aom_dsp/aom_dsp_rtcd_defs.pl

   1 ##
   2 ## Copyright (c) 2017, Alliance for Open Media. All rights reserved
   3 ##
   4 ## This source code is subject to the terms of the BSD 2 Clause License and
   5 ## the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
   6 ## was not distributed with this source code in the LICENSE file, you can
   7 ## obtain it at www.aomedia.org/license/software. If the Alliance for Open
   8 ## Media Patent License 1.0 was not distributed with this source code in the
   9 ## PATENTS file, you can obtain it at www.aomedia.org/license/patent.
  10 ##
  11 sub aom_dsp_forward_decls() {
  12 print <<EOF
  13 /*
  14  * DSP
  15  */
  16
  17 #include "aom/aom_integer.h"
  18 #include "aom_dsp/aom_dsp_common.h"
  19 #include "av1/common/enums.h"
  20 #include "av1/common/blockd.h"
  21
  22 EOF
  23 }
  24 forward_decls qw/aom_dsp_forward_decls/;
  25
  26 # optimizations which depend on multiple features
  27 $avx2_ssse3 = '';
  28 if ((aom_config("HAVE_AVX2") eq "yes") && (aom_config("HAVE_SSSE3") eq "yes")) {
  29   $avx2_ssse3 = 'avx2';
  30 }
  31
  32 # functions that are 64 bit only.
  33 $mmx_x86_64 = $sse2_x86_64 = $ssse3_x86_64 = $avx_x86_64 = $avx2_x86_64 = '';
  34 if ($opts{arch} eq "x86_64") {
  35   $mmx_x86_64 = 'mmx';
  36   $sse2_x86_64 = 'sse2';
  37   $ssse3_x86_64 = 'ssse3';
  38   $avx_x86_64 = 'avx';
  39   $avx2_x86_64 = 'avx2';
  40 }
  41
  42 @block_widths = (4, 8, 16, 32, 64, 128);
  43
  44 @block_sizes = ();
  45 foreach $w (@block_widths) {
  46   foreach $h (@block_widths) {
  47     push @block_sizes, [$w, $h] if ($w <= 2*$h && $h <= 2*$w) ;
  48   }
  49 }
  50 push @block_sizes, [4, 16];
  51 push @block_sizes, [16, 4];
  52 push @block_sizes, [8, 32];
  53 push @block_sizes, [32, 8];
  54 push @block_sizes, [16, 64];
  55 push @block_sizes, [64, 16];
  56
  57 @tx_dims = (2, 4, 8, 16, 32, 64);
  58 @tx_sizes = ();
  59 foreach $w (@tx_dims) {
  60   push @tx_sizes, [$w, $w];
  61   foreach $h (@tx_dims) {
  62     push @tx_sizes, [$w, $h] if ($w >=4 && $h >=4 && ($w == 2*$h || $h == 2*$w));
  63     push @tx_sizes, [$w, $h] if ($w >=4 && $h >=4 && ($w == 4*$h || $h == 4*$w));
  64   }
  65 }
  66
  67 @pred_names = qw/dc dc_top dc_left dc_128 v h paeth smooth smooth_v smooth_h/;
  68
  69 #
  70 # Intra prediction
  71 #
  72
  73 foreach (@tx_sizes) {
  74   ($w, $h) = @$_;
  75   foreach $pred_name (@pred_names) {
  76     add_proto "void", "aom_${pred_name}_predictor_${w}x${h}",
  77               "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
  78     add_proto "void", "aom_highbd_${pred_name}_predictor_${w}x${h}",
  79               "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  80   }
  81 }
  82
  83 specialize qw/aom_dc_top_predictor_4x4 msa neon sse2/;
  84 specialize qw/aom_dc_top_predictor_4x8 sse2/;
  85 specialize qw/aom_dc_top_predictor_4x16 sse2/;
  86 specialize qw/aom_dc_top_predictor_8x4 sse2/;
  87 specialize qw/aom_dc_top_predictor_8x8 neon msa sse2/;
  88 specialize qw/aom_dc_top_predictor_8x16 sse2/;
  89 specialize qw/aom_dc_top_predictor_8x32 sse2/;
  90 specialize qw/aom_dc_top_predictor_16x8 sse2/;
  91 specialize qw/aom_dc_top_predictor_16x16 neon msa sse2/;
  92 specialize qw/aom_dc_top_predictor_16x32 sse2/;
  93 specialize qw/aom_dc_top_predictor_16x64 sse2/;
  94 specialize qw/aom_dc_top_predictor_32x16 sse2 avx2/;
  95 specialize qw/aom_dc_top_predictor_32x32 msa neon sse2 avx2/;
  96 specialize qw/aom_dc_top_predictor_32x64 sse2 avx2/;
  97 specialize qw/aom_dc_top_predictor_64x64 sse2 avx2/;
  98 specialize qw/aom_dc_top_predictor_64x32 sse2 avx2/;
  99 specialize qw/aom_dc_top_predictor_64x16 sse2 avx2/;
 100 specialize qw/aom_dc_left_predictor_4x4 msa neon sse2/;
 101 specialize qw/aom_dc_left_predictor_4x8 sse2/;
 102 specialize qw/aom_dc_left_predictor_4x16 sse2/;
 103 specialize qw/aom_dc_left_predictor_8x4 sse2/;
 104 specialize qw/aom_dc_left_predictor_8x8 neon msa sse2/;
 105 specialize qw/aom_dc_left_predictor_8x16 sse2/;
 106 specialize qw/aom_dc_left_predictor_8x32 sse2/;
 107 specialize qw/aom_dc_left_predictor_16x8 sse2/;
 108 specialize qw/aom_dc_left_predictor_16x16 neon msa sse2/;
 109 specialize qw/aom_dc_left_predictor_16x32 sse2/;
 110 specialize qw/aom_dc_left_predictor_16x64 sse2/;
 111 specialize qw/aom_dc_left_predictor_32x16 sse2 avx2/;
 112 specialize qw/aom_dc_left_predictor_32x32 msa neon sse2 avx2/;
 113 specialize qw/aom_dc_left_predictor_32x64 sse2 avx2/;
 114 specialize qw/aom_dc_left_predictor_64x64 sse2 avx2/;
 115 specialize qw/aom_dc_left_predictor_64x32 sse2 avx2/;
 116 specialize qw/aom_dc_left_predictor_64x16 sse2 avx2/;
 117 specialize qw/aom_dc_128_predictor_4x4 msa neon sse2/;
 118 specialize qw/aom_dc_128_predictor_4x8 sse2/;
 119 specialize qw/aom_dc_128_predictor_4x16 sse2/;
 120 specialize qw/aom_dc_128_predictor_8x4 sse2/;
 121 specialize qw/aom_dc_128_predictor_8x8 neon msa sse2/;
 122 specialize qw/aom_dc_128_predictor_8x16 sse2/;
 123 specialize qw/aom_dc_128_predictor_8x32 sse2/;
 124 specialize qw/aom_dc_128_predictor_16x8 sse2/;
 125 specialize qw/aom_dc_128_predictor_16x16 neon msa sse2/;
 126 specialize qw/aom_dc_128_predictor_16x32 sse2/;
 127 specialize qw/aom_dc_128_predictor_16x64 sse2/;
 128 specialize qw/aom_dc_128_predictor_32x16 sse2 avx2/;
 129 specialize qw/aom_dc_128_predictor_32x32 msa neon sse2 avx2/;
 130 specialize qw/aom_dc_128_predictor_32x64 sse2 avx2/;
 131 specialize qw/aom_dc_128_predictor_64x64 sse2 avx2/;
 132 specialize qw/aom_dc_128_predictor_64x32 sse2 avx2/;
 133 specialize qw/aom_dc_128_predictor_64x16 sse2 avx2/;
 134 specialize qw/aom_v_predictor_4x4 neon msa sse2/;
 135 specialize qw/aom_v_predictor_4x8 sse2/;
 136 specialize qw/aom_v_predictor_4x16 sse2/;
 137 specialize qw/aom_v_predictor_8x4 sse2/;
 138 specialize qw/aom_v_predictor_8x8 neon msa sse2/;
 139 specialize qw/aom_v_predictor_8x16 sse2/;
 140 specialize qw/aom_v_predictor_8x32 sse2/;
 141 specialize qw/aom_v_predictor_16x8 sse2/;
 142 specialize qw/aom_v_predictor_16x16 neon msa sse2/;
 143 specialize qw/aom_v_predictor_16x32 sse2/;
 144 specialize qw/aom_v_predictor_16x64 sse2/;
 145 specialize qw/aom_v_predictor_32x16 sse2 avx2/;
 146 specialize qw/aom_v_predictor_32x32 neon msa sse2 avx2/;
 147 specialize qw/aom_v_predictor_32x64 sse2 avx2/;
 148 specialize qw/aom_v_predictor_64x64 sse2 avx2/;
 149 specialize qw/aom_v_predictor_64x32 sse2 avx2/;
 150 specialize qw/aom_v_predictor_64x16 sse2 avx2/;
 151 specialize qw/aom_h_predictor_4x8 sse2/;
 152 specialize qw/aom_h_predictor_4x16 sse2/;
 153 specialize qw/aom_h_predictor_4x4 neon dspr2 msa sse2/;
 154 specialize qw/aom_h_predictor_8x4 sse2/;
 155 specialize qw/aom_h_predictor_8x8 neon dspr2 msa sse2/;
 156 specialize qw/aom_h_predictor_8x16 sse2/;
 157 specialize qw/aom_h_predictor_8x32 sse2/;
 158 specialize qw/aom_h_predictor_16x8 sse2/;
 159 specialize qw/aom_h_predictor_16x16 neon dspr2 msa sse2/;
 160 specialize qw/aom_h_predictor_16x32 sse2/;
 161 specialize qw/aom_h_predictor_16x64 sse2/;
 162 specialize qw/aom_h_predictor_32x16 sse2/;
 163 specialize qw/aom_h_predictor_32x32 neon msa sse2 avx2/;
 164 specialize qw/aom_h_predictor_32x64 sse2/;
 165 specialize qw/aom_h_predictor_64x64 sse2/;
 166 specialize qw/aom_h_predictor_64x32 sse2/;
 167 specialize qw/aom_h_predictor_64x16 sse2/;
 168 specialize qw/aom_paeth_predictor_4x4 ssse3/;
 169 specialize qw/aom_paeth_predictor_4x8 ssse3/;
 170 specialize qw/aom_paeth_predictor_4x16 ssse3/;
 171 specialize qw/aom_paeth_predictor_8x4 ssse3/;
 172 specialize qw/aom_paeth_predictor_8x8 ssse3/;
 173 specialize qw/aom_paeth_predictor_8x16 ssse3/;
 174 specialize qw/aom_paeth_predictor_8x32 ssse3/;
 175 specialize qw/aom_paeth_predictor_16x8 ssse3 avx2/;
 176 specialize qw/aom_paeth_predictor_16x16 ssse3 avx2/;
 177 specialize qw/aom_paeth_predictor_16x32 ssse3 avx2/;
 178 specialize qw/aom_paeth_predictor_16x64 ssse3 avx2/;
 179 specialize qw/aom_paeth_predictor_32x16 ssse3 avx2/;
 180 specialize qw/aom_paeth_predictor_32x32 ssse3 avx2/;
 181 specialize qw/aom_paeth_predictor_32x64 ssse3 avx2/;
 182 specialize qw/aom_paeth_predictor_64x32 ssse3 avx2/;
 183 specialize qw/aom_paeth_predictor_64x64 ssse3 avx2/;
 184 specialize qw/aom_paeth_predictor_64x16 ssse3 avx2/;
 185 specialize qw/aom_paeth_predictor_16x8 ssse3/;
 186 specialize qw/aom_paeth_predictor_16x16 ssse3/;
 187 specialize qw/aom_paeth_predictor_16x32 ssse3/;
 188 specialize qw/aom_paeth_predictor_32x16 ssse3/;
 189 specialize qw/aom_paeth_predictor_32x32 ssse3/;
 190 specialize qw/aom_smooth_predictor_4x4 ssse3/;
 191 specialize qw/aom_smooth_predictor_4x8 ssse3/;
 192 specialize qw/aom_smooth_predictor_4x16 ssse3/;
 193 specialize qw/aom_smooth_predictor_8x4 ssse3/;
 194 specialize qw/aom_smooth_predictor_8x8 ssse3/;
 195 specialize qw/aom_smooth_predictor_8x16 ssse3/;
 196 specialize qw/aom_smooth_predictor_8x32 ssse3/;
 197 specialize qw/aom_smooth_predictor_16x8 ssse3/;
 198 specialize qw/aom_smooth_predictor_16x16 ssse3/;
 199 specialize qw/aom_smooth_predictor_16x32 ssse3/;
 200 specialize qw/aom_smooth_predictor_16x64 ssse3/;
 201 specialize qw/aom_smooth_predictor_32x16 ssse3/;
 202 specialize qw/aom_smooth_predictor_32x32 ssse3/;
 203 specialize qw/aom_smooth_predictor_32x64 ssse3/;
 204 specialize qw/aom_smooth_predictor_64x64 ssse3/;
 205 specialize qw/aom_smooth_predictor_64x32 ssse3/;
 206 specialize qw/aom_smooth_predictor_64x16 ssse3/;
 207
 208 specialize qw/aom_smooth_v_predictor_4x4 ssse3/;
 209 specialize qw/aom_smooth_v_predictor_4x8 ssse3/;
 210 specialize qw/aom_smooth_v_predictor_4x16 ssse3/;
 211 specialize qw/aom_smooth_v_predictor_8x4 ssse3/;
 212 specialize qw/aom_smooth_v_predictor_8x8 ssse3/;
 213 specialize qw/aom_smooth_v_predictor_8x16 ssse3/;
 214 specialize qw/aom_smooth_v_predictor_8x32 ssse3/;
 215 specialize qw/aom_smooth_v_predictor_16x4 ssse3/;
 216 specialize qw/aom_smooth_v_predictor_16x8 ssse3/;
 217 specialize qw/aom_smooth_v_predictor_16x16 ssse3/;
 218 specialize qw/aom_smooth_v_predictor_16x32 ssse3/;
 219 specialize qw/aom_smooth_v_predictor_16x64 ssse3/;
 220 specialize qw/aom_smooth_v_predictor_32x8 ssse3/;
 221 specialize qw/aom_smooth_v_predictor_32x16 ssse3/;
 222 specialize qw/aom_smooth_v_predictor_32x32 ssse3/;
 223 specialize qw/aom_smooth_v_predictor_32x64 ssse3/;
 224 specialize qw/aom_smooth_v_predictor_64x64 ssse3/;
 225 specialize qw/aom_smooth_v_predictor_64x32 ssse3/;
 226 specialize qw/aom_smooth_v_predictor_64x16 ssse3/;
 227
 228 specialize qw/aom_smooth_h_predictor_4x4 ssse3/;
 229 specialize qw/aom_smooth_h_predictor_4x8 ssse3/;
 230 specialize qw/aom_smooth_h_predictor_4x16 ssse3/;
 231 specialize qw/aom_smooth_h_predictor_8x4 ssse3/;
 232 specialize qw/aom_smooth_h_predictor_8x8 ssse3/;
 233 specialize qw/aom_smooth_h_predictor_8x16 ssse3/;
 234 specialize qw/aom_smooth_h_predictor_8x32 ssse3/;
 235 specialize qw/aom_smooth_h_predictor_16x4 ssse3/;
 236 specialize qw/aom_smooth_h_predictor_16x8 ssse3/;
 237 specialize qw/aom_smooth_h_predictor_16x16 ssse3/;
 238 specialize qw/aom_smooth_h_predictor_16x32 ssse3/;
 239 specialize qw/aom_smooth_h_predictor_16x64 ssse3/;
 240 specialize qw/aom_smooth_h_predictor_32x8 ssse3/;
 241 specialize qw/aom_smooth_h_predictor_32x16 ssse3/;
 242 specialize qw/aom_smooth_h_predictor_32x32 ssse3/;
 243 specialize qw/aom_smooth_h_predictor_32x64 ssse3/;
 244 specialize qw/aom_smooth_h_predictor_64x64 ssse3/;
 245 specialize qw/aom_smooth_h_predictor_64x32 ssse3/;
 246 specialize qw/aom_smooth_h_predictor_64x16 ssse3/;
 247
 248 # TODO(yunqingwang): optimize rectangular DC_PRED to replace division
 249 # by multiply and shift.
