Some cosmetic improvements since HBD variance 4x4 optimization
Change-Id: I414c1fabd2e3a9b1d9daa8a90f85a0bace8bd3cd
diff --git a/vp10/encoder/mcomp.c b/vp10/encoder/mcomp.c
index 0c8ec43..4327d97 100644
--- a/vp10/encoder/mcomp.c
+++ b/vp10/encoder/mcomp.c
@@ -367,8 +367,8 @@
if (second_pred != NULL) {
if (xd->cur_buf->flags & YV12_FLAG_HIGHBITDEPTH) {
DECLARE_ALIGNED(16, uint16_t, comp_pred16[MAX_SB_SQUARE]);
- vpx_highbd_comp_avg_pred_c(comp_pred16, second_pred, w, h, y + offset,
- y_stride);
+ vpx_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 {
diff --git a/vpx_dsp/variance.c b/vpx_dsp/variance.c
index 90c8bed..e6be1dd 100644
--- a/vpx_dsp/variance.c
+++ b/vpx_dsp/variance.c
@@ -433,7 +433,7 @@
return *sse; \
}
-void highbd_var_filter_block2d_bil_first_pass(
+void vpx_highbd_var_filter_block2d_bil_first_pass(
const uint8_t *src_ptr8,
uint16_t *output_ptr,
unsigned int src_pixels_per_line,
@@ -459,7 +459,7 @@
}
}
-void highbd_var_filter_block2d_bil_second_pass(
+void vpx_highbd_var_filter_block2d_bil_second_pass(
const uint16_t *src_ptr,
uint16_t *output_ptr,
unsigned int src_pixels_per_line,
@@ -492,13 +492,14 @@
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_2t[xoffset]); \
- highbd_var_filter_block2d_bil_second_pass(fdata3, temp2, W, W, H, W, \
- bilinear_filters_2t[yoffset]); \
+ vpx_highbd_var_filter_block2d_bil_first_pass( \
+ src, fdata3, src_stride, 1, H + 1, \
+ W, bilinear_filters_2t[xoffset]); \
+ vpx_highbd_var_filter_block2d_bil_second_pass(fdata3, temp2, W, W, H, W, \
+ bilinear_filters_2t[yoffset]); \
\
return vpx_highbd_8_variance##W##x##H##_c(CONVERT_TO_BYTEPTR(temp2), W, dst, \
- dst_stride, sse); \
+ dst_stride, sse); \
} \
\
uint32_t vpx_highbd_10_sub_pixel_variance##W##x##H##_c( \
@@ -509,10 +510,11 @@
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_2t[xoffset]); \
- highbd_var_filter_block2d_bil_second_pass(fdata3, temp2, W, W, H, W, \
- bilinear_filters_2t[yoffset]); \
+ vpx_highbd_var_filter_block2d_bil_first_pass( \
+ src, fdata3, src_stride, 1, H + 1, \
+ W, bilinear_filters_2t[xoffset]); \
+ vpx_highbd_var_filter_block2d_bil_second_pass(fdata3, temp2, W, W, H, W, \
+ bilinear_filters_2t[yoffset]); \
\
return vpx_highbd_10_variance##W##x##H##_c(CONVERT_TO_BYTEPTR(temp2), \
W, dst, dst_stride, sse); \
@@ -526,10 +528,11 @@
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_2t[xoffset]); \
- highbd_var_filter_block2d_bil_second_pass(fdata3, temp2, W, W, H, W, \
- bilinear_filters_2t[yoffset]); \
+ vpx_highbd_var_filter_block2d_bil_first_pass( \
+ src, fdata3, src_stride, 1, H + 1, \
+ W, bilinear_filters_2t[xoffset]); \
+ vpx_highbd_var_filter_block2d_bil_second_pass(fdata3, temp2, W, W, H, W, \
+ bilinear_filters_2t[yoffset]); \
\
return vpx_highbd_12_variance##W##x##H##_c(CONVERT_TO_BYTEPTR(temp2), \
W, dst, dst_stride, sse); \
@@ -546,16 +549,17 @@
uint16_t temp2[H * W]; \
DECLARE_ALIGNED(16, uint16_t, temp3[H * W]); \
