, aom_highbd_sad16x64_bits8,
+ aom_highbd_sad16x64_avg_bits8, aom_highbd_8_variance16x64,
+ aom_highbd_8_sub_pixel_variance16x64,
+ aom_highbd_8_sub_pixel_avg_variance16x64, NULL, NULL,
+ aom_highbd_sad16x64x4d_bits8,
+ aom_highbd_jnt_sad16x64_avg_bits8,
+ aom_highbd_8_jnt_sub_pixel_avg_variance16x64)
+
+ HIGHBD_BFP(
+ BLOCK_32X8, aom_highbd_sad32x8_bits8, aom_highbd_sad32x8_avg_bits8,
+ aom_highbd_8_variance32x8, aom_highbd_8_sub_pixel_variance32x8,
+ aom_highbd_8_sub_pixel_avg_variance32x8, NULL, NULL,
+ aom_highbd_sad32x8x4d_bits8, aom_highbd_jnt_sad32x8_avg_bits8,
+ aom_highbd_8_jnt_sub_pixel_avg_variance32x8)
+
+ HIGHBD_BFP(
+ BLOCK_8X32, aom_highbd_sad8x32_bits8, aom_highbd_sad8x32_avg_bits8,
+ aom_highbd_8_variance8x32, aom_highbd_8_sub_pixel_variance8x32,
+ aom_highbd_8_sub_pixel_avg_variance8x32, NULL, NULL,
+ aom_highbd_sad8x32x4d_bits8, aom_highbd_jnt_sad8x32_avg_bits8,
+ aom_highbd_8_jnt_sub_pixel_avg_variance8x32)
+
+ HIGHBD_BFP(
+ BLOCK_16X4, aom_highbd_sad16x4_bits8, aom_highbd_sad16x4_avg_bits8,
+ aom_highbd_8_variance16x4, aom_highbd_8_sub_pixel_variance16x4,
+ aom_highbd_8_sub_pixel_avg_variance16x4, NULL, NULL,
+ aom_highbd_sad16x4x4d_bits8, aom_highbd_jnt_sad16x4_avg_bits8,
+ aom_highbd_8_jnt_sub_pixel_avg_variance16x4)
+
+ HIGHBD_BFP(
+ BLOCK_4X16, aom_highbd_sad4x16_bits8, aom_highbd_sad4x16_avg_bits8,
+ aom_highbd_8_variance4x16, aom_highbd_8_sub_pixel_variance4x16,
+ aom_highbd_8_sub_pixel_avg_variance4x16, NULL, NULL,
+ aom_highbd_sad4x16x4d_bits8, aom_highbd_jnt_sad4x16_avg_bits8,
+ aom_highbd_8_jnt_sub_pixel_avg_variance4x16)
+#endif
+
+ HIGHBD_BFP(BLOCK_32X16, aom_highbd_sad32x16_bits8,
+ aom_highbd_sad32x16_avg_bits8, aom_highbd_8_variance32x16,
+ aom_highbd_8_sub_pixel_variance32x16,
+ aom_highbd_8_sub_pixel_avg_variance32x16, NULL, NULL,
+ aom_highbd_sad32x16x4d_bits8,
+ aom_highbd_jnt_sad32x16_avg_bits8,
+ aom_highbd_8_jnt_sub_pixel_avg_variance32x16)
+
+ HIGHBD_BFP(BLOCK_16X32, aom_highbd_sad16x32_bits8,
+ aom_highbd_sad16x32_avg_bits8, aom_highbd_8_variance16x32,
+ aom_highbd_8_sub_pixel_variance16x32,
+ aom_highbd_8_sub_pixel_avg_variance16x32, NULL, NULL,
+ aom_highbd_sad16x32x4d_bits8,
+ aom_highbd_jnt_sad16x32_avg_bits8,
+ aom_highbd_8_jnt_sub_pixel_avg_variance16x32)
+
+ HIGHBD_BFP(BLOCK_64X32, aom_highbd_sad64x32_bits8,
+ aom_highbd_sad64x32_avg_bits8, aom_highbd_8_variance64x32,
+ aom_highbd_8_sub_pixel_variance64x32,
+ aom_highbd_8_sub_pixel_avg_variance64x32, NULL, NULL,
+ aom_highbd_sad64x32x4d_bits8,
+ aom_highbd_jnt_sad64x32_avg_bits8,
+ aom_highbd_8_jnt_sub_pixel_avg_variance64x32)
+
+ HIGHBD_BFP(BLOCK_32X64, aom_highbd_sad32x64_bits8,
+ aom_highbd_sad32x64_avg_bits8, aom_highbd_8_variance32x64,
+ aom_highbd_8_sub_pixel_variance32x64,
+ aom_highbd_8_sub_pixel_avg_variance32x64, NULL, NULL,
+ aom_highbd_sad32x64x4d_bits8,
+ aom_highbd_jnt_sad32x64_avg_bits8,
+ aom_highbd_8_jnt_sub_pixel_avg_variance32x64)
+
+ HIGHBD_BFP(BLOCK_32X32, aom_highbd_sad32x32_bits8,
+ aom_highbd_sad32x32_avg_bits8, aom_highbd_8_variance32x32,
+ aom_highbd_8_sub_pixel_variance32x32,
+ aom_highbd_8_sub_pixel_avg_variance32x32,
+ aom_highbd_sad32x32x3_bits8, aom_highbd_sad32x32x8_bits8,
+ aom_highbd_sad32x32x4d_bits8,
+ aom_highbd_jnt_sad32x32_avg_bits8,
+ aom_highbd_8_jnt_sub_pixel_avg_variance32x32)
+
+ HIGHBD_BFP(BLOCK_64X64, aom_highbd_sad64x64_bits8,
+ aom_highbd_sad64x64_avg_bits8, aom_highbd_8_variance64x64,
+ aom_highbd_8_sub_pixel_variance64x64,
+ aom_highbd_8_sub_pixel_avg_variance64x64,
+ aom_highbd_sad64x64x3_bits8, aom_highbd_sad64x64x8_bits8,
+ aom_highbd_sad64x64x4d_bits8,