 250 specialize qw/aom_dc_predictor_4x4 dspr2 msa neon sse2/;
 251 specialize qw/aom_dc_predictor_4x8 sse2/;
 252 specialize qw/aom_dc_predictor_4x16 sse2/;
 253 specialize qw/aom_dc_predictor_8x4 sse2/;
 254 specialize qw/aom_dc_predictor_8x8 dspr2 neon msa sse2/;
 255 specialize qw/aom_dc_predictor_8x16 sse2/;
 256 specialize qw/aom_dc_predictor_8x32 sse2/;
 257 specialize qw/aom_dc_predictor_16x8 sse2/;
 258 specialize qw/aom_dc_predictor_16x16 dspr2 neon msa sse2/;
 259 specialize qw/aom_dc_predictor_16x32 sse2/;
 260 specialize qw/aom_dc_predictor_16x64 sse2/;
 261 specialize qw/aom_dc_predictor_32x16 sse2 avx2/;
 262 specialize qw/aom_dc_predictor_32x32 msa neon sse2 avx2/;
 263 specialize qw/aom_dc_predictor_32x64 sse2 avx2/;
 264 specialize qw/aom_dc_predictor_64x64 sse2 avx2/;
 265 specialize qw/aom_dc_predictor_64x32 sse2 avx2/;
 266 specialize qw/aom_dc_predictor_64x16 sse2 avx2/;
 267
 268   specialize qw/aom_highbd_v_predictor_4x4 sse2/;
 269   specialize qw/aom_highbd_v_predictor_4x8 sse2/;
 270   specialize qw/aom_highbd_v_predictor_8x4 sse2/;
 271   specialize qw/aom_highbd_v_predictor_8x8 sse2/;
 272   specialize qw/aom_highbd_v_predictor_8x16 sse2/;
 273   specialize qw/aom_highbd_v_predictor_16x8 sse2/;
 274   specialize qw/aom_highbd_v_predictor_16x16 sse2/;
 275   specialize qw/aom_highbd_v_predictor_16x32 sse2/;
 276   specialize qw/aom_highbd_v_predictor_32x16 sse2/;
 277   specialize qw/aom_highbd_v_predictor_32x32 sse2/;
 278
 279   # TODO(yunqingwang): optimize rectangular DC_PRED to replace division
 280   # by multiply and shift.
 281   specialize qw/aom_highbd_dc_predictor_4x4 sse2/;
 282   specialize qw/aom_highbd_dc_predictor_4x8 sse2/;
 283   specialize qw/aom_highbd_dc_predictor_8x4 sse2/;;
 284   specialize qw/aom_highbd_dc_predictor_8x8 sse2/;;
 285   specialize qw/aom_highbd_dc_predictor_8x16 sse2/;;
 286   specialize qw/aom_highbd_dc_predictor_16x8 sse2/;
 287   specialize qw/aom_highbd_dc_predictor_16x16 sse2/;
 288   specialize qw/aom_highbd_dc_predictor_16x32 sse2/;
 289   specialize qw/aom_highbd_dc_predictor_32x16 sse2/;
 290   specialize qw/aom_highbd_dc_predictor_32x32 sse2/;
 291
 292   specialize qw/aom_highbd_h_predictor_4x4 sse2/;
 293   specialize qw/aom_highbd_h_predictor_4x8 sse2/;
 294   specialize qw/aom_highbd_h_predictor_8x4 sse2/;
 295   specialize qw/aom_highbd_h_predictor_8x8 sse2/;
 296   specialize qw/aom_highbd_h_predictor_8x16 sse2/;
 297   specialize qw/aom_highbd_h_predictor_16x8 sse2/;
 298   specialize qw/aom_highbd_h_predictor_16x16 sse2/;
 299   specialize qw/aom_highbd_h_predictor_16x32 sse2/;
 300   specialize qw/aom_highbd_h_predictor_32x16 sse2/;
 301   specialize qw/aom_highbd_h_predictor_32x32 sse2/;
 302   specialize qw/aom_highbd_dc_left_predictor_4x4 sse2/;
 303   specialize qw/aom_highbd_dc_top_predictor_4x4 sse2/;
 304   specialize qw/aom_highbd_dc_128_predictor_4x4 sse2/;
 305   specialize qw/aom_highbd_dc_left_predictor_4x8 sse2/;
 306   specialize qw/aom_highbd_dc_top_predictor_4x8 sse2/;
 307   specialize qw/aom_highbd_dc_128_predictor_4x8 sse2/;
 308   specialize qw/aom_highbd_dc_left_predictor_8x4 sse2/;
 309   specialize qw/aom_highbd_dc_top_predictor_8x4 sse2/;
 310   specialize qw/aom_highbd_dc_128_predictor_8x4 sse2/;
 311   specialize qw/aom_highbd_dc_left_predictor_8x8 sse2/;
 312   specialize qw/aom_highbd_dc_top_predictor_8x8 sse2/;
 313   specialize qw/aom_highbd_dc_128_predictor_8x8 sse2/;
 314   specialize qw/aom_highbd_dc_left_predictor_8x16 sse2/;
 315   specialize qw/aom_highbd_dc_top_predictor_8x16 sse2/;
 316   specialize qw/aom_highbd_dc_128_predictor_8x16 sse2/;
 317   specialize qw/aom_highbd_dc_left_predictor_16x8 sse2/;
 318   specialize qw/aom_highbd_dc_top_predictor_16x8 sse2/;
 319   specialize qw/aom_highbd_dc_128_predictor_16x8 sse2/;
 320   specialize qw/aom_highbd_dc_left_predictor_16x16 sse2/;
 321   specialize qw/aom_highbd_dc_top_predictor_16x16 sse2/;
 322   specialize qw/aom_highbd_dc_128_predictor_16x16 sse2/;
 323   specialize qw/aom_highbd_dc_left_predictor_16x32 sse2/;
 324   specialize qw/aom_highbd_dc_top_predictor_16x32 sse2/;
 325   specialize qw/aom_highbd_dc_128_predictor_16x32 sse2/;
 326   specialize qw/aom_highbd_dc_left_predictor_32x16 sse2/;
 327   specialize qw/aom_highbd_dc_top_predictor_32x16 sse2/;
 328   specialize qw/aom_highbd_dc_128_predictor_32x16 sse2/;
 329   specialize qw/aom_highbd_dc_left_predictor_32x32 sse2/;
 330   specialize qw/aom_highbd_dc_top_predictor_32x32 sse2/;
 331   specialize qw/aom_highbd_dc_128_predictor_32x32 sse2/;
 332
 333 #
 334 # Sub Pixel Filters
 335 #
 336 add_proto qw/void aom_convolve_copy/,             "const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h";
 337 add_proto qw/void aom_convolve8_horiz/,           "const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h";
 338 add_proto qw/void aom_convolve8_vert/,            "const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h";
 339
 340 specialize qw/aom_convolve_copy       sse2      /;
 341 specialize qw/aom_convolve8_horiz     sse2 ssse3/, "$avx2_ssse3";
 342 specialize qw/aom_convolve8_vert      sse2 ssse3/, "$avx2_ssse3";
 343
 344 # TODO(any): These need to be extended to up to 128x128 block sizes
 345 if (!(aom_config("CONFIG_AV1") eq "yes")) {
 346   specialize qw/aom_convolve_copy       neon dspr2 msa/;
 347   specialize qw/aom_convolve8_horiz     neon dspr2 msa/;
 348   specialize qw/aom_convolve8_vert      neon dspr2 msa/;
 349 }
 350
 351 add_proto qw/void aom_highbd_convolve_copy/, "const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps";
 352 specialize qw/aom_highbd_convolve_copy sse2 avx2/;
 353
 354 add_proto qw/void aom_highbd_convolve8_horiz/, "const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps";
 355 specialize qw/aom_highbd_convolve8_horiz avx2/, "$sse2_x86_64";
 356
 357 add_proto qw/void aom_highbd_convolve8_vert/, "const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps";
 358 specialize qw/aom_highbd_convolve8_vert avx2/, "$sse2_x86_64";
 359
 360 #
 361 # Loopfilter
 362 #
 363 add_proto qw/void aom_lpf_vertical_14/, "uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh";
 364 specialize qw/aom_lpf_vertical_14 sse2/;
 365
 366 add_proto qw/void aom_lpf_vertical_14_dual/, "uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh";
 367 specialize qw/aom_lpf_vertical_14_dual sse2/;
 368
 369 add_proto qw/void aom_lpf_vertical_6/, "uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh";
 370 specialize qw/aom_lpf_vertical_6 sse2/;
 371
 372 add_proto qw/void aom_lpf_vertical_8/, "uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh";
 373 specialize qw/aom_lpf_vertical_8 sse2/;
 374
 375 add_proto qw/void aom_lpf_vertical_8_dual/, "uint8_t *s, int pitch, const uint8_t *blimit0, const uint8_t *limit0, const uint8_t *thresh0, const uint8_t *blimit1, const uint8_t *limit1, const uint8_t *thresh1";
 376
 377 add_proto qw/void aom_lpf_vertical_4/, "uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh";
 378 specialize qw/aom_lpf_vertical_4 sse2/;
 379
 380 add_proto qw/void aom_lpf_vertical_4_dual/, "uint8_t *s, int pitch, const uint8_t *blimit0, const uint8_t *limit0, const uint8_t *thresh0, const uint8_t *blimit1, const uint8_t *limit1, const uint8_t *thresh1";
 381
 382 add_proto qw/void aom_lpf_horizontal_14/, "uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh";
 383 specialize qw/aom_lpf_horizontal_14 sse2/;
 384
 385 add_proto qw/void aom_lpf_horizontal_14_dual/, "uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh";
 386 specialize qw/aom_lpf_horizontal_14_dual sse2/;
 387
 388 add_proto qw/void aom_lpf_horizontal_6/, "uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh";
 389 specialize qw/aom_lpf_horizontal_6 sse2/;
 390
 391 add_proto qw/void aom_lpf_horizontal_8/, "uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh";
 392 specialize qw/aom_lpf_horizontal_8 sse2/;
 393
 394 add_proto qw/void aom_lpf_horizontal_8_dual/, "uint8_t *s, int pitch, const uint8_t *blimit0, const uint8_t *limit0, const uint8_t *thresh0, const uint8_t *blimit1, const uint8_t *limit1, const uint8_t *thresh1";
 395
 396 add_proto qw/void aom_lpf_horizontal_4/, "uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh";
 397 specialize qw/aom_lpf_horizontal_4 sse2/;
 398
 399 add_proto qw/void aom_lpf_horizontal_4_dual/, "uint8_t *s, int pitch, const uint8_t *blimit0, const uint8_t *limit0, const uint8_t *thresh0, const uint8_t *blimit1, const uint8_t *limit1, const uint8_t *thresh1";
 400
 401 add_proto qw/void aom_highbd_lpf_vertical_14/, "uint16_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh, int bd";
 402 specialize qw/aom_highbd_lpf_vertical_14 sse2/;
 403
 404 add_proto qw/void aom_highbd_lpf_vertical_14_dual/, "uint16_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh,   int bd";
 405 specialize qw/aom_highbd_lpf_vertical_14_dual sse2 avx2/;
 406
 407 add_proto qw/void aom_highbd_lpf_vertical_8/, "uint16_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh, int bd";
 408 specialize qw/aom_highbd_lpf_vertical_8 sse2/;
 409
 410 add_proto qw/void aom_highbd_lpf_vertical_6/, "uint16_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh, int bd";
 411 specialize qw/aom_highbd_lpf_vertical_6 sse2/;
 412
 413 add_proto qw/void aom_highbd_lpf_vertical_8_dual/, "uint16_t *s, int pitch, const uint8_t *blimit0, const uint8_t *limit0, const uint8_t *thresh0, const uint8_t *blimit1, const uint8_t *limit1, const uint8_t *thresh1, int bd";
 414 specialize qw/aom_highbd_lpf_vertical_8_dual sse2 avx2/;
 415
 416 add_proto qw/void aom_highbd_lpf_vertical_4/, "uint16_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh, int bd";
 417 specialize qw/aom_highbd_lpf_vertical_4 sse2/;
 418
 419 add_proto qw/void aom_highbd_lpf_vertical_4_dual/, "uint16_t *s, int pitch, const uint8_t *blimit0, const uint8_t *limit0, const uint8_t *thresh0, const uint8_t *blimit1, const uint8_t *limit1, const uint8_t *thresh1, int bd";
 420 specialize qw/aom_highbd_lpf_vertical_4_dual sse2 avx2/;
 421
 422 add_proto qw/void aom_highbd_lpf_horizontal_14/, "uint16_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh, int bd";
 423 specialize qw/aom_highbd_lpf_horizontal_14 sse2/;
 424
 425 add_proto qw/void aom_highbd_lpf_horizontal_14_dual/, "uint16_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh, int bd";
 426 specialize qw/aom_highbd_lpf_horizontal_14_dual sse2 avx2/;
 427
 428 add_proto qw/void aom_highbd_lpf_horizontal_6/, "uint16_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh, int bd";
 429 specialize qw/aom_highbd_lpf_horizontal_6 sse2/;
 430
 431 add_proto qw/void aom_highbd_lpf_horizontal_8/, "uint16_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh, int bd";
 432 specialize qw/aom_highbd_lpf_horizontal_8 sse2/;
 433
 434 add_proto qw/void aom_highbd_lpf_horizontal_8_dual/, "uint16_t *s, int pitch, const uint8_t *blimit0, const uint8_t *limit0, const uint8_t *thresh0, const uint8_t *blimit1, const uint8_t *limit1, const uint8_t *thresh1, int bd";
 435 specialize qw/aom_highbd_lpf_horizontal_8_dual sse2 avx2/;
 436