\
- highbd_var_filter_block2d_bil_first_pass(src, fdata3, src_stride, 1, H + 1, \
- W, bilinear_filters_2t[xoffset]); \
- highbd_var_filter_block2d_bil_second_pass(fdata3, temp2, W, W, H, W, \
- bilinear_filters_2t[yoffset]); \
+ vpx_highbd_var_filter_block2d_bil_first_pass( \
+ src, fdata3, src_stride, 1, H + 1, \
+ W, bilinear_filters_2t[xoffset]); \
+ vpx_highbd_var_filter_block2d_bil_second_pass(fdata3, temp2, W, W, H, W, \
+ bilinear_filters_2t[yoffset]); \
\
vpx_highbd_comp_avg_pred_c(temp3, second_pred, W, H, \
CONVERT_TO_BYTEPTR(temp2), W); \
\
return vpx_highbd_8_variance##W##x##H##_c(CONVERT_TO_BYTEPTR(temp3), W, dst, \
- dst_stride, sse); \
+ dst_stride, sse); \
} \
\
uint32_t vpx_highbd_10_sub_pixel_avg_variance##W##x##H##_c( \
@@ -568,10 +572,11 @@
uint16_t temp2[H * W]; \
DECLARE_ALIGNED(16, uint16_t, temp3[H * W]); \
\
- highbd_var_filter_block2d_bil_first_pass(src, fdata3, src_stride, 1, H + 1, \
- W, bilinear_filters_2t[xoffset]); \
- highbd_var_filter_block2d_bil_second_pass(fdata3, temp2, W, W, H, W, \
- bilinear_filters_2t[yoffset]); \
+ vpx_highbd_var_filter_block2d_bil_first_pass( \
+ src, fdata3, src_stride, 1, H + 1, \
+ W, bilinear_filters_2t[xoffset]); \
+ vpx_highbd_var_filter_block2d_bil_second_pass(fdata3, temp2, W, W, H, W, \
+ bilinear_filters_2t[yoffset]); \
\
vpx_highbd_comp_avg_pred_c(temp3, second_pred, W, H, \
CONVERT_TO_BYTEPTR(temp2), W); \
@@ -590,10 +595,11 @@
uint16_t temp2[H * W]; \
DECLARE_ALIGNED(16, uint16_t, temp3[H * W]); \
\
- highbd_var_filter_block2d_bil_first_pass(src, fdata3, src_stride, 1, H + 1, \
- W, bilinear_filters_2t[xoffset]); \
- highbd_var_filter_block2d_bil_second_pass(fdata3, temp2, W, W, H, W, \
- bilinear_filters_2t[yoffset]); \
+ vpx_highbd_var_filter_block2d_bil_first_pass( \
+ src, fdata3, src_stride, 1, H + 1, \
+ W, bilinear_filters_2t[xoffset]); \
+ vpx_highbd_var_filter_block2d_bil_second_pass(fdata3, temp2, W, W, H, W, \
+ bilinear_filters_2t[yoffset]); \
\
vpx_highbd_comp_avg_pred_c(temp3, second_pred, W, H, \
CONVERT_TO_BYTEPTR(temp2), W); \
@@ -914,11 +920,11 @@
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_2t[xoffset]); \
- highbd_var_filter_block2d_bil_second_pass(fdata3, temp2, W, W, H, W, \
- bilinear_filters_2t[yoffset]); \
+ vpx_highbd_var_filter_block2d_bil_first_pass(src, fdata3, src_stride, 1, \
+ H + 1, W, \
+ bilinear_filters_2t[xoffset]); \
+ vpx_highbd_var_filter_block2d_bil_second_pass(fdata3, temp2, W, W, H, W, \
+ bilinear_filters_2t[yoffset]); \
\
return vpx_highbd_masked_variance##W##x##H##_c(CONVERT_TO_BYTEPTR(temp2), \
W, dst, dst_stride, \
@@ -934,11 +940,11 @@
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_2t[xoffset]); \
- highbd_var_filter_block2d_bil_second_pass(fdata3, temp2, W, W, H, W, \
- bilinear_filters_2t[yoffset]); \
+ vpx_highbd_var_filter_block2d_bil_first_pass(src, fdata3, src_stride, 1, \
+ H + 1, W, \
+ bilinear_filters_2t[xoffset]); \