+ aom_highbd_jnt_sad64x64_avg_bits8,
+ aom_highbd_8_jnt_sub_pixel_avg_variance64x64)
+
+ HIGHBD_BFP(BLOCK_16X16, aom_highbd_sad16x16_bits8,
+ aom_highbd_sad16x16_avg_bits8, aom_highbd_8_variance16x16,
+ aom_highbd_8_sub_pixel_variance16x16,
+ aom_highbd_8_sub_pixel_avg_variance16x16,
+ aom_highbd_sad16x16x3_bits8, aom_highbd_sad16x16x8_bits8,
+ aom_highbd_sad16x16x4d_bits8,
+ aom_highbd_jnt_sad16x16_avg_bits8,
+ aom_highbd_8_jnt_sub_pixel_avg_variance16x16)
+
+ HIGHBD_BFP(
+ BLOCK_16X8, aom_highbd_sad16x8_bits8, aom_highbd_sad16x8_avg_bits8,
+ aom_highbd_8_variance16x8, aom_highbd_8_sub_pixel_variance16x8,
+ aom_highbd_8_sub_pixel_avg_variance16x8, aom_highbd_sad16x8x3_bits8,
+ aom_highbd_sad16x8x8_bits8, aom_highbd_sad16x8x4d_bits8,
+ aom_highbd_jnt_sad16x8_avg_bits8,
+ aom_highbd_8_jnt_sub_pixel_avg_variance16x8)
+
+ HIGHBD_BFP(
+ BLOCK_8X16, aom_highbd_sad8x16_bits8, aom_highbd_sad8x16_avg_bits8,
+ aom_highbd_8_variance8x16, aom_highbd_8_sub_pixel_variance8x16,
+ aom_highbd_8_sub_pixel_avg_variance8x16, aom_highbd_sad8x16x3_bits8,
+ aom_highbd_sad8x16x8_bits8, aom_highbd_sad8x16x4d_bits8,
+ aom_highbd_jnt_sad8x16_avg_bits8,
+ aom_highbd_8_jnt_sub_pixel_avg_variance8x16)
+
+ HIGHBD_BFP(BLOCK_8X8, aom_highbd_sad8x8_bits8,
+ aom_highbd_sad8x8_avg_bits8, aom_highbd_8_variance8x8,
+ aom_highbd_8_sub_pixel_variance8x8,
+ aom_highbd_8_sub_pixel_avg_variance8x8,
+ aom_highbd_sad8x8x3_bits8, aom_highbd_sad8x8x8_bits8,
+ aom_highbd_sad8x8x4d_bits8, aom_highbd_jnt_sad8x8_avg_bits8,
+ aom_highbd_8_jnt_sub_pixel_avg_variance8x8)
+
+ HIGHBD_BFP(BLOCK_8X4, aom_highbd_sad8x4_bits8,
+ aom_highbd_sad8x4_avg_bits8, aom_highbd_8_variance8x4,
+ aom_highbd_8_sub_pixel_variance8x4,
+ aom_highbd_8_sub_pixel_avg_variance8x4, NULL,
+ aom_highbd_sad8x4x8_bits8, aom_highbd_sad8x4x4d_bits8,
+ aom_highbd_jnt_sad8x4_avg_bits8,
+ aom_highbd_8_jnt_sub_pixel_avg_variance8x4)
+
+ HIGHBD_BFP(BLOCK_4X8, aom_highbd_sad4x8_bits8,
+ aom_highbd_sad4x8_avg_bits8, aom_highbd_8_variance4x8,
+ aom_highbd_8_sub_pixel_variance4x8,
+ aom_highbd_8_sub_pixel_avg_variance4x8, NULL,
+ aom_highbd_sad4x8x8_bits8, aom_highbd_sad4x8x4d_bits8,
+ aom_highbd_jnt_sad4x8_avg_bits8,
+ aom_highbd_8_jnt_sub_pixel_avg_variance4x8)
+
+ HIGHBD_BFP(BLOCK_4X4, aom_highbd_sad4x4_bits8,
+ aom_highbd_sad4x4_avg_bits8, aom_highbd_8_variance4x4,
+ aom_highbd_8_sub_pixel_variance4x4,
+ aom_highbd_8_sub_pixel_avg_variance4x4,
+ aom_highbd_sad4x4x3_bits8, aom_highbd_sad4x4x8_bits8,
+ aom_highbd_sad4x4x4d_bits8, aom_highbd_jnt_sad4x4_avg_bits8,
+ aom_highbd_8_jnt_sub_pixel_avg_variance4x4)
+
+ HIGHBD_BFP(BLOCK_2X2, NULL, NULL, aom_highbd_8_variance2x2, NULL, NULL,
+ NULL, NULL, NULL, NULL, NULL)
+ HIGHBD_BFP(BLOCK_4X2, NULL, NULL, aom_highbd_8_variance4x2, NULL, NULL,
+ NULL, NULL, NULL, NULL, NULL)
+ HIGHBD_BFP(BLOCK_2X4, NULL, NULL, aom_highbd_8_variance2x4, NULL, NULL,
+ NULL, NULL, NULL, NULL, NULL)
+
+#if CONFIG_EXT_PARTITION
+ HIGHBD_BFP(
+ BLOCK_128X128, aom_highbd_sad128x128_bits8,
+ aom_highbd_sad128x128_avg_bits8, aom_highbd_8_variance128x128,
+ aom_highbd_8_sub_pixel_variance128x128,
+ aom_highbd_8_sub_pixel_avg_variance128x128,
+ aom_highbd_sad128x128x3_bits8, aom_highbd_sad128x128x8_bits8,
+ aom_highbd_sad128x128x4d_bits8, aom_highbd_jnt_sad128x128_avg_bits8,
+ aom_highbd_8_jnt_sub_pixel_avg_variance128x128)
+
+ HIGHBD_BFP(BLOCK_128X64, aom_highbd_sad128x64_bits8,
+ aom_highbd_sad128x64_avg_bits8, aom_highbd_8_variance128x64,
+ aom_highbd_8_sub_pixel_variance128x64,
+ aom_highbd_8_sub_pixel_avg_variance128x64, NULL, NULL,
+ aom_highbd_sad128x64x4d_bits8,
+ aom_highbd_jnt_sad128x64_avg_bits8,
