vp8/encoder/x86/variance_ssse3.c

   1 /*
   2  *  Copyright (c) 2010 The WebM project authors. All Rights Reserved.
   3  *
   4  *  Use of this source code is governed by a BSD-style license
   5  *  that can be found in the LICENSE file in the root of the source
   6  *  tree. An additional intellectual property rights grant can be found
   7  *  in the file PATENTS.  All contributing project authors may
   8  *  be found in the AUTHORS file in the root of the source tree.
   9  */
  10
  11
  12 #include "vp8/encoder/variance.h"
  13 #include "vp8/common/pragmas.h"
  14 #include "vpx_ports/mem.h"
  15
  16 extern unsigned int vp8_get16x16var_sse2
  17 (
  18     const unsigned char *src_ptr,
  19     int source_stride,
  20     const unsigned char *ref_ptr,
  21     int recon_stride,
  22     unsigned int *SSE,
  23     int *Sum
  24 );
  25 extern void vp8_half_horiz_vert_variance16x_h_sse2
  26 (
  27     const unsigned char *ref_ptr,
  28     int ref_pixels_per_line,
  29     const unsigned char *src_ptr,
  30     int src_pixels_per_line,
  31     unsigned int Height,
  32     int *sum,
  33     unsigned int *sumsquared
  34 );
  35 extern void vp8_half_horiz_variance16x_h_sse2
  36 (
  37     const unsigned char *ref_ptr,
  38     int ref_pixels_per_line,
  39     const unsigned char *src_ptr,
  40     int src_pixels_per_line,
  41     unsigned int Height,
  42     int *sum,
  43     unsigned int *sumsquared
  44 );
  45 extern void vp8_half_vert_variance16x_h_sse2
  46 (
  47     const unsigned char *ref_ptr,
  48     int ref_pixels_per_line,
  49     const unsigned char *src_ptr,
  50     int src_pixels_per_line,
  51     unsigned int Height,
  52     int *sum,
  53     unsigned int *sumsquared
  54 );
  55 extern void vp8_filter_block2d_bil_var_ssse3
  56 (
  57     const unsigned char *ref_ptr,
  58     int ref_pixels_per_line,
  59     const unsigned char *src_ptr,
  60     int src_pixels_per_line,
  61     unsigned int Height,
  62     int  xoffset,
  63     int  yoffset,
  64     int *sum,
  65     unsigned int *sumsquared
  66 );
  67
  68 unsigned int vp8_sub_pixel_variance16x16_ssse3
  69 (
  70     const unsigned char  *src_ptr,
  71     int  src_pixels_per_line,
  72     int  xoffset,
  73     int  yoffset,
  74     const unsigned char *dst_ptr,
  75     int dst_pixels_per_line,
  76     unsigned int *sse
  77 )
  78 {
  79     int xsum0;
  80     unsigned int xxsum0;
  81
  82     // note we could avoid these if statements if the calling function
  83     // just called the appropriate functions inside.
  84     if (xoffset == 4 && yoffset == 0)
  85     {
  86         vp8_half_horiz_variance16x_h_sse2(
  87             src_ptr, src_pixels_per_line,
  88             dst_ptr, dst_pixels_per_line, 16,
  89             &xsum0, &xxsum0);
  90     }
  91     else if (xoffset == 0 && yoffset == 4)
  92     {
  93         vp8_half_vert_variance16x_h_sse2(
  94             src_ptr, src_pixels_per_line,
  95             dst_ptr, dst_pixels_per_line, 16,
  96             &xsum0, &xxsum0);
  97     }
  98     else if (xoffset == 4 && yoffset == 4)
  99     {
 100         vp8_half_horiz_vert_variance16x_h_sse2(
 101             src_ptr, src_pixels_per_line,
 102             dst_ptr, dst_pixels_per_line, 16,
 103             &xsum0, &xxsum0);
 104     }
 105     else
 106     {
 107         vp8_filter_block2d_bil_var_ssse3(
 108             src_ptr, src_pixels_per_line,
 109             dst_ptr, dst_pixels_per_line, 16,
 110             xoffset, yoffset,
 111             &xsum0, &xxsum0);
 112     }
 113
 114     *sse = xxsum0;
 115     return (xxsum0 - ((xsum0 * xsum0) >> 8));
 116 }
 117
 118 unsigned int vp8_sub_pixel_variance16x8_ssse3
 119 (
 120     const unsigned char  *src_ptr,
 121     int  src_pixels_per_line,
 122     int  xoffset,
 123     int  yoffset,
 124     const unsigned char *dst_ptr,
 125     int dst_pixels_per_line,
 126     unsigned int *sse
 127
 128 )
 129 {
 130     int xsum0;
 131     unsigned int xxsum0;
 132
 133     if (xoffset == 4 && yoffset == 0)
 134     {
 135         vp8_half_horiz_variance16x_h_sse2(
 136             src_ptr, src_pixels_per_line,
 137             dst_ptr, dst_pixels_per_line, 8,
 138             &xsum0, &xxsum0);
 139     }
 140     else if (xoffset == 0 && yoffset == 4)
 141     {
 142         vp8_half_vert_variance16x_h_sse2(
 143             src_ptr, src_pixels_per_line,
 144             dst_ptr, dst_pixels_per_line, 8,
 145             &xsum0, &xxsum0);
 146     }
 147     else if (xoffset == 4 && yoffset == 4)
 148     {
 149         vp8_half_horiz_vert_variance16x_h_sse2(
 150             src_ptr, src_pixels_per_line,
 151             dst_ptr, dst_pixels_per_line, 8,
 152             &xsum0, &xxsum0);
 153     }
 154     else
 155     {
 156         vp8_filter_block2d_bil_var_ssse3(
 157             src_ptr, src_pixels_per_line,
 158             dst_ptr, dst_pixels_per_line, 8,
 159             xoffset, yoffset,
 160             &xsum0, &xxsum0);
 161     }
 162
 163     *sse = xxsum0;
 164     return (xxsum0 - ((xsum0 * xsum0) >> 7));
 165 }