 437 add_proto qw/void aom_highbd_lpf_horizontal_4/, "uint16_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh, int bd";
 438 specialize qw/aom_highbd_lpf_horizontal_4 sse2/;
 439
 440 add_proto qw/void aom_highbd_lpf_horizontal_4_dual/, "uint16_t *s, int pitch, const uint8_t *blimit0, const uint8_t *limit0, const uint8_t *thresh0, const uint8_t *blimit1, const uint8_t *limit1, const uint8_t *thresh1, int bd";
 441 specialize qw/aom_highbd_lpf_horizontal_4_dual sse2 avx2/;
 442
 443 # Helper functions.
 444 add_proto qw/void av1_round_shift_array/, "int32_t *arr, int size, int bit";
 445 specialize "av1_round_shift_array", qw/sse4_1/;
 446
 447 #
 448 # Encoder functions.
 449 #
 450
 451 #
 452 # Forward transform
 453 #
 454 if (aom_config("CONFIG_AV1_ENCODER") eq "yes"){
 455     add_proto qw/void aom_fdct4x4/, "const int16_t *input, tran_low_t *output, int stride";
 456     specialize qw/aom_fdct4x4 sse2/;
 457
 458     add_proto qw/void aom_fdct4x4_1/, "const int16_t *input, tran_low_t *output, int stride";
 459     specialize qw/aom_fdct4x4_1 sse2/;
 460
 461     add_proto qw/void aom_fdct8x8/, "const int16_t *input, tran_low_t *output, int stride";
 462     specialize qw/aom_fdct8x8 sse2/, "$ssse3_x86_64";
 463
 464     # High bit depth
 465     add_proto qw/void aom_highbd_fdct4x4/, "const int16_t *input, tran_low_t *output, int stride";
 466     specialize qw/aom_highbd_fdct4x4 sse2/;
 467
 468     add_proto qw/void aom_highbd_fdct8x8/, "const int16_t *input, tran_low_t *output, int stride";
 469     specialize qw/aom_highbd_fdct8x8 sse2/;
 470
 471 }  # CONFIG_AV1_ENCODER
 472
 473 #
 474 # Inverse transform
 475 if (aom_config("CONFIG_AV1") eq "yes") {
 476   add_proto qw/void aom_highbd_iwht4x4_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int bd";
 477
 478   add_proto qw/void aom_highbd_iwht4x4_16_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int bd";
 479 }  # CONFIG_AV1
 480
 481 #
 482 # Quantization
 483 #
 484 if (aom_config("CONFIG_AV1_ENCODER") eq "yes") {
 485   add_proto qw/void aom_quantize_b/, "const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan";
 486   specialize qw/aom_quantize_b sse2/, "$ssse3_x86_64", "$avx_x86_64";
 487
 488   add_proto qw/void aom_quantize_b_32x32/, "const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan";
 489   specialize qw/aom_quantize_b_32x32/, "$ssse3_x86_64", "$avx_x86_64";
 490
 491   add_proto qw/void aom_quantize_b_64x64/, "const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan";
 492 }  # CONFIG_AV1_ENCODER
 493
 494 if (aom_config("CONFIG_AV1_ENCODER") eq "yes") {
 495   add_proto qw/void aom_highbd_quantize_b/, "const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan";
 496   specialize qw/aom_highbd_quantize_b sse2 avx2/;
 497
 498   add_proto qw/void aom_highbd_quantize_b_32x32/, "const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan";
 499   specialize qw/aom_highbd_quantize_b_32x32 sse2/;
 500
 501   add_proto qw/void aom_highbd_quantize_b_64x64/, "const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan";
 502
 503 }  # CONFIG_AV1_ENCODER
 504 if (aom_config("CONFIG_AV1") eq "yes") {
 505   #
 506   # Alpha blending with mask
 507   #
 508   add_proto qw/void aom_lowbd_blend_a64_d16_mask/, "uint8_t *dst, uint32_t dst_stride, const CONV_BUF_TYPE *src0, uint32_t src0_stride, const CONV_BUF_TYPE *src1, uint32_t src1_stride, const uint8_t *mask, uint32_t mask_stride, int h, int w, int suby, int subx, ConvolveParams *conv_params";
 509   add_proto qw/void aom_highbd_blend_a64_d16_mask/, "uint8_t *dst, uint32_t dst_stride, const CONV_BUF_TYPE *src0, uint32_t src0_stride, const CONV_BUF_TYPE *src1, uint32_t src1_stride, const uint8_t *mask, uint32_t mask_stride, int h, int w, int suby, int subx, ConvolveParams *conv_params, const int bd";
 510   add_proto qw/void aom_blend_a64_mask/, "uint8_t *dst, uint32_t dst_stride, const uint8_t *src0, uint32_t src0_stride, const uint8_t *src1, uint32_t src1_stride, const uint8_t *mask, uint32_t mask_stride, int h, int w, int suby, int subx";
 511   add_proto qw/void aom_blend_a64_hmask/, "uint8_t *dst, uint32_t dst_stride, const uint8_t *src0, uint32_t src0_stride, const uint8_t *src1, uint32_t src1_stride, const uint8_t *mask, int h, int w";
 512   add_proto qw/void aom_blend_a64_vmask/, "uint8_t *dst, uint32_t dst_stride, const uint8_t *src0, uint32_t src0_stride, const uint8_t *src1, uint32_t src1_stride, const uint8_t *mask, int h, int w";
 513   specialize "aom_blend_a64_mask", qw/sse4_1/;
 514   specialize "aom_blend_a64_hmask", qw/sse4_1/;
 515   specialize "aom_blend_a64_vmask", qw/sse4_1/;
 516
 517   add_proto qw/void aom_highbd_blend_a64_mask/, "uint8_t *dst, uint32_t dst_stride, const uint8_t *src0, uint32_t src0_stride, const uint8_t *src1, uint32_t src1_stride, const uint8_t *mask, uint32_t mask_stride, int h, int w, int suby, int subx, int bd";
 518   add_proto qw/void aom_highbd_blend_a64_hmask/, "uint8_t *dst, uint32_t dst_stride, const uint8_t *src0, uint32_t src0_stride, const uint8_t *src1, uint32_t src1_stride, const uint8_t *mask, int h, int w, int bd";
 519   add_proto qw/void aom_highbd_blend_a64_vmask/, "uint8_t *dst, uint32_t dst_stride, const uint8_t *src0, uint32_t src0_stride, const uint8_t *src1, uint32_t src1_stride, const uint8_t *mask, int h, int w, int bd";
 520   specialize "aom_highbd_blend_a64_mask", qw/sse4_1/;
 521   specialize "aom_highbd_blend_a64_hmask", qw/sse4_1/;
 522   specialize "aom_highbd_blend_a64_vmask", qw/sse4_1/;
 523
 524 }  # CONFIG_AV1
 525
 526 if (aom_config("CONFIG_AV1_ENCODER") eq "yes") {
 527   #
 528   # Block subtraction
 529   #
 530   add_proto qw/void aom_subtract_block/, "int rows, int cols, int16_t *diff_ptr, ptrdiff_t diff_stride, const uint8_t *src_ptr, ptrdiff_t src_stride, const uint8_t *pred_ptr, ptrdiff_t pred_stride";
 531   specialize qw/aom_subtract_block neon msa sse2/;
 532
 533   if (aom_config("CONFIG_AV1_ENCODER") eq "yes") {
 534     #
 535     # Sum of Squares
 536     #
 537     add_proto qw/uint64_t aom_sum_squares_2d_i16/, "const int16_t *src, int stride, int width, int height";
 538     specialize qw/aom_sum_squares_2d_i16 sse2/;
 539
 540     add_proto qw/uint64_t aom_sum_squares_i16/, "const int16_t *src, uint32_t N";
 541     specialize qw/aom_sum_squares_i16 sse2/;
 542   }
 543
 544
 545   #
 546   # Avg
 547   #
 548   if (aom_config("CONFIG_AV1_ENCODER") eq "yes") {
 549     #
 550     # Avg
 551     #
 552     specialize qw/aom_avg_8x8 sse2 neon msa/;
 553       add_proto qw/void aom_highbd_subtract_block/, "int rows, int cols, int16_t *diff_ptr, ptrdiff_t diff_stride, const uint8_t *src_ptr, ptrdiff_t src_stride, const uint8_t *pred_ptr, ptrdiff_t pred_stride, int bd";
 554       specialize qw/aom_highbd_subtract_block sse2/;
 555
 556     #
 557     # Minmax
 558     #
 559     add_proto qw/void aom_minmax_8x8/, "const uint8_t *s, int p, const uint8_t *d, int dp, int *min, int *max";
 560     specialize qw/aom_minmax_8x8 sse2 neon/;
 561       add_proto qw/void aom_highbd_minmax_8x8/, "const uint8_t *s, int p, const uint8_t *d, int dp, int *min, int *max";
 562
 563     add_proto qw/void aom_hadamard_8x8/, "const int16_t *src_diff, int src_stride, int16_t *coeff";
 564     specialize qw/aom_hadamard_8x8 sse2 neon/, "$ssse3_x86_64";
 565
 566     add_proto qw/void aom_hadamard_16x16/, "const int16_t *src_diff, int src_stride, int16_t *coeff";
 567     specialize qw/aom_hadamard_16x16 sse2 neon/;
 568
 569     add_proto qw/int aom_satd/, "const int16_t *coeff, int length";
 570     specialize qw/aom_satd sse2 neon/;
 571
 572     add_proto qw/void aom_int_pro_row/, "int16_t *hbuf, const uint8_t *ref, int ref_stride, int height";
 573     specialize qw/aom_int_pro_row sse2 neon/;
 574
 575     add_proto qw/int16_t aom_int_pro_col/, "const uint8_t *ref, int width";
 576     specialize qw/aom_int_pro_col sse2 neon/;
 577
 578     add_proto qw/int aom_vector_var/, "const int16_t *ref, const int16_t *src, int bwl";
 579     specialize qw/aom_vector_var neon sse2/;
 580   }  # CONFIG_AV1_ENCODER
 581
 582   #
 583   # Single block SAD / Single block Avg SAD
 584   #
 585   foreach (@block_sizes) {
 586     ($w, $h) = @$_;
 587     add_proto qw/unsigned int/, "aom_sad${w}x${h}", "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride";
 588     add_proto qw/unsigned int/, "aom_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";
 589     add_proto qw/unsigned int/, "aom_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";
 590   }
 591
 592   specialize qw/aom_sad128x128    avx2          sse2/;
 593   specialize qw/aom_sad128x64     avx2          sse2/;
 594   specialize qw/aom_sad64x128     avx2          sse2/;
 595   specialize qw/aom_sad64x64      avx2 neon msa sse2/;
 596   specialize qw/aom_sad64x32      avx2      msa sse2/;
 597   specialize qw/aom_sad32x64      avx2      msa sse2/;
 598   specialize qw/aom_sad32x32      avx2 neon msa sse2/;
 599   specialize qw/aom_sad32x16      avx2      msa sse2/;
 600   specialize qw/aom_sad16x32                msa sse2/;
 601   specialize qw/aom_sad16x16           neon msa sse2/;
 602   specialize qw/aom_sad16x8            neon msa sse2/;
 603   specialize qw/aom_sad8x16            neon msa sse2/;
 604   specialize qw/aom_sad8x8             neon msa sse2/;
 605   specialize qw/aom_sad8x4                  msa sse2/;
 606   specialize qw/aom_sad4x8                  msa sse2/;
 607   specialize qw/aom_sad4x4             neon msa sse2/;
 608
 609   specialize qw/aom_sad128x128_avg avx2     sse2/;
 610   specialize qw/aom_sad128x64_avg  avx2     sse2/;
 611   specialize qw/aom_sad64x128_avg  avx2     sse2/;
 612   specialize qw/aom_sad64x64_avg   avx2 msa sse2/;
 613   specialize qw/aom_sad64x32_avg   avx2 msa sse2/;
 614   specialize qw/aom_sad32x64_avg   avx2 msa sse2/;
 615   specialize qw/aom_sad32x32_avg   avx2 msa sse2/;
 616   specialize qw/aom_sad32x16_avg   avx2 msa sse2/;
 617   specialize qw/aom_sad16x32_avg        msa sse2/;
 618   specialize qw/aom_sad16x16_avg        msa sse2/;
 619   specialize qw/aom_sad16x8_avg         msa sse2/;
 620   specialize qw/aom_sad8x16_avg         msa sse2/;
 621   specialize qw/aom_sad8x8_avg          msa sse2/;
 622   specialize qw/aom_sad8x4_avg          msa sse2/;
 623   specialize qw/aom_sad4x8_avg          msa sse2/;
 624   specialize qw/aom_sad4x4_avg          msa sse2/;
 625
 626   specialize qw/aom_sad4x16      sse2/;
 627   specialize qw/aom_sad16x4      sse2/;
 628   specialize qw/aom_sad8x32      sse2/;
 629   specialize qw/aom_sad32x8      sse2/;
 630   specialize qw/aom_sad16x64     sse2/;
 631   specialize qw/aom_sad64x16     sse2/;
 632
 633   specialize qw/aom_sad4x16_avg  sse2/;
 634   specialize qw/aom_sad16x4_avg  sse2/;
 635   specialize qw/aom_sad8x32_avg  sse2/;
 636   specialize qw/aom_sad32x8_avg  sse2/;
 637   specialize qw/aom_sad16x64_avg sse2/;
 638   specialize qw/aom_sad64x16_avg sse2/;
 639
 640   specialize qw/aom_jnt_sad128x128_avg ssse3/;
 641   specialize qw/aom_jnt_sad128x64_avg  ssse3/;
 642   specialize qw/aom_jnt_sad64x128_avg  ssse3/;