+ aom_highbd_8_jnt_sub_pixel_avg_variance128x64)
+
+ HIGHBD_BFP(BLOCK_64X128, aom_highbd_sad64x128_bits8,
+ aom_highbd_sad64x128_avg_bits8, aom_highbd_8_variance64x128,
+ aom_highbd_8_sub_pixel_variance64x128,
+ aom_highbd_8_sub_pixel_avg_variance64x128, NULL, NULL,
+ aom_highbd_sad64x128x4d_bits8,
+ aom_highbd_jnt_sad64x128_avg_bits8,
+ aom_highbd_8_jnt_sub_pixel_avg_variance64x128)
+#endif // CONFIG_EXT_PARTITION
+#else // CONFIG_JNT_COMP
#if CONFIG_EXT_PARTITION_TYPES
#if CONFIG_EXT_PARTITION
HIGHBD_BFP(BLOCK_128X32, aom_highbd_sad128x32_bits8,
@@ -1590,6 +1871,7 @@
aom_highbd_8_sub_pixel_avg_variance64x128, NULL, NULL,
aom_highbd_sad64x128x4d_bits8)
#endif // CONFIG_EXT_PARTITION
+#endif // CONFIG_JNT_COMP
#if CONFIG_EXT_PARTITION
HIGHBD_MBFP(BLOCK_128X128, aom_highbd_masked_sad128x128_bits8,
@@ -1740,6 +2022,219 @@
break;
case AOM_BITS_10:
+#if CONFIG_JNT_COMP
+#if CONFIG_EXT_PARTITION_TYPES
+#if CONFIG_EXT_PARTITION
+ HIGHBD_BFP(
+ BLOCK_128X32, aom_highbd_sad128x32_bits10,
+ aom_highbd_sad128x32_avg_bits10, aom_highbd_10_variance128x32,
+ aom_highbd_10_sub_pixel_variance128x32,
+ aom_highbd_10_sub_pixel_avg_variance128x32, NULL, NULL,
+ aom_highbd_sad128x32x4d_bits10, aom_highbd_jnt_sad128x32_avg_bits10,
+ aom_highbd_10_jnt_sub_pixel_avg_variance128x32);
+
+ HIGHBD_BFP(
+ BLOCK_32X128, aom_highbd_sad32x128_bits10,
+ aom_highbd_sad32x128_avg_bits10, aom_highbd_10_variance32x128,
+ aom_highbd_10_sub_pixel_variance32x128,
+ aom_highbd_10_sub_pixel_avg_variance32x128, NULL, NULL,
+ aom_highbd_sad32x128x4d_bits10, aom_highbd_jnt_sad32x128_avg_bits10,
+ aom_highbd_10_jnt_sub_pixel_avg_variance32x128);
+#endif // CONFIG_EXT_PARTITION
+
+ HIGHBD_BFP(BLOCK_64X16, aom_highbd_sad64x16_bits10,
+ aom_highbd_sad64x16_avg_bits10, aom_highbd_10_variance64x16,
+ aom_highbd_10_sub_pixel_variance64x16,
+ aom_highbd_10_sub_pixel_avg_variance64x16, NULL, NULL,
+ aom_highbd_sad64x16x4d_bits10,
+ aom_highbd_jnt_sad64x16_avg_bits10,
+ aom_highbd_10_jnt_sub_pixel_avg_variance64x16);
+
+ HIGHBD_BFP(BLOCK_16X64, aom_highbd_sad16x64_bits10,
+ aom_highbd_sad16x64_avg_bits10, aom_highbd_10_variance16x64,
+ aom_highbd_10_sub_pixel_variance16x64,
+ aom_highbd_10_sub_pixel_avg_variance16x64, NULL, NULL,
+ aom_highbd_sad16x64x4d_bits10,
+ aom_highbd_jnt_sad16x64_avg_bits10,
+ aom_highbd_10_jnt_sub_pixel_avg_variance16x64);
+
+ HIGHBD_BFP(BLOCK_32X8, aom_highbd_sad32x8_bits10,
+ aom_highbd_sad32x8_avg_bits10, aom_highbd_10_variance32x8,
+ aom_highbd_10_sub_pixel_variance32x8,
+ aom_highbd_10_sub_pixel_avg_variance32x8, NULL, NULL,
+ aom_highbd_sad32x8x4d_bits10,
+ aom_highbd_jnt_sad32x8_avg_bits10,
+ aom_highbd_10_jnt_sub_pixel_avg_variance32x8);
+
+ HIGHBD_BFP(BLOCK_8X32, aom_highbd_sad8x32_bits10,
+ aom_highbd_sad8x32_avg_bits10, aom_highbd_10_variance8x32,
+ aom_highbd_10_sub_pixel_variance8x32,
+ aom_highbd_10_sub_pixel_avg_variance8x32, NULL, NULL,
+ aom_highbd_sad8x32x4d_bits10,
+ aom_highbd_jnt_sad8x32_avg_bits10,
+ aom_highbd_10_jnt_sub_pixel_avg_variance8x32);
+
+ HIGHBD_BFP(BLOCK_16X4, aom_highbd_sad16x4_bits10,
+ aom_highbd_sad16x4_avg_bits10, aom_highbd_10_variance16x4,
+ aom_highbd_10_sub_pixel_variance16x4,
+ aom_highbd_10_sub_pixel_avg_variance16x4, NULL, NULL,
+ aom_highbd_sad16x4x4d_bits10,
+ aom_highbd_jnt_sad16x4_avg_bits10,
+ aom_highbd_10_jnt_sub_pixel_avg_variance16x4);
+
+ HIGHBD_BFP(BLOCK_4X16, aom_highbd_sad4x16_bits10,
+ aom_highbd_sad4x16_avg_bits10, aom_highbd_10_variance4x16,
+ aom_highbd_10_sub_pixel_variance4x16,
+ aom_highbd_10_sub_pixel_avg_variance4x16, NULL, NULL,
+ aom_highbd_sad4x16x4d_bits10,