+ vpx_highbd_var_filter_block2d_bil_second_pass(fdata3, temp2, W, W, H, W, \
+ bilinear_filters_2t[yoffset]); \
\
return vpx_highbd_10_masked_variance##W##x##H##_c(CONVERT_TO_BYTEPTR(temp2), \
W, dst, dst_stride, \
@@ -954,11 +960,11 @@
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_2t[xoffset]); \
- highbd_var_filter_block2d_bil_second_pass(fdata3, temp2, W, W, H, W, \
- bilinear_filters_2t[yoffset]); \
+ vpx_highbd_var_filter_block2d_bil_first_pass(src, fdata3, src_stride, 1, \
+ H + 1, W, \
+ bilinear_filters_2t[xoffset]); \
+ vpx_highbd_var_filter_block2d_bil_second_pass(fdata3, temp2, W, W, H, W, \
+ bilinear_filters_2t[yoffset]); \
\
return vpx_highbd_12_masked_variance##W##x##H##_c(CONVERT_TO_BYTEPTR(temp2), \
W, dst, dst_stride, \
diff --git a/vpx_dsp/variance.h b/vpx_dsp/variance.h
index 4ad23f8..1759854 100644
--- a/vpx_dsp/variance.h
+++ b/vpx_dsp/variance.h
@@ -130,7 +130,7 @@
} vp10_variance_fn_ptr_t;
#endif // CONFIG_VP10
-void highbd_var_filter_block2d_bil_first_pass(
+void vpx_highbd_var_filter_block2d_bil_first_pass(
const uint8_t *src_ptr8,
uint16_t *output_ptr,
unsigned int src_pixels_per_line,
@@ -139,7 +139,7 @@
unsigned int output_width,
const uint8_t *filter);
-void highbd_var_filter_block2d_bil_second_pass(
+void vpx_highbd_var_filter_block2d_bil_second_pass(
const uint16_t *src_ptr,
uint16_t *output_ptr,
unsigned int src_pixels_per_line,
diff --git a/vpx_dsp/x86/highbd_variance_sse4.c b/vpx_dsp/x86/highbd_variance_sse4.c
index 18ecc7e..5c1dfe4 100644
--- a/vpx_dsp/x86/highbd_variance_sse4.c
+++ b/vpx_dsp/x86/highbd_variance_sse4.c
@@ -119,10 +119,12 @@
uint16_t fdata3[(4 + 1) * 4];
uint16_t temp2[4 * 4];
- highbd_var_filter_block2d_bil_first_pass(src, fdata3, src_stride, 1, 4 + 1,
- 4, bilinear_filters_2t[xoffset]);
- highbd_var_filter_block2d_bil_second_pass(fdata3, temp2, 4, 4, 4, 4,
- bilinear_filters_2t[yoffset]);
+ vpx_highbd_var_filter_block2d_bil_first_pass(
+ src, fdata3, src_stride, 1, 4 + 1,
+ 4, bilinear_filters_2t[xoffset]);
+ vpx_highbd_var_filter_block2d_bil_second_pass(
+ fdata3, temp2, 4, 4, 4, 4,
+ bilinear_filters_2t[yoffset]);
return vpx_highbd_8_variance4x4(CONVERT_TO_BYTEPTR(temp2),
4, dst, dst_stride, sse);
@@ -137,10 +139,12 @@
uint16_t fdata3[(4 + 1) * 4];
uint16_t temp2[4 * 4];
- highbd_var_filter_block2d_bil_first_pass(src, fdata3, src_stride, 1, 4 + 1,
- 4, bilinear_filters_2t[xoffset]);
- highbd_var_filter_block2d_bil_second_pass(fdata3, temp2, 4, 4, 4, 4,
- bilinear_filters_2t[yoffset]);
+ vpx_highbd_var_filter_block2d_bil_first_pass(
+ src, fdata3, src_stride, 1, 4 + 1,
+ 4, bilinear_filters_2t[xoffset]);
+ vpx_highbd_var_filter_block2d_bil_second_pass(
+ fdata3, temp2, 4, 4, 4, 4,
+ bilinear_filters_2t[yoffset]);
return vpx_highbd_10_variance4x4(CONVERT_TO_BYTEPTR(temp2),
4, dst, dst_stride, sse);
@@ -155,10 +159,12 @@
uint16_t fdata3[(4 + 1) * 4];
uint16_t temp2[4 * 4];
