summaryrefslogtreecommitdiff
path: root/vp8/encoder/x86/variance_sse2.c
diff options
context:
space:
mode:
Diffstat (limited to 'vp8/encoder/x86/variance_sse2.c')
-rw-r--r--vp8/encoder/x86/variance_sse2.c191
1 files changed, 46 insertions, 145 deletions
diff --git a/vp8/encoder/x86/variance_sse2.c b/vp8/encoder/x86/variance_sse2.c
index 7cf6a6308..0edda3062 100644
--- a/vp8/encoder/x86/variance_sse2.c
+++ b/vp8/encoder/x86/variance_sse2.c
@@ -81,6 +81,16 @@ void vp8_filter_block2d_bil_var_sse2
int *sum,
unsigned int *sumsquared
);
+void vp8_half_horiz_vert_variance8x_h_sse2
+(
+ const unsigned char *ref_ptr,
+ int ref_pixels_per_line,
+ const unsigned char *src_ptr,
+ int src_pixels_per_line,
+ unsigned int Height,
+ int *sum,
+ unsigned int *sumsquared
+);
void vp8_half_horiz_vert_variance16x_h_sse2
(
const unsigned char *ref_ptr,
@@ -91,6 +101,16 @@ void vp8_half_horiz_vert_variance16x_h_sse2
int *sum,
unsigned int *sumsquared
);
+void vp8_half_horiz_variance8x_h_sse2
+(
+ const unsigned char *ref_ptr,
+ int ref_pixels_per_line,
+ const unsigned char *src_ptr,
+ int src_pixels_per_line,
+ unsigned int Height,
+ int *sum,
+ unsigned int *sumsquared
+);
void vp8_half_horiz_variance16x_h_sse2
(
const unsigned char *ref_ptr,
@@ -101,6 +121,16 @@ void vp8_half_horiz_variance16x_h_sse2
int *sum,
unsigned int *sumsquared
);
+void vp8_half_vert_variance8x_h_sse2
+(
+ const unsigned char *ref_ptr,
+ int ref_pixels_per_line,
+ const unsigned char *src_ptr,
+ int src_pixels_per_line,
+ unsigned int Height,
+ int *sum,
+ unsigned int *sumsquared
+);
void vp8_half_vert_variance16x_h_sse2
(
const unsigned char *ref_ptr,
@@ -262,21 +292,21 @@ unsigned int vp8_sub_pixel_variance8x8_wmt
if (xoffset == 4 && yoffset == 0)
{
- vp8_half_horiz_variance16x_h_sse2(
+ vp8_half_horiz_variance8x_h_sse2(
src_ptr, src_pixels_per_line,
dst_ptr, dst_pixels_per_line, 8,
&xsum, &xxsum);
}
else if (xoffset == 0 && yoffset == 4)
{
- vp8_half_vert_variance16x_h_sse2(
+ vp8_half_vert_variance8x_h_sse2(
src_ptr, src_pixels_per_line,
dst_ptr, dst_pixels_per_line, 8,
&xsum, &xxsum);
}
else if (xoffset == 4 && yoffset == 4)
{
- vp8_half_horiz_vert_variance16x_h_sse2(
+ vp8_half_horiz_vert_variance8x_h_sse2(
src_ptr, src_pixels_per_line,
dst_ptr, dst_pixels_per_line, 8,
&xsum, &xxsum);
@@ -317,11 +347,6 @@ unsigned int vp8_sub_pixel_variance16x16_wmt
src_ptr, src_pixels_per_line,
dst_ptr, dst_pixels_per_line, 16,
&xsum0, &xxsum0);
-
- vp8_half_horiz_variance16x_h_sse2(
- src_ptr + 8, src_pixels_per_line,
- dst_ptr + 8, dst_pixels_per_line, 16,
- &xsum1, &xxsum1);
}
else if (xoffset == 0 && yoffset == 4)
{
@@ -329,11 +354,6 @@ unsigned int vp8_sub_pixel_variance16x16_wmt
src_ptr, src_pixels_per_line,
dst_ptr, dst_pixels_per_line, 16,
&xsum0, &xxsum0);
-
- vp8_half_vert_variance16x_h_sse2(
- src_ptr + 8, src_pixels_per_line,
- dst_ptr + 8, dst_pixels_per_line, 16,
- &xsum1, &xxsum1);
}
else if (xoffset == 4 && yoffset == 4)
{
@@ -341,11 +361,6 @@ unsigned int vp8_sub_pixel_variance16x16_wmt
src_ptr, src_pixels_per_line,
dst_ptr, dst_pixels_per_line, 16,
&xsum0, &xxsum0);
-
- vp8_half_horiz_vert_variance16x_h_sse2(
- src_ptr + 8, src_pixels_per_line,