+ aom_highbd_jnt_sad4x16_avg_bits10,
+ aom_highbd_10_jnt_sub_pixel_avg_variance4x16);
+#endif
+
+ HIGHBD_BFP(BLOCK_32X16, aom_highbd_sad32x16_bits10,
+ aom_highbd_sad32x16_avg_bits10, aom_highbd_10_variance32x16,
+ aom_highbd_10_sub_pixel_variance32x16,
+ aom_highbd_10_sub_pixel_avg_variance32x16, NULL, NULL,
+ aom_highbd_sad32x16x4d_bits10,
+ aom_highbd_jnt_sad32x16_avg_bits10,
+ aom_highbd_10_jnt_sub_pixel_avg_variance32x16);
+
+ HIGHBD_BFP(BLOCK_16X32, aom_highbd_sad16x32_bits10,
+ aom_highbd_sad16x32_avg_bits10, aom_highbd_10_variance16x32,
+ aom_highbd_10_sub_pixel_variance16x32,
+ aom_highbd_10_sub_pixel_avg_variance16x32, NULL, NULL,
+ aom_highbd_sad16x32x4d_bits10,
+ aom_highbd_jnt_sad16x32_avg_bits10,
+ aom_highbd_10_jnt_sub_pixel_avg_variance16x32);
+
+ HIGHBD_BFP(BLOCK_64X32, aom_highbd_sad64x32_bits10,
+ aom_highbd_sad64x32_avg_bits10, aom_highbd_10_variance64x32,
+ aom_highbd_10_sub_pixel_variance64x32,
+ aom_highbd_10_sub_pixel_avg_variance64x32, NULL, NULL,
+ aom_highbd_sad64x32x4d_bits10,
+ aom_highbd_jnt_sad64x32_avg_bits10,
+ aom_highbd_10_jnt_sub_pixel_avg_variance64x32);
+
+ HIGHBD_BFP(BLOCK_32X64, aom_highbd_sad32x64_bits10,
+ aom_highbd_sad32x64_avg_bits10, aom_highbd_10_variance32x64,
+ aom_highbd_10_sub_pixel_variance32x64,
+ aom_highbd_10_sub_pixel_avg_variance32x64, NULL, NULL,
+ aom_highbd_sad32x64x4d_bits10,
+ aom_highbd_jnt_sad32x64_avg_bits10,
+ aom_highbd_10_jnt_sub_pixel_avg_variance32x64);
+
+ HIGHBD_BFP(BLOCK_32X32, aom_highbd_sad32x32_bits10,
+ aom_highbd_sad32x32_avg_bits10, aom_highbd_10_variance32x32,
+ aom_highbd_10_sub_pixel_variance32x32,
+ aom_highbd_10_sub_pixel_avg_variance32x32,
+ aom_highbd_sad32x32x3_bits10, aom_highbd_sad32x32x8_bits10,
+ aom_highbd_sad32x32x4d_bits10,
+ aom_highbd_jnt_sad32x32_avg_bits10,
+ aom_highbd_10_jnt_sub_pixel_avg_variance32x32);
+
+ HIGHBD_BFP(BLOCK_64X64, aom_highbd_sad64x64_bits10,
+ aom_highbd_sad64x64_avg_bits10, aom_highbd_10_variance64x64,
+ aom_highbd_10_sub_pixel_variance64x64,
+ aom_highbd_10_sub_pixel_avg_variance64x64,
+ aom_highbd_sad64x64x3_bits10, aom_highbd_sad64x64x8_bits10,
+ aom_highbd_sad64x64x4d_bits10,
+ aom_highbd_jnt_sad64x64_avg_bits10,
+ aom_highbd_10_jnt_sub_pixel_avg_variance64x64);
+
+ HIGHBD_BFP(BLOCK_16X16, aom_highbd_sad16x16_bits10,
+ aom_highbd_sad16x16_avg_bits10, aom_highbd_10_variance16x16,
+ aom_highbd_10_sub_pixel_variance16x16,
+ aom_highbd_10_sub_pixel_avg_variance16x16,
+ aom_highbd_sad16x16x3_bits10, aom_highbd_sad16x16x8_bits10,
+ aom_highbd_sad16x16x4d_bits10,
+ aom_highbd_jnt_sad16x16_avg_bits10,
+ aom_highbd_10_jnt_sub_pixel_avg_variance16x16);
+
+ HIGHBD_BFP(BLOCK_16X8, aom_highbd_sad16x8_bits10,
+ aom_highbd_sad16x8_avg_bits10, aom_highbd_10_variance16x8,
+ aom_highbd_10_sub_pixel_variance16x8,
+ aom_highbd_10_sub_pixel_avg_variance16x8,
+ aom_highbd_sad16x8x3_bits10, aom_highbd_sad16x8x8_bits10,
+ aom_highbd_sad16x8x4d_bits10,
+ aom_highbd_jnt_sad16x8_avg_bits10,
+ aom_highbd_10_jnt_sub_pixel_avg_variance16x8);
+
+ HIGHBD_BFP(BLOCK_8X16, aom_highbd_sad8x16_bits10,
+ aom_highbd_sad8x16_avg_bits10, aom_highbd_10_variance8x16,
+ aom_highbd_10_sub_pixel_variance8x16,
+ aom_highbd_10_sub_pixel_avg_variance8x16,
+ aom_highbd_sad8x16x3_bits10, aom_highbd_sad8x16x8_bits10,
+ aom_highbd_sad8x16x4d_bits10,
+ aom_highbd_jnt_sad8x16_avg_bits10,
+ aom_highbd_10_jnt_sub_pixel_avg_variance8x16);
+
+ HIGHBD_BFP(
+ BLOCK_8X8, aom_highbd_sad8x8_bits10, aom_highbd_sad8x8_avg_bits10,
+ aom_highbd_10_variance8x8, aom_highbd_10_sub_pixel_variance8x8,
+ aom_highbd_10_sub_pixel_avg_variance8x8, aom_highbd_sad8x8x3_bits10,