 643   specialize qw/aom_jnt_sad64x64_avg   ssse3/;
 644   specialize qw/aom_jnt_sad64x32_avg   ssse3/;
 645   specialize qw/aom_jnt_sad32x64_avg   ssse3/;
 646   specialize qw/aom_jnt_sad32x32_avg   ssse3/;
 647   specialize qw/aom_jnt_sad32x16_avg   ssse3/;
 648   specialize qw/aom_jnt_sad16x32_avg   ssse3/;
 649   specialize qw/aom_jnt_sad16x16_avg   ssse3/;
 650   specialize qw/aom_jnt_sad16x8_avg    ssse3/;
 651   specialize qw/aom_jnt_sad8x16_avg    ssse3/;
 652   specialize qw/aom_jnt_sad8x8_avg     ssse3/;
 653   specialize qw/aom_jnt_sad8x4_avg     ssse3/;
 654   specialize qw/aom_jnt_sad4x8_avg     ssse3/;
 655   specialize qw/aom_jnt_sad4x4_avg     ssse3/;
 656
 657   specialize qw/aom_jnt_sad4x16_avg     ssse3/;
 658   specialize qw/aom_jnt_sad16x4_avg     ssse3/;
 659   specialize qw/aom_jnt_sad8x32_avg     ssse3/;
 660   specialize qw/aom_jnt_sad32x8_avg     ssse3/;
 661   specialize qw/aom_jnt_sad16x64_avg     ssse3/;
 662   specialize qw/aom_jnt_sad64x16_avg     ssse3/;
 663
 664   add_proto qw/unsigned int/, "aom_sad4xh", "const uint8_t *a, int a_stride, const uint8_t *b, int b_stride, int width, int height";
 665   add_proto qw/unsigned int/, "aom_sad8xh", "const uint8_t *a, int a_stride, const uint8_t *b, int b_stride, int width, int height";
 666   add_proto qw/unsigned int/, "aom_sad16xh", "const uint8_t *a, int a_stride, const uint8_t *b, int b_stride, int width, int height";
 667   add_proto qw/unsigned int/, "aom_sad32xh", "const uint8_t *a, int a_stride, const uint8_t *b, int b_stride, int width, int height";
 668   add_proto qw/unsigned int/, "aom_sad64xh", "const uint8_t *a, int a_stride, const uint8_t *b, int b_stride, int width, int height";
 669   add_proto qw/unsigned int/, "aom_sad128xh", "const uint8_t *a, int a_stride, const uint8_t *b, int b_stride, int width, int height";
 670
 671   specialize qw/aom_sad4xh   sse2/;
 672   specialize qw/aom_sad8xh   sse2/;
 673   specialize qw/aom_sad16xh  sse2/;
 674   specialize qw/aom_sad32xh  sse2/;
 675   specialize qw/aom_sad64xh  sse2/;
 676   specialize qw/aom_sad128xh sse2/;
 677
 678
 679     foreach (@block_sizes) {
 680       ($w, $h) = @$_;
 681       add_proto qw/unsigned int/, "aom_highbd_sad${w}x${h}", "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride";
 682       add_proto qw/unsigned int/, "aom_highbd_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";
 683       if ($w != 128 && $h != 128 && $w != 4) {
 684         specialize "aom_highbd_sad${w}x${h}", qw/sse2/;
 685         specialize "aom_highbd_sad${w}x${h}_avg", qw/sse2/;
 686       }
 687       add_proto qw/unsigned int/, "aom_highbd_jnt_sad${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";
 688     }
 689     specialize qw/aom_highbd_sad128x128 avx2/;
 690     specialize qw/aom_highbd_sad128x64  avx2/;
 691     specialize qw/aom_highbd_sad64x128  avx2/;
 692     specialize qw/aom_highbd_sad64x64   avx2 sse2/;
 693     specialize qw/aom_highbd_sad64x32   avx2 sse2/;
 694     specialize qw/aom_highbd_sad32x64   avx2 sse2/;
 695     specialize qw/aom_highbd_sad32x32   avx2 sse2/;
 696     specialize qw/aom_highbd_sad32x16   avx2 sse2/;
 697     specialize qw/aom_highbd_sad16x32   avx2 sse2/;
 698     specialize qw/aom_highbd_sad16x16   avx2 sse2/;
 699     specialize qw/aom_highbd_sad16x8    avx2 sse2/;
 700     specialize qw/aom_highbd_sad8x4     sse2/;
 701
 702     specialize qw/aom_highbd_sad128x128_avg avx2/;
 703     specialize qw/aom_highbd_sad128x64_avg  avx2/;
 704     specialize qw/aom_highbd_sad64x128_avg  avx2/;
 705     specialize qw/aom_highbd_sad64x64_avg   avx2 sse2/;
 706     specialize qw/aom_highbd_sad64x32_avg   avx2 sse2/;
 707     specialize qw/aom_highbd_sad32x64_avg   avx2 sse2/;
 708     specialize qw/aom_highbd_sad32x32_avg   avx2 sse2/;
 709     specialize qw/aom_highbd_sad32x16_avg   avx2 sse2/;
 710     specialize qw/aom_highbd_sad16x32_avg   avx2 sse2/;
 711     specialize qw/aom_highbd_sad16x16_avg   avx2 sse2/;
 712     specialize qw/aom_highbd_sad16x8_avg    avx2 sse2/;
 713     specialize qw/aom_highbd_sad8x4_avg     sse2/;
 714
 715     specialize qw/aom_highbd_sad16x4       sse2/;
 716     specialize qw/aom_highbd_sad8x32       sse2/;
 717     specialize qw/aom_highbd_sad32x8       sse2/;
 718     specialize qw/aom_highbd_sad16x64      sse2/;
 719     specialize qw/aom_highbd_sad64x16      sse2/;
 720
 721     specialize qw/aom_highbd_sad16x4_avg   sse2/;
 722     specialize qw/aom_highbd_sad8x32_avg   sse2/;
 723     specialize qw/aom_highbd_sad32x8_avg   sse2/;
 724     specialize qw/aom_highbd_sad16x64_avg  sse2/;
 725     specialize qw/aom_highbd_sad64x16_avg  sse2/;
 726
 727   #
 728   # Masked SAD
 729   #
 730   foreach (@block_sizes) {
 731     ($w, $h) = @$_;
 732     add_proto qw/unsigned int/, "aom_masked_sad${w}x${h}", "const uint8_t *src, int src_stride, const uint8_t *ref, int ref_stride, const uint8_t *second_pred, const uint8_t *msk, int msk_stride, int invert_mask";
 733     specialize "aom_masked_sad${w}x${h}", qw/ssse3/;
 734   }
 735
 736
 737     foreach (@block_sizes) {
 738       ($w, $h) = @$_;
 739       add_proto qw/unsigned int/, "aom_highbd_masked_sad${w}x${h}", "const uint8_t *src8, int src_stride, const uint8_t *ref8, int ref_stride, const uint8_t *second_pred8, const uint8_t *msk, int msk_stride, int invert_mask";
 740       specialize "aom_highbd_masked_sad${w}x${h}", qw/ssse3/;
 741     }
 742
 743
 744   #
 745   # OBMC SAD
 746   #
 747   foreach (@block_sizes) {
 748     ($w, $h) = @$_;
 749     add_proto qw/unsigned int/, "aom_obmc_sad${w}x${h}", "const uint8_t *pre, int pre_stride, const int32_t *wsrc, const int32_t *mask";
 750     if (! (($w == 128 && $h == 32) || ($w == 32 && $h == 128))) {
 751        specialize "aom_obmc_sad${w}x${h}", qw/sse4_1/;
 752     }
 753   }
 754
 755
 756     foreach (@block_sizes) {
 757       ($w, $h) = @$_;
 758       add_proto qw/unsigned int/, "aom_highbd_obmc_sad${w}x${h}", "const uint8_t *pre, int pre_stride, const int32_t *wsrc, const int32_t *mask";
 759       if (! (($w == 128 && $h == 32) || ($w == 32 && $h == 128))) {
 760         specialize "aom_highbd_obmc_sad${w}x${h}", qw/sse4_1/;
 761       }
 762     }
 763
 764
 765   #
 766   # Multi-block SAD, comparing a reference to N blocks 1 pixel apart horizontally
 767   #
 768   # Blocks of 3
 769   foreach $s (@block_widths) {
 770     add_proto qw/void/, "aom_sad${s}x${s}x3", "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, uint32_t *sad_array";
 771   }
 772   specialize qw/aom_sad64x64x3            msa/;
 773   specialize qw/aom_sad32x32x3            msa/;
 774   specialize qw/aom_sad16x16x3 sse3 ssse3 msa/;
 775   specialize qw/aom_sad8x8x3   sse3       msa/;
 776   specialize qw/aom_sad4x4x3   sse3       msa/;
 777
 778   add_proto qw/void/, "aom_sad16x8x3", "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, uint32_t *sad_array";
 779   specialize qw/aom_sad16x8x3 sse3 ssse3 msa/;
 780   add_proto qw/void/, "aom_sad8x16x3", "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, uint32_t *sad_array";
 781   specialize qw/aom_sad8x16x3 sse3 msa/;
 782
 783   # Blocks of 8
 784   foreach $s (@block_widths) {
 785     add_proto qw/void/, "aom_sad${s}x${s}x8", "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, uint32_t *sad_array";
 786   }
 787   specialize qw/aom_sad64x64x8        msa/;
 788   specialize qw/aom_sad32x32x8        msa/;
 789   specialize qw/aom_sad16x16x8 sse4_1 msa/;
 790   specialize qw/aom_sad8x8x8   sse4_1 msa/;
 791   specialize qw/aom_sad4x4x8   sse4_1 msa/;
 792
 793   add_proto qw/void/, "aom_sad16x8x8", "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, uint32_t *sad_array";
 794   specialize qw/aom_sad16x8x8 sse4_1 msa/;
 795   add_proto qw/void/, "aom_sad8x16x8", "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, uint32_t *sad_array";
 796   specialize qw/aom_sad8x16x8 sse4_1 msa/;
 797   add_proto qw/void/, "aom_sad8x4x8", "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, uint32_t *sad_array";
 798   specialize qw/aom_sad8x4x8 msa/;
 799   add_proto qw/void/, "aom_sad4x8x8", "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, uint32_t *sad_array";
 800   specialize qw/aom_sad4x8x8 msa/;
 801
 802
 803   foreach $s (@block_widths) {
 804     # Blocks of 3
 805     add_proto qw/void/, "aom_highbd_sad${s}x${s}x3", "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, uint32_t *sad_array";
 806     # Blocks of 8
 807     add_proto qw/void/, "aom_highbd_sad${s}x${s}x8", "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, uint32_t *sad_array";
 808   }
 809   # Blocks of 3
 810   add_proto qw/void/, "aom_highbd_sad16x8x3", "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, uint32_t *sad_array";
 811   add_proto qw/void/, "aom_highbd_sad8x16x3", "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, uint32_t *sad_array";
 812   # Blocks of 8
 813   add_proto qw/void/, "aom_highbd_sad16x8x8", "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, uint32_t *sad_array";
 814   add_proto qw/void/, "aom_highbd_sad8x16x8", "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, uint32_t *sad_array";
 815   add_proto qw/void/, "aom_highbd_sad8x4x8", "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, uint32_t *sad_array";
 816   add_proto qw/void/, "aom_highbd_sad4x8x8", "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, uint32_t *sad_array";
 817
 818
 819   #
 820   # Multi-block SAD, comparing a reference to N independent blocks
 821   #
 822   foreach (@block_sizes) {
 823     ($w, $h) = @$_;
 824     add_proto qw/void/, "aom_sad${w}x${h}x4d", "const uint8_t *src_ptr, int src_stride, const uint8_t * const ref_ptr[], int ref_stride, uint32_t *sad_array";
 825   }
 826
 827   specialize qw/aom_sad128x128x4d avx2          sse2/;
 828   specialize qw/aom_sad128x64x4d  avx2          sse2/;
 829   specialize qw/aom_sad64x128x4d  avx2          sse2/;
 830   specialize qw/aom_sad64x64x4d   avx2 neon msa sse2/;
 831   specialize qw/aom_sad64x32x4d   avx2      msa sse2/;
 832   specialize qw/aom_sad32x64x4d   avx2      msa sse2/;
 833   specialize qw/aom_sad32x32x4d   avx2 neon msa sse2/;
 834   specialize qw/aom_sad32x16x4d             msa sse2/;
 835   specialize qw/aom_sad16x32x4d             msa sse2/;
 836   specialize qw/aom_sad16x16x4d        neon msa sse2/;
 837   specialize qw/aom_sad16x8x4d              msa sse2/;
 838   specialize qw/aom_sad8x16x4d              msa sse2/;
 839   specialize qw/aom_sad8x8x4d               msa sse2/;
 840   specialize qw/aom_sad8x4x4d               msa sse2/;
 841   specialize qw/aom_sad4x8x4d               msa sse2/;
 842   specialize qw/aom_sad4x4x4d               msa sse2/;
 843
 844   specialize qw/aom_sad4x16x4d  sse2/;
 845   specialize qw/aom_sad16x4x4d  sse2/;
 846   specialize qw/aom_sad8x32x4d  sse2/;
 847   specialize qw/aom_sad32x8x4d  sse2/;
 848   specialize qw/aom_sad16x64x4d sse2/;
 849   specialize qw/aom_sad64x16x4d sse2/;
 850
 851   #
 852   # Multi-block SAD, comparing a reference to N independent blocks
 853   #
 854   foreach (@block_sizes) {
 855     ($w, $h) = @$_;
