vpx/vp9/common/x86/vp9_subpixel_variance_sse2.c
Yunqing Wang 8d568312a2 vp9_sub_pixel_variance16x2 SSE2 optimization
About 5% decoder speedup.

Change-Id: Ib6687d337af758a536a0e7e289f400990f1f9794
2013-01-08 12:01:55 -08:00

91 lines
3.8 KiB
C

/*
* Copyright (c) 2010 The WebM project authors. All Rights Reserved.
*
* Use of this source code is governed by a BSD-style license
* that can be found in the LICENSE file in the root of the source
* tree. An additional intellectual property rights grant can be found
* in the file PATENTS. All contributing project authors may
* be found in the AUTHORS file in the root of the source tree.
*/
#define HALFNDX 8
void vp9_half_horiz_variance16x_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 vp9_half_vert_variance16x_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 vp9_half_horiz_vert_variance16x_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 vp9_filter_block2d_bil_var_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 xoffset,
int yoffset,
int *sum,
unsigned int *sumsquared);
unsigned int vp9_sub_pixel_variance16x2_sse2(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) {
int xsum0, xsum1;
unsigned int xxsum0, xxsum1;
if (xoffset == HALFNDX && yoffset == 0) {
vp9_half_horiz_variance16x_h_sse2(
src_ptr, src_pixels_per_line,
dst_ptr, dst_pixels_per_line, 2,
&xsum0, &xxsum0);
} else if (xoffset == 0 && yoffset == HALFNDX) {
vp9_half_vert_variance16x_h_sse2(
src_ptr, src_pixels_per_line,
dst_ptr, dst_pixels_per_line, 2,
&xsum0, &xxsum0);
} else if (xoffset == HALFNDX && yoffset == HALFNDX) {
vp9_half_horiz_vert_variance16x_h_sse2(
src_ptr, src_pixels_per_line,
dst_ptr, dst_pixels_per_line, 2,
&xsum0, &xxsum0);
} else {
vp9_filter_block2d_bil_var_sse2(
src_ptr, src_pixels_per_line,
dst_ptr, dst_pixels_per_line, 2,
xoffset, yoffset,
&xsum0, &xxsum0);
vp9_filter_block2d_bil_var_sse2(
src_ptr + 8, src_pixels_per_line,
dst_ptr + 8, dst_pixels_per_line, 2,
xoffset, yoffset,
&xsum1, &xxsum1);
xsum0 += xsum1;
xxsum0 += xxsum1;
}
*sse = xxsum0;
return (xxsum0 - (((unsigned int)xsum0 * xsum0) >> 5));
}