+ aom_highbd_sad8x8x8_bits10, aom_highbd_sad8x8x4d_bits10,
+ aom_highbd_jnt_sad8x8_avg_bits10,
+ aom_highbd_10_jnt_sub_pixel_avg_variance8x8);
+
+ HIGHBD_BFP(BLOCK_8X4, aom_highbd_sad8x4_bits10,
+ aom_highbd_sad8x4_avg_bits10, aom_highbd_10_variance8x4,
+ aom_highbd_10_sub_pixel_variance8x4,
+ aom_highbd_10_sub_pixel_avg_variance8x4, NULL,
+ aom_highbd_sad8x4x8_bits10, aom_highbd_sad8x4x4d_bits10,
+ aom_highbd_jnt_sad8x4_avg_bits10,
+ aom_highbd_10_jnt_sub_pixel_avg_variance8x4);
+
+ HIGHBD_BFP(BLOCK_4X8, aom_highbd_sad4x8_bits10,
+ aom_highbd_sad4x8_avg_bits10, aom_highbd_10_variance4x8,
+ aom_highbd_10_sub_pixel_variance4x8,
+ aom_highbd_10_sub_pixel_avg_variance4x8, NULL,
+ aom_highbd_sad4x8x8_bits10, aom_highbd_sad4x8x4d_bits10,
+ aom_highbd_jnt_sad4x8_avg_bits10,
+ aom_highbd_10_jnt_sub_pixel_avg_variance4x8);
+
+ HIGHBD_BFP(
+ BLOCK_4X4, aom_highbd_sad4x4_bits10, aom_highbd_sad4x4_avg_bits10,
+ aom_highbd_10_variance4x4, aom_highbd_10_sub_pixel_variance4x4,
+ aom_highbd_10_sub_pixel_avg_variance4x4, aom_highbd_sad4x4x3_bits10,
+ aom_highbd_sad4x4x8_bits10, aom_highbd_sad4x4x4d_bits10,
+ aom_highbd_jnt_sad4x4_avg_bits10,
+ aom_highbd_10_jnt_sub_pixel_avg_variance4x4);
+
+ HIGHBD_BFP(BLOCK_2X2, NULL, NULL, aom_highbd_10_variance2x2, NULL, NULL,
+ NULL, NULL, NULL, NULL, NULL)
+ HIGHBD_BFP(BLOCK_4X2, NULL, NULL, aom_highbd_10_variance4x2, NULL, NULL,
+ NULL, NULL, NULL, NULL, NULL)
+ HIGHBD_BFP(BLOCK_2X4, NULL, NULL, aom_highbd_10_variance2x4, NULL, NULL,
+ NULL, NULL, NULL, NULL, NULL)
+
+#if CONFIG_EXT_PARTITION
+ HIGHBD_BFP(
+ BLOCK_128X128, aom_highbd_sad128x128_bits10,
+ aom_highbd_sad128x128_avg_bits10, aom_highbd_10_variance128x128,
+ aom_highbd_10_sub_pixel_variance128x128,
+ aom_highbd_10_sub_pixel_avg_variance128x128,
+ aom_highbd_sad128x128x3_bits10, aom_highbd_sad128x128x8_bits10,
+ aom_highbd_sad128x128x4d_bits10,
+ aom_highbd_jnt_sad128x128_avg_bits10,
+ aom_highbd_10_jnt_sub_pixel_avg_variance128x128);
+
+ HIGHBD_BFP(
+ BLOCK_128X64, aom_highbd_sad128x64_bits10,
+ aom_highbd_sad128x64_avg_bits10, aom_highbd_10_variance128x64,
+ aom_highbd_10_sub_pixel_variance128x64,
+ aom_highbd_10_sub_pixel_avg_variance128x64, NULL, NULL,
+ aom_highbd_sad128x64x4d_bits10, aom_highbd_jnt_sad128x64_avg_bits10,
+ aom_highbd_10_jnt_sub_pixel_avg_variance128x64);
+
+ HIGHBD_BFP(
+ BLOCK_64X128, aom_highbd_sad64x128_bits10,
+ aom_highbd_sad64x128_avg_bits10, aom_highbd_10_variance64x128,
+ aom_highbd_10_sub_pixel_variance64x128,
+ aom_highbd_10_sub_pixel_avg_variance64x128, NULL, NULL,
+ aom_highbd_sad64x128x4d_bits10, aom_highbd_jnt_sad64x128_avg_bits10,
+ aom_highbd_10_jnt_sub_pixel_avg_variance64x128);
+#endif // CONFIG_EXT_PARTITION
+#else // CONFIG_JNT_COMP
#if CONFIG_EXT_PARTITION_TYPES
#if CONFIG_EXT_PARTITION
HIGHBD_BFP(BLOCK_128X32, aom_highbd_sad128x32_bits10,
@@ -1900,6 +2395,7 @@
aom_highbd_10_sub_pixel_avg_variance64x128, NULL, NULL,
aom_highbd_sad64x128x4d_bits10)
#endif // CONFIG_EXT_PARTITION
+#endif // CONFIG_JNT_COMP
#if CONFIG_EXT_PARTITION
HIGHBD_MBFP(BLOCK_128X128, aom_highbd_masked_sad128x128_bits10,
@@ -2050,6 +2546,219 @@
break;
case AOM_BITS_12:
+#if CONFIG_JNT_COMP
+#if CONFIG_EXT_PARTITION_TYPES
+#if CONFIG_EXT_PARTITION
+ HIGHBD_BFP(
+ BLOCK_128X32, aom_highbd_sad128x32_bits12,
+ aom_highbd_sad128x32_avg_bits12, aom_highbd_12_variance128x32,
+ aom_highbd_12_sub_pixel_variance128x32,
+ aom_highbd_12_sub_pixel_avg_variance128x32, NULL, NULL,
+ aom_highbd_sad128x32x4d_bits12, aom_highbd_jnt_sad128x32_avg_bits12,
+ aom_highbd_12_jnt_sub_pixel_avg_variance128x32);
+
+ HIGHBD_BFP(