 856     add_proto qw/void/, "aom_highbd_sad${w}x${h}x4d", "const uint8_t *src_ptr, int src_stride, const uint8_t * const ref_ptr[], int ref_stride, uint32_t *sad_array";
 857     if ($w != 128 && $h != 128) {
 858       specialize "aom_highbd_sad${w}x${h}x4d", qw/sse2/;
 859     }
 860   }
 861   specialize qw/aom_highbd_sad128x128x4d avx2/;
 862   specialize qw/aom_highbd_sad128x64x4d  avx2/;
 863   specialize qw/aom_highbd_sad64x128x4d  avx2/;
 864   specialize qw/aom_highbd_sad64x64x4d   sse2 avx2/;
 865   specialize qw/aom_highbd_sad64x32x4d   sse2 avx2/;
 866   specialize qw/aom_highbd_sad32x64x4d   sse2 avx2/;
 867   specialize qw/aom_highbd_sad32x32x4d   sse2 avx2/;
 868   specialize qw/aom_highbd_sad32x16x4d   sse2 avx2/;
 869   specialize qw/aom_highbd_sad16x32x4d   sse2 avx2/;
 870   specialize qw/aom_highbd_sad16x16x4d   sse2 avx2/;
 871   specialize qw/aom_highbd_sad16x8x4d    sse2 avx2/;
 872   specialize qw/aom_highbd_sad8x16x4d    sse2/;
 873   specialize qw/aom_highbd_sad8x8x4d     sse2/;
 874   specialize qw/aom_highbd_sad8x4x4d     sse2/;
 875   specialize qw/aom_highbd_sad4x8x4d     sse2/;
 876   specialize qw/aom_highbd_sad4x4x4d     sse2/;
 877
 878   specialize qw/aom_highbd_sad4x16x4d  sse2/;
 879   specialize qw/aom_highbd_sad16x4x4d  sse2/;
 880   specialize qw/aom_highbd_sad8x32x4d  sse2/;
 881   specialize qw/aom_highbd_sad32x8x4d  sse2/;
 882   specialize qw/aom_highbd_sad16x64x4d sse2/;
 883   specialize qw/aom_highbd_sad64x16x4d sse2/;
 884
 885
 886   #
 887   # Structured Similarity (SSIM)
 888   #
 889   if (aom_config("CONFIG_INTERNAL_STATS") eq "yes") {
 890     add_proto qw/void aom_ssim_parms_8x8/, "const uint8_t *s, int sp, const uint8_t *r, int rp, uint32_t *sum_s, uint32_t *sum_r, uint32_t *sum_sq_s, uint32_t *sum_sq_r, uint32_t *sum_sxr";
 891     specialize qw/aom_ssim_parms_8x8/, "$sse2_x86_64";
 892
 893     add_proto qw/void aom_ssim_parms_16x16/, "const uint8_t *s, int sp, const uint8_t *r, int rp, uint32_t *sum_s, uint32_t *sum_r, uint32_t *sum_sq_s, uint32_t *sum_sq_r, uint32_t *sum_sxr";
 894     specialize qw/aom_ssim_parms_16x16/, "$sse2_x86_64";
 895
 896     add_proto qw/void aom_highbd_ssim_parms_8x8/, "const uint16_t *s, int sp, const uint16_t *r, int rp, uint32_t *sum_s, uint32_t *sum_r, uint32_t *sum_sq_s, uint32_t *sum_sq_r, uint32_t *sum_sxr";
 897
 898   }
 899 }  # CONFIG_AV1_ENCODER
 900
 901 if (aom_config("CONFIG_AV1_ENCODER") eq "yes") {
 902
 903   #
 904   # Specialty Variance
 905   #
 906   add_proto qw/void aom_get16x16var/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, int *sum";
 907
 908   add_proto qw/void aom_get8x8var/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, int *sum";
 909
 910   specialize qw/aom_get16x16var sse2 avx2 neon msa/;
 911   specialize qw/aom_get8x8var   sse2      neon msa/;
 912
 913
 914   add_proto qw/unsigned int aom_mse16x16/, "const uint8_t *src_ptr, int  source_stride, const uint8_t *ref_ptr, int  recon_stride, unsigned int *sse";
 915   add_proto qw/unsigned int aom_mse16x8/, "const uint8_t *src_ptr, int  source_stride, const uint8_t *ref_ptr, int  recon_stride, unsigned int *sse";
 916   add_proto qw/unsigned int aom_mse8x16/, "const uint8_t *src_ptr, int  source_stride, const uint8_t *ref_ptr, int  recon_stride, unsigned int *sse";
 917   add_proto qw/unsigned int aom_mse8x8/, "const uint8_t *src_ptr, int  source_stride, const uint8_t *ref_ptr, int  recon_stride, unsigned int *sse";
 918
 919   specialize qw/aom_mse16x16          sse2 avx2 neon msa/;
 920   specialize qw/aom_mse16x8           sse2           msa/;
 921   specialize qw/aom_mse8x16           sse2           msa/;
 922   specialize qw/aom_mse8x8            sse2           msa/;
 923
 924     foreach $bd (8, 10, 12) {
 925       add_proto qw/void/, "aom_highbd_${bd}_get16x16var", "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, int *sum";
 926       add_proto qw/void/, "aom_highbd_${bd}_get8x8var", "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, int *sum";
 927
 928       add_proto qw/unsigned int/, "aom_highbd_${bd}_mse16x16", "const uint8_t *src_ptr, int  source_stride, const uint8_t *ref_ptr, int  recon_stride, unsigned int *sse";
 929       add_proto qw/unsigned int/, "aom_highbd_${bd}_mse16x8", "const uint8_t *src_ptr, int  source_stride, const uint8_t *ref_ptr, int  recon_stride, unsigned int *sse";
 930       add_proto qw/unsigned int/, "aom_highbd_${bd}_mse8x16", "const uint8_t *src_ptr, int  source_stride, const uint8_t *ref_ptr, int  recon_stride, unsigned int *sse";
 931       add_proto qw/unsigned int/, "aom_highbd_${bd}_mse8x8", "const uint8_t *src_ptr, int  source_stride, const uint8_t *ref_ptr, int  recon_stride, unsigned int *sse";
 932
 933       specialize "aom_highbd_${bd}_mse16x16", qw/sse2/;
 934       specialize "aom_highbd_${bd}_mse8x8", qw/sse2/;
 935     }
 936
 937
 938   #
 939   #
 940   #
 941   add_proto qw/void aom_upsampled_pred/, "uint8_t *comp_pred, int width, int height, int subsample_x_q3, int subsample_y_q3, const uint8_t *ref, int ref_stride";
 942   specialize qw/aom_upsampled_pred sse2/;
 943
 944   add_proto qw/void aom_comp_avg_upsampled_pred/, "uint8_t *comp_pred, const uint8_t *pred, int width, int height, int subsample_x_q3, int subsample_y_q3, const uint8_t *ref, int ref_stride";
 945   specialize qw/aom_comp_avg_upsampled_pred sse2/;
 946
 947   add_proto qw/void aom_jnt_comp_avg_upsampled_pred/, "uint8_t *comp_pred, const uint8_t *pred, int width, int height, int subsample_x_q3, int subsample_y_q3, const uint8_t *ref, int ref_stride, const JNT_COMP_PARAMS *jcp_param";
 948   specialize qw/aom_jnt_comp_avg_upsampled_pred ssse3/;
 949
 950
 951   add_proto qw/void aom_highbd_upsampled_pred/, "uint16_t *comp_pred, int width, int height, int subsample_x_q3, int subsample_y_q3, const uint8_t *ref8, int ref_stride, int bd";
 952   specialize qw/aom_highbd_upsampled_pred sse2/;
 953   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";
 954   specialize qw/aom_highbd_comp_avg_upsampled_pred sse2/;
 955
 956   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";
 957   specialize qw/aom_highbd_jnt_comp_avg_upsampled_pred sse2/;
 958
 959
 960   #
 961   #
 962   #
 963   add_proto qw/unsigned int aom_get_mb_ss/, "const int16_t *";
 964   add_proto qw/unsigned int aom_get4x4sse_cs/, "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int ref_stride";
 965
 966   specialize qw/aom_get_mb_ss sse2 msa/;
 967   specialize qw/aom_get4x4sse_cs neon msa/;
 968
 969   #
 970   # Variance / Subpixel Variance / Subpixel Avg Variance
 971   #
 972   add_proto qw/unsigned int/, "aom_variance2x2", "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
 973
 974   add_proto qw/unsigned int/, "aom_variance2x4", "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
 975
 976   add_proto qw/unsigned int/, "aom_variance4x2", "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
 977
 978   foreach (@block_sizes) {
 979     ($w, $h) = @$_;
 980     add_proto qw/unsigned int/, "aom_variance${w}x${h}", "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
 981     add_proto qw/uint32_t/, "aom_sub_pixel_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";
 982     add_proto qw/uint32_t/, "aom_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";
 983     add_proto qw/uint32_t/, "aom_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";
 984   }
 985   specialize qw/aom_variance128x128   sse2 avx2         /;
 986   specialize qw/aom_variance128x64    sse2 avx2         /;
 987   specialize qw/aom_variance64x128    sse2 avx2         /;
 988   specialize qw/aom_variance64x64     sse2 avx2 neon msa/;
 989   specialize qw/aom_variance64x32     sse2 avx2 neon msa/;
 990   specialize qw/aom_variance32x64     sse2      neon msa/;
 991   specialize qw/aom_variance32x32     sse2 avx2 neon msa/;
 992   specialize qw/aom_variance32x16     sse2 avx2 msa/;
 993   specialize qw/aom_variance16x32     sse2      msa/;
 994   specialize qw/aom_variance16x16     sse2 avx2 neon msa/;
 995   specialize qw/aom_variance16x8      sse2      neon msa/;
 996   specialize qw/aom_variance8x16      sse2      neon msa/;
 997   specialize qw/aom_variance8x8       sse2      neon msa/;
 998   specialize qw/aom_variance8x4       sse2           msa/;
 999   specialize qw/aom_variance4x8       sse2           msa/;
1000   specialize qw/aom_variance4x4       sse2           msa/;
1001
1002   specialize qw/aom_sub_pixel_variance64x64     avx2 neon msa sse2 ssse3/;
1003   specialize qw/aom_sub_pixel_variance64x32               msa sse2 ssse3/;
1004   specialize qw/aom_sub_pixel_variance32x64               msa sse2 ssse3/;
1005   specialize qw/aom_sub_pixel_variance32x32     avx2 neon msa sse2 ssse3/;
1006   specialize qw/aom_sub_pixel_variance32x16               msa sse2 ssse3/;
1007   specialize qw/aom_sub_pixel_variance16x32               msa sse2 ssse3/;
1008   specialize qw/aom_sub_pixel_variance16x16          neon msa sse2 ssse3/;
1009   specialize qw/aom_sub_pixel_variance16x8                msa sse2 ssse3/;
1010   specialize qw/aom_sub_pixel_variance8x16                msa sse2 ssse3/;
1011   specialize qw/aom_sub_pixel_variance8x8            neon msa sse2 ssse3/;
1012   specialize qw/aom_sub_pixel_variance8x4                 msa sse2 ssse3/;
1013   specialize qw/aom_sub_pixel_variance4x8                 msa sse2 ssse3/;
1014   specialize qw/aom_sub_pixel_variance4x4                 msa sse2 ssse3/;
1015
1016   specialize qw/aom_sub_pixel_avg_variance64x64 avx2 msa sse2 ssse3/;
1017   specialize qw/aom_sub_pixel_avg_variance64x32      msa sse2 ssse3/;
1018   specialize qw/aom_sub_pixel_avg_variance32x64      msa sse2 ssse3/;
1019   specialize qw/aom_sub_pixel_avg_variance32x32 avx2 msa sse2 ssse3/;
1020   specialize qw/aom_sub_pixel_avg_variance32x16      msa sse2 ssse3/;
1021   specialize qw/aom_sub_pixel_avg_variance16x32      msa sse2 ssse3/;
1022   specialize qw/aom_sub_pixel_avg_variance16x16      msa sse2 ssse3/;
1023   specialize qw/aom_sub_pixel_avg_variance16x8       msa sse2 ssse3/;
1024   specialize qw/aom_sub_pixel_avg_variance8x16       msa sse2 ssse3/;
1025   specialize qw/aom_sub_pixel_avg_variance8x8        msa sse2 ssse3/;
1026   specialize qw/aom_sub_pixel_avg_variance8x4        msa sse2 ssse3/;
1027   specialize qw/aom_sub_pixel_avg_variance4x8        msa sse2 ssse3/;
1028   specialize qw/aom_sub_pixel_avg_variance4x4        msa sse2 ssse3/;
1029
1030   specialize qw/aom_variance4x16 sse2/;
1031   specialize qw/aom_variance16x4 sse2/;
1032   specialize qw/aom_variance8x32 sse2/;
1033   specialize qw/aom_variance32x8 sse2/;
1034   specialize qw/aom_variance16x64 sse2/;
1035   specialize qw/aom_variance64x16 sse2/;
1036   specialize qw/aom_sub_pixel_variance4x16 sse2 ssse3/;
1037   specialize qw/aom_sub_pixel_variance16x4 sse2 ssse3/;
1038   specialize qw/aom_sub_pixel_variance8x32 sse2 ssse3/;
1039   specialize qw/aom_sub_pixel_variance32x8 sse2 ssse3/;
1040   specialize qw/aom_sub_pixel_variance16x64 sse2 ssse3/;
1041   specialize qw/aom_sub_pixel_variance64x16 sse2 ssse3/;