- dst_ptr + 8, dst_pixels_per_line, 16,
- &xsum1, &xxsum1);
}
else
{
@@ -356,17 +371,16 @@ unsigned int vp8_sub_pixel_variance16x16_wmt
&xsum0, &xxsum0
);
-
vp8_filter_block2d_bil_var_sse2(
src_ptr + 8, src_pixels_per_line,
dst_ptr + 8, dst_pixels_per_line, 16,
xoffset, yoffset,
&xsum1, &xxsum1
);
+ xsum0 += xsum1;
+ xxsum0 += xxsum1;
}
- xsum0 += xsum1;
- xxsum0 += xxsum1;
*sse = xxsum0;
return (xxsum0 - ((xsum0 * xsum0) >> 8));
}
@@ -406,11 +420,6 @@ unsigned int vp8_sub_pixel_variance16x8_wmt
src_ptr, src_pixels_per_line,
dst_ptr, dst_pixels_per_line, 8,
&xsum0, &xxsum0);
-
- vp8_half_horiz_variance16x_h_sse2(
- src_ptr + 8, src_pixels_per_line,
- dst_ptr + 8, dst_pixels_per_line, 8,
- &xsum1, &xxsum1);
}
else if (xoffset == 0 && yoffset == 4)
{
@@ -418,11 +427,6 @@ unsigned int vp8_sub_pixel_variance16x8_wmt
src_ptr, src_pixels_per_line,
dst_ptr, dst_pixels_per_line, 8,
&xsum0, &xxsum0);
-
- vp8_half_vert_variance16x_h_sse2(
- src_ptr + 8, src_pixels_per_line,
- dst_ptr + 8, dst_pixels_per_line, 8,
- &xsum1, &xxsum1);
}
else if (xoffset == 4 && yoffset == 4)
{
@@ -430,11 +434,6 @@ unsigned int vp8_sub_pixel_variance16x8_wmt
src_ptr, src_pixels_per_line,
dst_ptr, dst_pixels_per_line, 8,
&xsum0, &xxsum0);
-
- vp8_half_horiz_vert_variance16x_h_sse2(
- src_ptr + 8, src_pixels_per_line,
- dst_ptr + 8, dst_pixels_per_line, 8,
- &xsum1, &xxsum1);
}
else
{
@@ -449,11 +448,10 @@ unsigned int vp8_sub_pixel_variance16x8_wmt
dst_ptr + 8, dst_pixels_per_line, 8,
xoffset, yoffset,
&xsum1, &xxsum1);
+ xsum0 += xsum1;
+ xxsum0 += xxsum1;
}
- xsum0 += xsum1;
- xxsum0 += xxsum1;
-
*sse = xxsum0;
return (xxsum0 - ((xsum0 * xsum0) >> 7));
}
@@ -474,21 +472,21 @@ unsigned int vp8_sub_pixel_variance8x16_wmt
if (xoffset == 4 && yoffset == 0)
{
- vp8_half_horiz_variance16x_h_sse2(
+ vp8_half_horiz_variance8x_h_sse2(
src_ptr, src_pixels_per_line,
dst_ptr, dst_pixels_per_line, 16,
&xsum, &xxsum);
}
else if (xoffset == 0 && yoffset == 4)
{
- vp8_half_vert_variance16x_h_sse2(
+ vp8_half_vert_variance8x_h_sse2(
src_ptr, src_pixels_per_line,
dst_ptr, dst_pixels_per_line, 16,
&xsum, &xxsum);
}
else if (xoffset == 4 && yoffset == 4)
{
- vp8_half_horiz_vert_variance16x_h_sse2(
+ vp8_half_horiz_vert_variance8x_h_sse2(
src_ptr, src_pixels_per_line,
dst_ptr, dst_pixels_per_line, 16,
&xsum, &xxsum);
@@ -506,81 +504,6 @@ unsigned int vp8_sub_pixel_variance8x16_wmt
return (xxsum - ((xsum * xsum) >> 7));
}
-unsigned int vp8_i_variance16x16_wmt(
- const unsigned char *src_ptr,
- int source_stride,
- const unsigned char *ref_ptr,
- int recon_stride,
- unsigned int *sse)
-{
- unsigned int sse0, sse1, sse2, sse3, var;
- int sum0, sum1, sum2, sum3, avg;
-
-
- vp8_get8x8var_sse2(src_ptr, source_stride, ref_ptr, recon_stride, &sse0, &sum0) ;
- vp8_get8x8var_sse2(src_ptr + 8, source_stride, ref_ptr + 8, recon_stride, &sse1, &sum1);
- vp8_get8x8var_sse2(src_ptr + (source_stride >> 1), source_stride, ref_ptr + (recon_stride >> 1), recon_stride, &sse2, &sum2) ;
- vp8_get8x8var_sse2(src_ptr + (source_stride >> 1) + 8, source_stride, ref_ptr + (recon_stride >> 1) + 8, recon_stride, &sse3, &sum3);
-
- var = sse0 + sse1 + sse2 + sse3;