+ BLOCK_32X128, aom_highbd_sad32x128_bits12,
+ aom_highbd_sad32x128_avg_bits12, aom_highbd_12_variance32x128,
+ aom_highbd_12_sub_pixel_variance32x128,
+ aom_highbd_12_sub_pixel_avg_variance32x128, NULL, NULL,
+ aom_highbd_sad32x128x4d_bits12, aom_highbd_jnt_sad32x128_avg_bits12,
+ aom_highbd_12_jnt_sub_pixel_avg_variance32x128);
+#endif // CONFIG_EXT_PARTITION
+
+ HIGHBD_BFP(BLOCK_64X16, aom_highbd_sad64x16_bits12,
+ aom_highbd_sad64x16_avg_bits12, aom_highbd_12_variance64x16,
+ aom_highbd_12_sub_pixel_variance64x16,
+ aom_highbd_12_sub_pixel_avg_variance64x16, NULL, NULL,
+ aom_highbd_sad64x16x4d_bits12,
+ aom_highbd_jnt_sad64x16_avg_bits12,
+ aom_highbd_12_jnt_sub_pixel_avg_variance64x16);
+
+ HIGHBD_BFP(BLOCK_16X64, aom_highbd_sad16x64_bits12,
+ aom_highbd_sad16x64_avg_bits12, aom_highbd_12_variance16x64,
+ aom_highbd_12_sub_pixel_variance16x64,
+ aom_highbd_12_sub_pixel_avg_variance16x64, NULL, NULL,
+ aom_highbd_sad16x64x4d_bits12,
+ aom_highbd_jnt_sad16x64_avg_bits12,
+ aom_highbd_12_jnt_sub_pixel_avg_variance16x64);
+
+ HIGHBD_BFP(BLOCK_32X8, aom_highbd_sad32x8_bits12,
+ aom_highbd_sad32x8_avg_bits12, aom_highbd_12_variance32x8,
+ aom_highbd_12_sub_pixel_variance32x8,
+ aom_highbd_12_sub_pixel_avg_variance32x8, NULL, NULL,
+ aom_highbd_sad32x8x4d_bits12,
+ aom_highbd_jnt_sad32x8_avg_bits12,
+ aom_highbd_12_jnt_sub_pixel_avg_variance32x8);
+
+ HIGHBD_BFP(BLOCK_8X32, aom_highbd_sad8x32_bits12,
+ aom_highbd_sad8x32_avg_bits12, aom_highbd_12_variance8x32,
+ aom_highbd_12_sub_pixel_variance8x32,
+ aom_highbd_12_sub_pixel_avg_variance8x32, NULL, NULL,
+ aom_highbd_sad8x32x4d_bits12,
+ aom_highbd_jnt_sad8x32_avg_bits12,
+ aom_highbd_12_jnt_sub_pixel_avg_variance8x32);
+
+ HIGHBD_BFP(BLOCK_16X4, aom_highbd_sad16x4_bits12,
+ aom_highbd_sad16x4_avg_bits12, aom_highbd_12_variance16x4,
+ aom_highbd_12_sub_pixel_variance16x4,
+ aom_highbd_12_sub_pixel_avg_variance16x4, NULL, NULL,
+ aom_highbd_sad16x4x4d_bits12,
+ aom_highbd_jnt_sad16x4_avg_bits12,
+ aom_highbd_12_jnt_sub_pixel_avg_variance16x4);
+
+ HIGHBD_BFP(BLOCK_4X16, aom_highbd_sad4x16_bits12,
+ aom_highbd_sad4x16_avg_bits12, aom_highbd_12_variance4x16,
+ aom_highbd_12_sub_pixel_variance4x16,
+ aom_highbd_12_sub_pixel_avg_variance4x16, NULL, NULL,
+ aom_highbd_sad4x16x4d_bits12,
+ aom_highbd_jnt_sad4x16_avg_bits12,
+ aom_highbd_12_jnt_sub_pixel_avg_variance4x16);
+#endif
+
+ HIGHBD_BFP(BLOCK_32X16, aom_highbd_sad32x16_bits12,
+ aom_highbd_sad32x16_avg_bits12, aom_highbd_12_variance32x16,
+ aom_highbd_12_sub_pixel_variance32x16,
+ aom_highbd_12_sub_pixel_avg_variance32x16, NULL, NULL,
+ aom_highbd_sad32x16x4d_bits12,
+ aom_highbd_jnt_sad32x16_avg_bits12,
+ aom_highbd_12_jnt_sub_pixel_avg_variance32x16);
+
+ HIGHBD_BFP(BLOCK_16X32, aom_highbd_sad16x32_bits12,
+ aom_highbd_sad16x32_avg_bits12, aom_highbd_12_variance16x32,
+ aom_highbd_12_sub_pixel_variance16x32,
+ aom_highbd_12_sub_pixel_avg_variance16x32, NULL, NULL,
+ aom_highbd_sad16x32x4d_bits12,
+ aom_highbd_jnt_sad16x32_avg_bits12,
+ aom_highbd_12_jnt_sub_pixel_avg_variance16x32);
+
+ HIGHBD_BFP(BLOCK_64X32, aom_highbd_sad64x32_bits12,
+ aom_highbd_sad64x32_avg_bits12, aom_highbd_12_variance64x32,
+ aom_highbd_12_sub_pixel_variance64x32,
+ aom_highbd_12_sub_pixel_avg_variance64x32, NULL, NULL,
+ aom_highbd_sad64x32x4d_bits12,
+ aom_highbd_jnt_sad64x32_avg_bits12,
+ aom_highbd_12_jnt_sub_pixel_avg_variance64x32);
+
+ HIGHBD_BFP(BLOCK_32X64, aom_highbd_sad32x64_bits12,
+ aom_highbd_sad32x64_avg_bits12, aom_highbd_12_variance32x64,
+ aom_highbd_12_sub_pixel_variance32x64,
+ aom_highbd_12_sub_pixel_avg_variance32x64, NULL, NULL,