1042   specialize qw/aom_sub_pixel_avg_variance4x16 sse2 ssse3/;
1043   specialize qw/aom_sub_pixel_avg_variance16x4 sse2 ssse3/;
1044   specialize qw/aom_sub_pixel_avg_variance8x32 sse2 ssse3/;
1045   specialize qw/aom_sub_pixel_avg_variance32x8 sse2 ssse3/;
1046   specialize qw/aom_sub_pixel_avg_variance16x64 sse2 ssse3/;
1047   specialize qw/aom_sub_pixel_avg_variance64x16 sse2 ssse3/;
1048
1049   specialize qw/aom_jnt_sub_pixel_avg_variance64x64 ssse3/;
1050   specialize qw/aom_jnt_sub_pixel_avg_variance64x32 ssse3/;
1051   specialize qw/aom_jnt_sub_pixel_avg_variance32x64 ssse3/;
1052   specialize qw/aom_jnt_sub_pixel_avg_variance32x32 ssse3/;
1053   specialize qw/aom_jnt_sub_pixel_avg_variance32x16 ssse3/;
1054   specialize qw/aom_jnt_sub_pixel_avg_variance16x32 ssse3/;
1055   specialize qw/aom_jnt_sub_pixel_avg_variance16x16 ssse3/;
1056   specialize qw/aom_jnt_sub_pixel_avg_variance16x8  ssse3/;
1057   specialize qw/aom_jnt_sub_pixel_avg_variance8x16  ssse3/;
1058   specialize qw/aom_jnt_sub_pixel_avg_variance8x8   ssse3/;
1059   specialize qw/aom_jnt_sub_pixel_avg_variance8x4   ssse3/;
1060   specialize qw/aom_jnt_sub_pixel_avg_variance4x8   ssse3/;
1061   specialize qw/aom_jnt_sub_pixel_avg_variance4x4   ssse3/;
1062
1063   specialize qw/aom_jnt_sub_pixel_avg_variance4x16  ssse3/;
1064   specialize qw/aom_jnt_sub_pixel_avg_variance16x4  ssse3/;
1065   specialize qw/aom_jnt_sub_pixel_avg_variance8x32  ssse3/;
1066   specialize qw/aom_jnt_sub_pixel_avg_variance32x8  ssse3/;
1067   specialize qw/aom_jnt_sub_pixel_avg_variance16x64 ssse3/;
1068   specialize qw/aom_jnt_sub_pixel_avg_variance64x16 ssse3/;
1069
1070   specialize qw/aom_jnt_sub_pixel_avg_variance128x128  ssse3/;
1071   specialize qw/aom_jnt_sub_pixel_avg_variance128x64   ssse3/;
1072   specialize qw/aom_jnt_sub_pixel_avg_variance64x128   ssse3/;
1073
1074
1075   foreach $bd (8, 10, 12) {
1076     add_proto qw/unsigned int/, "aom_highbd_${bd}_variance2x2", "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
1077
1078     add_proto qw/unsigned int/, "aom_highbd_${bd}_variance2x4", "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
1079
1080     add_proto qw/unsigned int/, "aom_highbd_${bd}_variance4x2", "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
1081
1082     foreach (@block_sizes) {
1083       ($w, $h) = @$_;
1084       add_proto qw/unsigned int/, "aom_highbd_${bd}_variance${w}x${h}", "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
1085       add_proto qw/uint32_t/, "aom_highbd_${bd}_sub_pixel_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";
1086       add_proto qw/uint32_t/, "aom_highbd_${bd}_sub_pixel_avg_variance${w}x${h}", "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse, const uint8_t *second_pred";
1087       if ($w != 128 && $h != 128 && $w != 4 && $h != 4) {
1088         specialize "aom_highbd_${bd}_variance${w}x${h}", "sse2";
1089       }
1090       # TODO(david.barker): When ext-partition-types is enabled, we currently
1091       # don't have vectorized 4x16 highbd variance functions
1092       if ($w == 4 && $h == 4) {
1093           specialize "aom_highbd_${bd}_variance${w}x${h}", "sse4_1";
1094         }
1095       if ($w != 128 && $h != 128 && $w != 4) {
1096         specialize "aom_highbd_${bd}_sub_pixel_variance${w}x${h}", qw/sse2/;
1097         specialize "aom_highbd_${bd}_sub_pixel_avg_variance${w}x${h}", qw/sse2/;
1098       }
1099       if ($w == 4 && $h == 4) {
1100         specialize "aom_highbd_${bd}_sub_pixel_variance${w}x${h}", "sse4_1";
1101         specialize "aom_highbd_${bd}_sub_pixel_avg_variance${w}x${h}", "sse4_1";
1102       }
1103
1104       add_proto qw/uint32_t/, "aom_highbd_${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";
1105     }
1106   }
1107
1108   #
1109   # Masked Variance / Masked Subpixel Variance
1110   #
1111   foreach (@block_sizes) {
1112     ($w, $h) = @$_;
1113     add_proto qw/unsigned int/, "aom_masked_sub_pixel_variance${w}x${h}", "const uint8_t *src, int src_stride, int xoffset, int yoffset, const uint8_t *ref, int ref_stride, const uint8_t *second_pred, const uint8_t *msk, int msk_stride, int invert_mask, unsigned int *sse";
1114     specialize "aom_masked_sub_pixel_variance${w}x${h}", qw/ssse3/;
1115   }
1116
1117
1118     foreach $bd ("_8_", "_10_", "_12_") {
1119       foreach (@block_sizes) {
1120         ($w, $h) = @$_;
1121         add_proto qw/unsigned int/, "aom_highbd${bd}masked_sub_pixel_variance${w}x${h}", "const uint8_t *src, int src_stride, int xoffset, int yoffset, const uint8_t *ref, int ref_stride, const uint8_t *second_pred, const uint8_t *msk, int msk_stride, int invert_mask, unsigned int *sse";
1122         specialize "aom_highbd${bd}masked_sub_pixel_variance${w}x${h}", qw/ssse3/;
1123       }
1124     }
1125
1126
1127   #
1128   # OBMC Variance / OBMC Subpixel Variance
1129   #
1130   foreach (@block_sizes) {
1131     ($w, $h) = @$_;
1132     add_proto qw/unsigned int/, "aom_obmc_variance${w}x${h}", "const uint8_t *pre, int pre_stride, const int32_t *wsrc, const int32_t *mask, unsigned int *sse";
1133     add_proto qw/unsigned int/, "aom_obmc_sub_pixel_variance${w}x${h}", "const uint8_t *pre, int pre_stride, int xoffset, int yoffset, const int32_t *wsrc, const int32_t *mask, unsigned int *sse";
1134     specialize "aom_obmc_variance${w}x${h}", q/sse4_1/;
1135   }
1136
1137
1138     foreach $bd ("_", "_10_", "_12_") {
1139       foreach (@block_sizes) {
1140         ($w, $h) = @$_;
1141         add_proto qw/unsigned int/, "aom_highbd${bd}obmc_variance${w}x${h}", "const uint8_t *pre, int pre_stride, const int32_t *wsrc, const int32_t *mask, unsigned int *sse";
1142         add_proto qw/unsigned int/, "aom_highbd${bd}obmc_sub_pixel_variance${w}x${h}", "const uint8_t *pre, int pre_stride, int xoffset, int yoffset, const int32_t *wsrc, const int32_t *mask, unsigned int *sse";
1143         specialize "aom_highbd${bd}obmc_variance${w}x${h}", qw/sse4_1/;
1144       }
1145     }
1146
1147
1148   add_proto qw/uint32_t aom_sub_pixel_avg_variance64x64/, "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";
1149   specialize qw/aom_sub_pixel_avg_variance64x64 avx2 msa sse2 ssse3/;
1150
1151   add_proto qw/uint32_t aom_sub_pixel_avg_variance64x32/, "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";
1152   specialize qw/aom_sub_pixel_avg_variance64x32 msa sse2 ssse3/;
1153
1154   add_proto qw/uint32_t aom_sub_pixel_avg_variance32x64/, "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";
1155   specialize qw/aom_sub_pixel_avg_variance32x64 msa sse2 ssse3/;
1156
1157   add_proto qw/uint32_t aom_sub_pixel_avg_variance32x32/, "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";
1158   specialize qw/aom_sub_pixel_avg_variance32x32 avx2 msa sse2 ssse3/;
1159
1160   add_proto qw/uint32_t aom_sub_pixel_avg_variance32x16/, "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";
1161   specialize qw/aom_sub_pixel_avg_variance32x16 msa sse2 ssse3/;
1162
1163   add_proto qw/uint32_t aom_sub_pixel_avg_variance16x32/, "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";
1164   specialize qw/aom_sub_pixel_avg_variance16x32 msa sse2 ssse3/;
1165
1166   add_proto qw/uint32_t aom_sub_pixel_avg_variance16x16/, "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";
1167   specialize qw/aom_sub_pixel_avg_variance16x16 msa sse2 ssse3/;
1168
1169   add_proto qw/uint32_t aom_sub_pixel_avg_variance16x8/, "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";
1170   specialize qw/aom_sub_pixel_avg_variance16x8 msa sse2 ssse3/;
1171
1172   add_proto qw/uint32_t aom_sub_pixel_avg_variance8x16/, "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";
1173   specialize qw/aom_sub_pixel_avg_variance8x16 msa sse2 ssse3/;
1174
1175   add_proto qw/uint32_t aom_sub_pixel_avg_variance8x8/, "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";
1176   specialize qw/aom_sub_pixel_avg_variance8x8 msa sse2 ssse3/;
1177
1178   add_proto qw/uint32_t aom_sub_pixel_avg_variance8x4/, "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";
1179   specialize qw/aom_sub_pixel_avg_variance8x4 msa sse2 ssse3/;
1180
1181   add_proto qw/uint32_t aom_sub_pixel_avg_variance4x8/, "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";
1182   specialize qw/aom_sub_pixel_avg_variance4x8 msa sse2 ssse3/;
1183
1184   add_proto qw/uint32_t aom_sub_pixel_avg_variance4x4/, "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";
1185   specialize qw/aom_sub_pixel_avg_variance4x4 msa sse2 ssse3/;
1186   #
1187   # Specialty Subpixel
1188   #
1189   add_proto qw/uint32_t aom_variance_halfpixvar16x16_h/, "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int  ref_stride, uint32_t *sse";
1190   specialize qw/aom_variance_halfpixvar16x16_h sse2/;
1191
1192   add_proto qw/uint32_t aom_variance_halfpixvar16x16_v/, "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int  ref_stride, uint32_t *sse";
1193   specialize qw/aom_variance_halfpixvar16x16_v sse2/;
1194
1195   add_proto qw/uint32_t aom_variance_halfpixvar16x16_hv/, "const unsigned char *src_ptr, int source_stride, const unsigned char *ref_ptr, int  ref_stride, uint32_t *sse";
1196   specialize qw/aom_variance_halfpixvar16x16_hv sse2/;
1197
1198   #
1199   # Comp Avg
1200   #
1201   add_proto qw/void aom_comp_avg_pred/, "uint8_t *comp_pred, const uint8_t *pred, int width, int height, const uint8_t *ref, int ref_stride";
1202
1203   add_proto qw/void aom_jnt_comp_avg_pred/, "uint8_t *comp_pred, const uint8_t *pred, int width, int height, const uint8_t *ref, int ref_stride, const JNT_COMP_PARAMS *jcp_param";
1204   specialize qw/aom_jnt_comp_avg_pred ssse3/;
1205
1206
1207     add_proto qw/unsigned int aom_highbd_12_variance64x64/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
1208     specialize qw/aom_highbd_12_variance64x64 sse2/;
1209
1210     add_proto qw/unsigned int aom_highbd_12_variance64x32/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
1211     specialize qw/aom_highbd_12_variance64x32 sse2/;
1212
1213     add_proto qw/unsigned int aom_highbd_12_variance32x64/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
1214     specialize qw/aom_highbd_12_variance32x64 sse2/;
1215
1216     add_proto qw/unsigned int aom_highbd_12_variance32x32/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
1217     specialize qw/aom_highbd_12_variance32x32 sse2/;
1218
1219     add_proto qw/unsigned int aom_highbd_12_variance32x16/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
1220     specialize qw/aom_highbd_12_variance32x16 sse2/;
1221
1222     add_proto qw/unsigned int aom_highbd_12_variance16x32/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
1223     specialize qw/aom_highbd_12_variance16x32 sse2/;
1224
1225     add_proto qw/unsigned int aom_highbd_12_variance16x16/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
1226     specialize qw/aom_highbd_12_variance16x16 sse2/;
1227
1228     add_proto qw/unsigned int aom_highbd_12_variance16x8/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
1229     specialize qw/aom_highbd_12_variance16x8 sse2/;
1230
1231     add_proto qw/unsigned int aom_highbd_12_variance8x16/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