- avg = sum0 + sum1 + sum2 + sum3;
-
- *sse = var;
- return (var - ((avg * avg) >> 8));
-
-}
-
-unsigned int vp8_i_variance8x16_wmt(
- const unsigned char *src_ptr,
- int source_stride,
- const unsigned char *ref_ptr,
- int recon_stride,
- unsigned int *sse)
-{
- unsigned int sse0, sse1, var;
- int sum0, sum1, avg;
- vp8_get8x8var_sse2(src_ptr, source_stride, ref_ptr, recon_stride, &sse0, &sum0) ;
- vp8_get8x8var_sse2(src_ptr + (source_stride >> 1), source_stride, ref_ptr + (recon_stride >> 1), recon_stride, &sse1, &sum1) ;
-
- var = sse0 + sse1;
- avg = sum0 + sum1;
-
- *sse = var;
- return (var - ((avg * avg) >> 7));
-
-}
-
-
-unsigned int vp8_i_sub_pixel_variance16x16_wmt
-(
- const unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- const unsigned char *dst_ptr,
- int dst_pixels_per_line,
- unsigned int *sse
-)
-{
- return vp8_sub_pixel_variance16x16_wmt(src_ptr, (src_pixels_per_line >> 1), xoffset, yoffset, dst_ptr, (dst_pixels_per_line >> 1), sse);
-}
-
-
-unsigned int vp8_i_sub_pixel_variance8x16_wmt
-(
- const unsigned char *src_ptr,
- int src_pixels_per_line,
- int xoffset,
- int yoffset,
- const unsigned char *dst_ptr,
- int dst_pixels_per_line,
- unsigned int *sse
-)
-{
-
- return vp8_sub_pixel_variance8x16_wmt(src_ptr, (src_pixels_per_line >> 1), xoffset, yoffset, dst_ptr, (dst_pixels_per_line >> 1), sse);
-}
-
unsigned int vp8_variance_halfpixvar16x16_h_wmt(
const unsigned char *src_ptr,
@@ -589,21 +512,14 @@ unsigned int vp8_variance_halfpixvar16x16_h_wmt(
int dst_pixels_per_line,
unsigned int *sse)
{
- int xsum0, xsum1;
- unsigned int xxsum0, xxsum1;
+ int xsum0;
+ unsigned int xxsum0;
vp8_half_horiz_variance16x_h_sse2(
src_ptr, src_pixels_per_line,
dst_ptr, dst_pixels_per_line, 16,
&xsum0, &xxsum0);
- vp8_half_horiz_variance16x_h_sse2(
- src_ptr + 8, src_pixels_per_line,
- dst_ptr + 8, dst_pixels_per_line, 16,
- &xsum1, &xxsum1);
-
- xsum0 += xsum1;
- xxsum0 += xxsum1;
*sse = xxsum0;
return (xxsum0 - ((xsum0 * xsum0) >> 8));
}
@@ -616,21 +532,13 @@ unsigned int vp8_variance_halfpixvar16x16_v_wmt(
int dst_pixels_per_line,
unsigned int *sse)
{
- int xsum0, xsum1;
- unsigned int xxsum0, xxsum1;
-
+ int xsum0;
+ unsigned int xxsum0;
vp8_half_vert_variance16x_h_sse2(
src_ptr, src_pixels_per_line,
dst_ptr, dst_pixels_per_line, 16,
&xsum0, &xxsum0);
- vp8_half_vert_variance16x_h_sse2(
- src_ptr + 8, src_pixels_per_line,
- dst_ptr + 8, dst_pixels_per_line, 16,
- &xsum1, &xxsum1);
-
- xsum0 += xsum1;
- xxsum0 += xxsum1;
*sse = xxsum0;
return (xxsum0 - ((xsum0 * xsum0) >> 8));
}
@@ -643,21 +551,14 @@ unsigned int vp8_variance_halfpixvar16x16_hv_wmt(
int dst_pixels_per_line,
unsigned int *sse)
{
- int xsum0, xsum1;
- unsigned int xxsum0, xxsum1;
+ int xsum0;
+ unsigned int xxsum0;
vp8_half_horiz_vert_variance16x_h_sse2(
src_ptr, src_pixels_per_line,
dst_ptr, dst_pixels_per_line, 16,
&xsum0, &xxsum0);
- vp8_half_horiz_vert_variance16x_h_sse2(
- src_ptr + 8, src_pixels_per_line,
- dst_ptr + 8, dst_pixels_per_line, 16,
- &xsum1, &xxsum1);
-
- xsum0 += xsum1;
- xxsum0 += xxsum1;
*sse = xxsum0;
return (xxsum0 - ((xsum0 * xsum0) >> 8));
}
generated by cgit v1.2.3 (git 2.25.1) at 2024-10-13 19:15:58 +0000