+ aom_highbd_sad32x64x4d_bits12,
+ aom_highbd_jnt_sad32x64_avg_bits12,
+ aom_highbd_12_jnt_sub_pixel_avg_variance32x64);
+
+ HIGHBD_BFP(BLOCK_32X32, aom_highbd_sad32x32_bits12,
+ aom_highbd_sad32x32_avg_bits12, aom_highbd_12_variance32x32,
+ aom_highbd_12_sub_pixel_variance32x32,
+ aom_highbd_12_sub_pixel_avg_variance32x32,
+ aom_highbd_sad32x32x3_bits12, aom_highbd_sad32x32x8_bits12,
+ aom_highbd_sad32x32x4d_bits12,
+ aom_highbd_jnt_sad32x32_avg_bits12,
+ aom_highbd_12_jnt_sub_pixel_avg_variance32x32);
+
+ HIGHBD_BFP(BLOCK_64X64, aom_highbd_sad64x64_bits12,
+ aom_highbd_sad64x64_avg_bits12, aom_highbd_12_variance64x64,
+ aom_highbd_12_sub_pixel_variance64x64,
+ aom_highbd_12_sub_pixel_avg_variance64x64,
+ aom_highbd_sad64x64x3_bits12, aom_highbd_sad64x64x8_bits12,
+ aom_highbd_sad64x64x4d_bits12,
+ aom_highbd_jnt_sad64x64_avg_bits12,
+ aom_highbd_12_jnt_sub_pixel_avg_variance64x64);
+
+ HIGHBD_BFP(BLOCK_16X16, aom_highbd_sad16x16_bits12,
+ aom_highbd_sad16x16_avg_bits12, aom_highbd_12_variance16x16,
+ aom_highbd_12_sub_pixel_variance16x16,
+ aom_highbd_12_sub_pixel_avg_variance16x16,
+ aom_highbd_sad16x16x3_bits12, aom_highbd_sad16x16x8_bits12,
+ aom_highbd_sad16x16x4d_bits12,
+ aom_highbd_jnt_sad16x16_avg_bits12,
+ aom_highbd_12_jnt_sub_pixel_avg_variance16x16);
+
+ HIGHBD_BFP(BLOCK_16X8, aom_highbd_sad16x8_bits12,
+ aom_highbd_sad16x8_avg_bits12, aom_highbd_12_variance16x8,
+ aom_highbd_12_sub_pixel_variance16x8,
+ aom_highbd_12_sub_pixel_avg_variance16x8,
+ aom_highbd_sad16x8x3_bits12, aom_highbd_sad16x8x8_bits12,
+ aom_highbd_sad16x8x4d_bits12,
+ aom_highbd_jnt_sad16x8_avg_bits12,
+ aom_highbd_12_jnt_sub_pixel_avg_variance16x8);
+
+ HIGHBD_BFP(BLOCK_8X16, aom_highbd_sad8x16_bits12,
+ aom_highbd_sad8x16_avg_bits12, aom_highbd_12_variance8x16,
+ aom_highbd_12_sub_pixel_variance8x16,
+ aom_highbd_12_sub_pixel_avg_variance8x16,
+ aom_highbd_sad8x16x3_bits12, aom_highbd_sad8x16x8_bits12,
+ aom_highbd_sad8x16x4d_bits12,
+ aom_highbd_jnt_sad8x16_avg_bits12,
+ aom_highbd_12_jnt_sub_pixel_avg_variance8x16);
+
+ HIGHBD_BFP(
+ BLOCK_8X8, aom_highbd_sad8x8_bits12, aom_highbd_sad8x8_avg_bits12,
+ aom_highbd_12_variance8x8, aom_highbd_12_sub_pixel_variance8x8,
+ aom_highbd_12_sub_pixel_avg_variance8x8, aom_highbd_sad8x8x3_bits12,
+ aom_highbd_sad8x8x8_bits12, aom_highbd_sad8x8x4d_bits12,
+ aom_highbd_jnt_sad8x8_avg_bits12,
+ aom_highbd_12_jnt_sub_pixel_avg_variance8x8);
+
+ HIGHBD_BFP(BLOCK_8X4, aom_highbd_sad8x4_bits12,
+ aom_highbd_sad8x4_avg_bits12, aom_highbd_12_variance8x4,
+ aom_highbd_12_sub_pixel_variance8x4,
+ aom_highbd_12_sub_pixel_avg_variance8x4, NULL,
+ aom_highbd_sad8x4x8_bits12, aom_highbd_sad8x4x4d_bits12,
+ aom_highbd_jnt_sad8x4_avg_bits12,
+ aom_highbd_12_jnt_sub_pixel_avg_variance8x4);
+
+ HIGHBD_BFP(BLOCK_4X8, aom_highbd_sad4x8_bits12,
+ aom_highbd_sad4x8_avg_bits12, aom_highbd_12_variance4x8,
+ aom_highbd_12_sub_pixel_variance4x8,
+ aom_highbd_12_sub_pixel_avg_variance4x8, NULL,
+ aom_highbd_sad4x8x8_bits12, aom_highbd_sad4x8x4d_bits12,
+ aom_highbd_jnt_sad4x8_avg_bits12,
+ aom_highbd_12_jnt_sub_pixel_avg_variance4x8);
+
+ HIGHBD_BFP(
+ BLOCK_4X4, aom_highbd_sad4x4_bits12, aom_highbd_sad4x4_avg_bits12,
+ aom_highbd_12_variance4x4, aom_highbd_12_sub_pixel_variance4x4,
+ aom_highbd_12_sub_pixel_avg_variance4x4, aom_highbd_sad4x4x3_bits12,
+ aom_highbd_sad4x4x8_bits12, aom_highbd_sad4x4x4d_bits12,
+ aom_highbd_jnt_sad4x4_avg_bits12,
+ aom_highbd_12_jnt_sub_pixel_avg_variance4x4);
+
+ HIGHBD_BFP(BLOCK_2X2, NULL, NULL, aom_highbd_12_variance2x2, NULL, NULL,
+ NULL, NULL, NULL, NULL, NULL)