1232     specialize qw/aom_highbd_12_variance8x16 sse2/;
1233
1234     add_proto qw/unsigned int aom_highbd_12_variance8x8/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
1235     specialize qw/aom_highbd_12_variance8x8 sse2/;
1236
1237     add_proto qw/unsigned int aom_highbd_12_variance8x4/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
1238     add_proto qw/unsigned int aom_highbd_12_variance4x8/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
1239     add_proto qw/unsigned int aom_highbd_12_variance4x4/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
1240
1241     add_proto qw/unsigned int aom_highbd_10_variance64x64/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
1242     specialize qw/aom_highbd_10_variance64x64 sse2/;
1243
1244     add_proto qw/unsigned int aom_highbd_10_variance64x32/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
1245     specialize qw/aom_highbd_10_variance64x32 sse2/;
1246
1247     add_proto qw/unsigned int aom_highbd_10_variance32x64/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
1248     specialize qw/aom_highbd_10_variance32x64 sse2/;
1249
1250     add_proto qw/unsigned int aom_highbd_10_variance32x32/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
1251     specialize qw/aom_highbd_10_variance32x32 sse2/;
1252
1253     add_proto qw/unsigned int aom_highbd_10_variance32x16/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
1254     specialize qw/aom_highbd_10_variance32x16 sse2/;
1255
1256     add_proto qw/unsigned int aom_highbd_10_variance16x32/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
1257     specialize qw/aom_highbd_10_variance16x32 sse2/;
1258
1259     add_proto qw/unsigned int aom_highbd_10_variance16x16/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
1260     specialize qw/aom_highbd_10_variance16x16 sse2/;
1261
1262     add_proto qw/unsigned int aom_highbd_10_variance16x8/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
1263     specialize qw/aom_highbd_10_variance16x8 sse2/;
1264
1265     add_proto qw/unsigned int aom_highbd_10_variance8x16/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
1266     specialize qw/aom_highbd_10_variance8x16 sse2/;
1267
1268     add_proto qw/unsigned int aom_highbd_10_variance8x8/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
1269     specialize qw/aom_highbd_10_variance8x8 sse2/;
1270
1271     add_proto qw/unsigned int aom_highbd_10_variance8x4/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
1272     add_proto qw/unsigned int aom_highbd_10_variance4x8/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
1273     add_proto qw/unsigned int aom_highbd_10_variance4x4/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
1274
1275     add_proto qw/unsigned int aom_highbd_8_variance64x64/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
1276     specialize qw/aom_highbd_8_variance64x64 sse2/;
1277
1278     add_proto qw/unsigned int aom_highbd_8_variance64x32/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
1279     specialize qw/aom_highbd_8_variance64x32 sse2/;
1280
1281     add_proto qw/unsigned int aom_highbd_8_variance32x64/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
1282     specialize qw/aom_highbd_8_variance32x64 sse2/;
1283
1284     add_proto qw/unsigned int aom_highbd_8_variance32x32/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
1285     specialize qw/aom_highbd_8_variance32x32 sse2/;
1286
1287     add_proto qw/unsigned int aom_highbd_8_variance32x16/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
1288     specialize qw/aom_highbd_8_variance32x16 sse2/;
1289
1290     add_proto qw/unsigned int aom_highbd_8_variance16x32/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
1291     specialize qw/aom_highbd_8_variance16x32 sse2/;
1292
1293     add_proto qw/unsigned int aom_highbd_8_variance16x16/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
1294     specialize qw/aom_highbd_8_variance16x16 sse2/;
1295
1296     add_proto qw/unsigned int aom_highbd_8_variance16x8/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
1297     specialize qw/aom_highbd_8_variance16x8 sse2/;
1298
1299     add_proto qw/unsigned int aom_highbd_8_variance8x16/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
1300     specialize qw/aom_highbd_8_variance8x16 sse2/;
1301
1302     add_proto qw/unsigned int aom_highbd_8_variance8x8/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
1303     specialize qw/aom_highbd_8_variance8x8 sse2/;
1304
1305     add_proto qw/unsigned int aom_highbd_8_variance8x4/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
1306     add_proto qw/unsigned int aom_highbd_8_variance4x8/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
1307     add_proto qw/unsigned int aom_highbd_8_variance4x4/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
1308
1309     add_proto qw/void aom_highbd_8_get16x16var/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, int *sum";
1310     add_proto qw/void aom_highbd_8_get8x8var/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, int *sum";
1311
1312     add_proto qw/void aom_highbd_10_get16x16var/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, int *sum";
1313     add_proto qw/void aom_highbd_10_get8x8var/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, int *sum";
1314
1315     add_proto qw/void aom_highbd_12_get16x16var/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, int *sum";
1316     add_proto qw/void aom_highbd_12_get8x8var/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, int *sum";
1317
1318     add_proto qw/unsigned int aom_highbd_8_mse16x16/, "const uint8_t *src_ptr, int  source_stride, const uint8_t *ref_ptr, int  recon_stride, unsigned int *sse";
1319     specialize qw/aom_highbd_8_mse16x16 sse2/;
1320
1321     add_proto qw/unsigned int aom_highbd_8_mse16x8/, "const uint8_t *src_ptr, int  source_stride, const uint8_t *ref_ptr, int  recon_stride, unsigned int *sse";
1322     add_proto qw/unsigned int aom_highbd_8_mse8x16/, "const uint8_t *src_ptr, int  source_stride, const uint8_t *ref_ptr, int  recon_stride, unsigned int *sse";
1323     add_proto qw/unsigned int aom_highbd_8_mse8x8/, "const uint8_t *src_ptr, int  source_stride, const uint8_t *ref_ptr, int  recon_stride, unsigned int *sse";
1324     specialize qw/aom_highbd_8_mse8x8 sse2/;
1325
1326     add_proto qw/unsigned int aom_highbd_10_mse16x16/, "const uint8_t *src_ptr, int  source_stride, const uint8_t *ref_ptr, int  recon_stride, unsigned int *sse";
1327     specialize qw/aom_highbd_10_mse16x16 sse2/;
1328
1329     add_proto qw/unsigned int aom_highbd_10_mse16x8/, "const uint8_t *src_ptr, int  source_stride, const uint8_t *ref_ptr, int  recon_stride, unsigned int *sse";
1330     add_proto qw/unsigned int aom_highbd_10_mse8x16/, "const uint8_t *src_ptr, int  source_stride, const uint8_t *ref_ptr, int  recon_stride, unsigned int *sse";
1331     add_proto qw/unsigned int aom_highbd_10_mse8x8/, "const uint8_t *src_ptr, int  source_stride, const uint8_t *ref_ptr, int  recon_stride, unsigned int *sse";
1332     specialize qw/aom_highbd_10_mse8x8 sse2/;
1333
1334     add_proto qw/unsigned int aom_highbd_12_mse16x16/, "const uint8_t *src_ptr, int  source_stride, const uint8_t *ref_ptr, int  recon_stride, unsigned int *sse";
1335     specialize qw/aom_highbd_12_mse16x16 sse2/;
1336
1337     add_proto qw/unsigned int aom_highbd_12_mse16x8/, "const uint8_t *src_ptr, int  source_stride, const uint8_t *ref_ptr, int  recon_stride, unsigned int *sse";
1338     add_proto qw/unsigned int aom_highbd_12_mse8x16/, "const uint8_t *src_ptr, int  source_stride, const uint8_t *ref_ptr, int  recon_stride, unsigned int *sse";
1339     add_proto qw/unsigned int aom_highbd_12_mse8x8/, "const uint8_t *src_ptr, int  source_stride, const uint8_t *ref_ptr, int  recon_stride, unsigned int *sse";
1340     specialize qw/aom_highbd_12_mse8x8 sse2/;
1341
1342     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";
1343
1344     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";
1345     specialize qw/aom_highbd_jnt_comp_avg_pred c/;
1346
1347     #
1348     # Subpixel Variance
1349     #
1350     add_proto qw/uint32_t aom_highbd_12_sub_pixel_variance64x64/, "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
1351     specialize qw/aom_highbd_12_sub_pixel_variance64x64 sse2/;
1352
1353     add_proto qw/uint32_t aom_highbd_12_sub_pixel_variance64x32/, "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
1354     specialize qw/aom_highbd_12_sub_pixel_variance64x32 sse2/;
1355
1356     add_proto qw/uint32_t aom_highbd_12_sub_pixel_variance32x64/, "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
1357     specialize qw/aom_highbd_12_sub_pixel_variance32x64 sse2/;
1358
1359     add_proto qw/uint32_t aom_highbd_12_sub_pixel_variance32x32/, "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
1360     specialize qw/aom_highbd_12_sub_pixel_variance32x32 sse2/;
1361
1362     add_proto qw/uint32_t aom_highbd_12_sub_pixel_variance32x16/, "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
1363     specialize qw/aom_highbd_12_sub_pixel_variance32x16 sse2/;
1364
1365     add_proto qw/uint32_t aom_highbd_12_sub_pixel_variance16x32/, "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
1366     specialize qw/aom_highbd_12_sub_pixel_variance16x32 sse2/;
1367
1368     add_proto qw/uint32_t aom_highbd_12_sub_pixel_variance16x16/, "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
1369     specialize qw/aom_highbd_12_sub_pixel_variance16x16 sse2/;
1370
1371     add_proto qw/uint32_t aom_highbd_12_sub_pixel_variance16x8/, "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
1372     specialize qw/aom_highbd_12_sub_pixel_variance16x8 sse2/;
1373
1374     add_proto qw/uint32_t aom_highbd_12_sub_pixel_variance8x16/, "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
1375     specialize qw/aom_highbd_12_sub_pixel_variance8x16 sse2/;
1376
1377     add_proto qw/uint32_t aom_highbd_12_sub_pixel_variance8x8/, "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
1378     specialize qw/aom_highbd_12_sub_pixel_variance8x8 sse2/;
1379
1380     add_proto qw/uint32_t aom_highbd_12_sub_pixel_variance8x4/, "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
1381     specialize qw/aom_highbd_12_sub_pixel_variance8x4 sse2/;
1382
1383     add_proto qw/uint32_t aom_highbd_12_sub_pixel_variance4x8/, "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
1384     add_proto qw/uint32_t aom_highbd_12_sub_pixel_variance4x4/, "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
1385
1386     add_proto qw/uint32_t aom_highbd_10_sub_pixel_variance64x64/, "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
1387     specialize qw/aom_highbd_10_sub_pixel_variance64x64 sse2/;
1388
1389     add_proto qw/uint32_t aom_highbd_10_sub_pixel_variance64x32/, "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
1390     specialize qw/aom_highbd_10_sub_pixel_variance64x32 sse2/;
1391
1392     add_proto qw/uint32_t aom_highbd_10_sub_pixel_variance32x64/, "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