- highbd_var_filter_block2d_bil_first_pass(src, fdata3, src_stride, 1, 4 + 1,
- 4, bilinear_filters_2t[xoffset]);
- highbd_var_filter_block2d_bil_second_pass(fdata3, temp2, 4, 4, 4, 4,
- bilinear_filters_2t[yoffset]);
+ vpx_highbd_var_filter_block2d_bil_first_pass(
+ src, fdata3, src_stride, 1, 4 + 1,
+ 4, bilinear_filters_2t[xoffset]);
+ vpx_highbd_var_filter_block2d_bil_second_pass(
+ fdata3, temp2, 4, 4, 4, 4,
+ bilinear_filters_2t[yoffset]);
return vpx_highbd_12_variance4x4(CONVERT_TO_BYTEPTR(temp2),
4, dst, dst_stride, sse);
@@ -177,13 +183,15 @@
uint16_t temp2[4 * 4];
DECLARE_ALIGNED(16, uint16_t, temp3[4 * 4]);
- highbd_var_filter_block2d_bil_first_pass(src, fdata3, src_stride, 1, 4 + 1,
- 4, bilinear_filters_2t[xoffset]);
- highbd_var_filter_block2d_bil_second_pass(fdata3, temp2, 4, 4, 4, 4,
- bilinear_filters_2t[yoffset]);
+ vpx_highbd_var_filter_block2d_bil_first_pass(
+ src, fdata3, src_stride, 1, 4 + 1,
+ 4, bilinear_filters_2t[xoffset]);
+ vpx_highbd_var_filter_block2d_bil_second_pass(
+ fdata3, temp2, 4, 4, 4, 4,
+ bilinear_filters_2t[yoffset]);
- vpx_highbd_comp_avg_pred_c(temp3, second_pred, 4, 4,
- CONVERT_TO_BYTEPTR(temp2), 4);
+ vpx_highbd_comp_avg_pred(temp3, second_pred, 4, 4,
+ CONVERT_TO_BYTEPTR(temp2), 4);
return vpx_highbd_8_variance4x4(CONVERT_TO_BYTEPTR(temp3),
4, dst, dst_stride, sse);
@@ -200,13 +208,15 @@
uint16_t temp2[4 * 4];
DECLARE_ALIGNED(16, uint16_t, temp3[4 * 4]);
- highbd_var_filter_block2d_bil_first_pass(src, fdata3, src_stride, 1, 4 + 1,
- 4, bilinear_filters_2t[xoffset]);
- highbd_var_filter_block2d_bil_second_pass(fdata3, temp2, 4, 4, 4, 4,
- bilinear_filters_2t[yoffset]);
+ vpx_highbd_var_filter_block2d_bil_first_pass(
+ src, fdata3, src_stride, 1, 4 + 1,
+ 4, bilinear_filters_2t[xoffset]);
+ vpx_highbd_var_filter_block2d_bil_second_pass(
+ fdata3, temp2, 4, 4, 4, 4,
+ bilinear_filters_2t[yoffset]);
- vpx_highbd_comp_avg_pred_c(temp3, second_pred, 4, 4,
- CONVERT_TO_BYTEPTR(temp2), 4);
+ vpx_highbd_comp_avg_pred(temp3, second_pred, 4, 4,
+ CONVERT_TO_BYTEPTR(temp2), 4);
return vpx_highbd_10_variance4x4(CONVERT_TO_BYTEPTR(temp3),
4, dst, dst_stride, sse);
@@ -223,13 +233,15 @@
uint16_t temp2[4 * 4];
DECLARE_ALIGNED(16, uint16_t, temp3[4 * 4]);
- highbd_var_filter_block2d_bil_first_pass(src, fdata3, src_stride, 1, 4 + 1,
- 4, bilinear_filters_2t[xoffset]);
- highbd_var_filter_block2d_bil_second_pass(fdata3, temp2, 4, 4, 4, 4,
- bilinear_filters_2t[yoffset]);
+ vpx_highbd_var_filter_block2d_bil_first_pass(
+ src, fdata3, src_stride, 1, 4 + 1,
+ 4, bilinear_filters_2t[xoffset]);
+ vpx_highbd_var_filter_block2d_bil_second_pass(
+ fdata3, temp2, 4, 4, 4, 4,
+ bilinear_filters_2t[yoffset]);
- vpx_highbd_comp_avg_pred_c(temp3, second_pred, 4, 4,
- CONVERT_TO_BYTEPTR(temp2), 4);
+ vpx_highbd_comp_avg_pred(temp3, second_pred, 4, 4,
+ CONVERT_TO_BYTEPTR(temp2), 4);
return vpx_highbd_12_variance4x4(CONVERT_TO_BYTEPTR(temp3),
4, dst, dst_stride, sse);