+ HIGHBD_BFP(BLOCK_4X2, NULL, NULL, aom_highbd_12_variance4x2, NULL, NULL,
+ NULL, NULL, NULL, NULL, NULL)
+ HIGHBD_BFP(BLOCK_2X4, NULL, NULL, aom_highbd_12_variance2x4, NULL, NULL,
+ NULL, NULL, NULL, NULL, NULL)
+
+#if CONFIG_EXT_PARTITION
+ HIGHBD_BFP(
+ BLOCK_128X128, aom_highbd_sad128x128_bits12,
+ aom_highbd_sad128x128_avg_bits12, aom_highbd_12_variance128x128,
+ aom_highbd_12_sub_pixel_variance128x128,
+ aom_highbd_12_sub_pixel_avg_variance128x128,
+ aom_highbd_sad128x128x3_bits12, aom_highbd_sad128x128x8_bits12,
+ aom_highbd_sad128x128x4d_bits12,
+ aom_highbd_jnt_sad128x128_avg_bits12,
+ aom_highbd_12_jnt_sub_pixel_avg_variance128x128);
+
+ HIGHBD_BFP(
+ BLOCK_128X64, aom_highbd_sad128x64_bits12,
+ aom_highbd_sad128x64_avg_bits12, aom_highbd_12_variance128x64,
+ aom_highbd_12_sub_pixel_variance128x64,
+ aom_highbd_12_sub_pixel_avg_variance128x64, NULL, NULL,
+ aom_highbd_sad128x64x4d_bits12, aom_highbd_jnt_sad128x64_avg_bits12,
+ aom_highbd_12_jnt_sub_pixel_avg_variance128x64);
+
+ HIGHBD_BFP(
+ BLOCK_64X128, aom_highbd_sad64x128_bits12,
+ aom_highbd_sad64x128_avg_bits12, aom_highbd_12_variance64x128,
+ aom_highbd_12_sub_pixel_variance64x128,
+ aom_highbd_12_sub_pixel_avg_variance64x128, NULL, NULL,
+ aom_highbd_sad64x128x4d_bits12, aom_highbd_jnt_sad64x128_avg_bits12,
+ aom_highbd_12_jnt_sub_pixel_avg_variance64x128);
+#endif // CONFIG_EXT_PARTITION
+#else // CONFIG_JNT_COMP
#if CONFIG_EXT_PARTITION_TYPES
#if CONFIG_EXT_PARTITION
HIGHBD_BFP(BLOCK_128X32, aom_highbd_sad128x32_bits12,
@@ -2210,6 +2919,7 @@
aom_highbd_12_sub_pixel_avg_variance64x128, NULL, NULL,
aom_highbd_sad64x128x4d_bits12)
#endif // CONFIG_EXT_PARTITION
+#endif // CONFIG_JNT_COMP
#if CONFIG_EXT_PARTITION
HIGHBD_MBFP(BLOCK_128X128, aom_highbd_masked_sad128x128_bits12,
diff --git a/av1/encoder/mcomp.c b/av1/encoder/mcomp.c
index e0c5ba2..5ad7127 100644
--- a/av1/encoder/mcomp.c
+++ b/av1/encoder/mcomp.c
@@ -369,12 +369,19 @@
if (second_pred != NULL) {
if (xd->cur_buf->flags & YV12_FLAG_HIGHBITDEPTH) {
DECLARE_ALIGNED(16, uint16_t, comp_pred16[MAX_SB_SQUARE]);
- if (mask)
+ if (mask) {
aom_highbd_comp_mask_pred(comp_pred16, second_pred, w, h, y + offset,
y_stride, mask, mask_stride, invert_mask);
- else
- aom_highbd_comp_avg_pred(comp_pred16, second_pred, w, h, y + offset,
- y_stride);
+ } else {
+#if CONFIG_JNT_COMP
+ if (xd->jcp_param.use_jnt_comp_avg)
+ aom_highbd_jnt_comp_avg_pred(comp_pred16, second_pred, w, h,
+ y + offset, y_stride, &xd->jcp_param);
+ else
+#endif // CONFIG_JNT_COMP
+ aom_highbd_comp_avg_pred(comp_pred16, second_pred, w, h, y + offset,
+ y_stride);
+ }
besterr =
vfp->vf(CONVERT_TO_BYTEPTR(comp_pred16), w, src, src_stride, sse1);
} else {
@@ -685,14 +692,22 @@
if (xd->cur_buf->flags & YV12_FLAG_HIGHBITDEPTH) {
DECLARE_ALIGNED(16, uint16_t, pred16[MAX_SB_SQUARE]);
if (second_pred != NULL) {
- if (mask)
+ if (mask) {
aom_highbd_comp_mask_upsampled_pred(
pred16, second_pred, w, h, subpel_x_q3, subpel_y_q3, y, y_stride,
mask, mask_stride, invert_mask, xd->bd);
- else
- aom_highbd_comp_avg_upsampled_pred(pred16, second_pred, w, h,
- subpel_x_q3, subpel_y_q3, y,
- y_stride, xd->bd);
+ } else {
+#if CONFIG_JNT_COMP
+ if (xd->jcp_param.use_jnt_comp_avg)
+ aom_highbd_jnt_comp_avg_upsampled_pred(
+ pred16, second_pred, w, h, subpel_x_q3, subpel_y_q3, y, y_stride,
+ xd->bd, &xd->jcp_param);
+ else
+#endif // CONFIG_JNT_COMP
+ aom_highbd_comp_avg_upsampled_pred(pred16, second_pred, w, h,
+ subpel_x_q3, subpel_y_q3, y,
+ y_stride, xd->bd);
+ }
} else {
aom_highbd_upsampled_pred(pred16, w, h, subpel_x_q3, subpel_y_q3, y,
y_stride, xd->bd);