1393     specialize qw/aom_highbd_10_sub_pixel_variance32x64 sse2/;
1394
1395     add_proto qw/uint32_t aom_highbd_10_sub_pixel_variance32x32/, "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
1396     specialize qw/aom_highbd_10_sub_pixel_variance32x32 sse2/;
1397
1398     add_proto qw/uint32_t aom_highbd_10_sub_pixel_variance32x16/, "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
1399     specialize qw/aom_highbd_10_sub_pixel_variance32x16 sse2/;
1400
1401     add_proto qw/uint32_t aom_highbd_10_sub_pixel_variance16x32/, "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
1402     specialize qw/aom_highbd_10_sub_pixel_variance16x32 sse2/;
1403
1404     add_proto qw/uint32_t aom_highbd_10_sub_pixel_variance16x16/, "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
1405     specialize qw/aom_highbd_10_sub_pixel_variance16x16 sse2/;
1406
1407     add_proto qw/uint32_t aom_highbd_10_sub_pixel_variance16x8/, "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
1408     specialize qw/aom_highbd_10_sub_pixel_variance16x8 sse2/;
1409
1410     add_proto qw/uint32_t aom_highbd_10_sub_pixel_variance8x16/, "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
1411     specialize qw/aom_highbd_10_sub_pixel_variance8x16 sse2/;
1412
1413     add_proto qw/uint32_t aom_highbd_10_sub_pixel_variance8x8/, "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
1414     specialize qw/aom_highbd_10_sub_pixel_variance8x8 sse2/;
1415
1416     add_proto qw/uint32_t aom_highbd_10_sub_pixel_variance8x4/, "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
1417     specialize qw/aom_highbd_10_sub_pixel_variance8x4 sse2/;
1418
1419     add_proto qw/uint32_t aom_highbd_10_sub_pixel_variance4x8/, "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
1420     add_proto qw/uint32_t aom_highbd_10_sub_pixel_variance4x4/, "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
1421
1422     add_proto qw/uint32_t aom_highbd_8_sub_pixel_variance64x64/, "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
1423     specialize qw/aom_highbd_8_sub_pixel_variance64x64 sse2/;
1424
1425     add_proto qw/uint32_t aom_highbd_8_sub_pixel_variance64x32/, "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
1426     specialize qw/aom_highbd_8_sub_pixel_variance64x32 sse2/;
1427
1428     add_proto qw/uint32_t aom_highbd_8_sub_pixel_variance32x64/, "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
1429     specialize qw/aom_highbd_8_sub_pixel_variance32x64 sse2/;
1430
1431     add_proto qw/uint32_t aom_highbd_8_sub_pixel_variance32x32/, "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
1432     specialize qw/aom_highbd_8_sub_pixel_variance32x32 sse2/;
1433
1434     add_proto qw/uint32_t aom_highbd_8_sub_pixel_variance32x16/, "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
1435     specialize qw/aom_highbd_8_sub_pixel_variance32x16 sse2/;
1436
1437     add_proto qw/uint32_t aom_highbd_8_sub_pixel_variance16x32/, "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
1438     specialize qw/aom_highbd_8_sub_pixel_variance16x32 sse2/;
1439
1440     add_proto qw/uint32_t aom_highbd_8_sub_pixel_variance16x16/, "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
1441     specialize qw/aom_highbd_8_sub_pixel_variance16x16 sse2/;
1442
1443     add_proto qw/uint32_t aom_highbd_8_sub_pixel_variance16x8/, "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
1444     specialize qw/aom_highbd_8_sub_pixel_variance16x8 sse2/;
1445
1446     add_proto qw/uint32_t aom_highbd_8_sub_pixel_variance8x16/, "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
1447     specialize qw/aom_highbd_8_sub_pixel_variance8x16 sse2/;
1448
1449     add_proto qw/uint32_t aom_highbd_8_sub_pixel_variance8x8/, "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
1450     specialize qw/aom_highbd_8_sub_pixel_variance8x8 sse2/;
1451
1452     add_proto qw/uint32_t aom_highbd_8_sub_pixel_variance8x4/, "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
1453     specialize qw/aom_highbd_8_sub_pixel_variance8x4 sse2/;
1454
1455     add_proto qw/uint32_t aom_highbd_8_sub_pixel_variance4x8/, "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
1456     add_proto qw/uint32_t aom_highbd_8_sub_pixel_variance4x4/, "const uint8_t *src_ptr, int source_stride, int xoffset, int  yoffset, const uint8_t *ref_ptr, int ref_stride, uint32_t *sse";
1457
1458     add_proto qw/uint32_t aom_highbd_12_sub_pixel_avg_variance64x64/, "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";
1459     specialize qw/aom_highbd_12_sub_pixel_avg_variance64x64 sse2/;
1460
1461     add_proto qw/uint32_t aom_highbd_12_sub_pixel_avg_variance64x32/, "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";
1462     specialize qw/aom_highbd_12_sub_pixel_avg_variance64x32 sse2/;
1463
1464     add_proto qw/uint32_t aom_highbd_12_sub_pixel_avg_variance32x64/, "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";
1465     specialize qw/aom_highbd_12_sub_pixel_avg_variance32x64 sse2/;
1466
1467     add_proto qw/uint32_t aom_highbd_12_sub_pixel_avg_variance32x32/, "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";
1468     specialize qw/aom_highbd_12_sub_pixel_avg_variance32x32 sse2/;
1469
1470     add_proto qw/uint32_t aom_highbd_12_sub_pixel_avg_variance32x16/, "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";
1471     specialize qw/aom_highbd_12_sub_pixel_avg_variance32x16 sse2/;
1472
1473     add_proto qw/uint32_t aom_highbd_12_sub_pixel_avg_variance16x32/, "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";
1474     specialize qw/aom_highbd_12_sub_pixel_avg_variance16x32 sse2/;
1475
1476     add_proto qw/uint32_t aom_highbd_12_sub_pixel_avg_variance16x16/, "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";
1477     specialize qw/aom_highbd_12_sub_pixel_avg_variance16x16 sse2/;
1478
1479     add_proto qw/uint32_t aom_highbd_12_sub_pixel_avg_variance16x8/, "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";
1480     specialize qw/aom_highbd_12_sub_pixel_avg_variance16x8 sse2/;
1481
1482     add_proto qw/uint32_t aom_highbd_12_sub_pixel_avg_variance8x16/, "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";
1483     specialize qw/aom_highbd_12_sub_pixel_avg_variance8x16 sse2/;
1484
1485     add_proto qw/uint32_t aom_highbd_12_sub_pixel_avg_variance8x8/, "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";
1486     specialize qw/aom_highbd_12_sub_pixel_avg_variance8x8 sse2/;
1487
1488     add_proto qw/uint32_t aom_highbd_12_sub_pixel_avg_variance8x4/, "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";
1489     specialize qw/aom_highbd_12_sub_pixel_avg_variance8x4 sse2/;
1490
1491     add_proto qw/uint32_t aom_highbd_12_sub_pixel_avg_variance4x8/, "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";
1492     add_proto qw/uint32_t aom_highbd_12_sub_pixel_avg_variance4x4/, "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";
1493
1494     add_proto qw/uint32_t aom_highbd_10_sub_pixel_avg_variance64x64/, "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";
1495     specialize qw/aom_highbd_10_sub_pixel_avg_variance64x64 sse2/;
1496
1497     add_proto qw/uint32_t aom_highbd_10_sub_pixel_avg_variance64x32/, "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";
1498     specialize qw/aom_highbd_10_sub_pixel_avg_variance64x32 sse2/;
1499
1500     add_proto qw/uint32_t aom_highbd_10_sub_pixel_avg_variance32x64/, "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";
1501     specialize qw/aom_highbd_10_sub_pixel_avg_variance32x64 sse2/;
1502
1503     add_proto qw/uint32_t aom_highbd_10_sub_pixel_avg_variance32x32/, "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";
1504     specialize qw/aom_highbd_10_sub_pixel_avg_variance32x32 sse2/;
1505
1506     add_proto qw/uint32_t aom_highbd_10_sub_pixel_avg_variance32x16/, "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";
1507     specialize qw/aom_highbd_10_sub_pixel_avg_variance32x16 sse2/;
1508
1509     add_proto qw/uint32_t aom_highbd_10_sub_pixel_avg_variance16x32/, "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";
1510     specialize qw/aom_highbd_10_sub_pixel_avg_variance16x32 sse2/;
1511
1512     add_proto qw/uint32_t aom_highbd_10_sub_pixel_avg_variance16x16/, "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";
1513     specialize qw/aom_highbd_10_sub_pixel_avg_variance16x16 sse2/;
1514
1515     add_proto qw/uint32_t aom_highbd_10_sub_pixel_avg_variance16x8/, "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";
1516     specialize qw/aom_highbd_10_sub_pixel_avg_variance16x8 sse2/;
1517
1518     add_proto qw/uint32_t aom_highbd_10_sub_pixel_avg_variance8x16/, "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";
1519     specialize qw/aom_highbd_10_sub_pixel_avg_variance8x16 sse2/;
1520
1521     add_proto qw/uint32_t aom_highbd_10_sub_pixel_avg_variance8x8/, "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";
1522     specialize qw/aom_highbd_10_sub_pixel_avg_variance8x8 sse2/;
1523
1524     add_proto qw/uint32_t aom_highbd_10_sub_pixel_avg_variance8x4/, "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";
1525     specialize qw/aom_highbd_10_sub_pixel_avg_variance8x4 sse2/;
1526
1527     add_proto qw/uint32_t aom_highbd_10_sub_pixel_avg_variance4x8/, "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";
1528     add_proto qw/uint32_t aom_highbd_10_sub_pixel_avg_variance4x4/, "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";
1529
1530     add_proto qw/uint32_t aom_highbd_8_sub_pixel_avg_variance64x64/, "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";
1531     specialize qw/aom_highbd_8_sub_pixel_avg_variance64x64 sse2/;
1532
1533     add_proto qw/uint32_t aom_highbd_8_sub_pixel_avg_variance64x32/, "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";
1534     specialize qw/aom_highbd_8_sub_pixel_avg_variance64x32 sse2/;
1535
1536     add_proto qw/uint32_t aom_highbd_8_sub_pixel_avg_variance32x64/, "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";
1537     specialize qw/aom_highbd_8_sub_pixel_avg_variance32x64 sse2/;
1538
1539     add_proto qw/uint32_t aom_highbd_8_sub_pixel_avg_variance32x32/, "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";
1540     specialize qw/aom_highbd_8_sub_pixel_avg_variance32x32 sse2/;
1541
1542     add_proto qw/uint32_t aom_highbd_8_sub_pixel_avg_variance32x16/, "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";
1543     specialize qw/aom_highbd_8_sub_pixel_avg_variance32x16 sse2/;
1544
1545     add_proto qw/uint32_t aom_highbd_8_sub_pixel_avg_variance16x32/, "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";
1546     specialize qw/aom_highbd_8_sub_pixel_avg_variance16x32 sse2/;
1547
1548     add_proto qw/uint32_t aom_highbd_8_sub_pixel_avg_variance16x16/, "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";
1549     specialize qw/aom_highbd_8_sub_pixel_avg_variance16x16 sse2/;
1550
1551     add_proto qw/uint32_t aom_highbd_8_sub_pixel_avg_variance16x8/, "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";
1552     specialize qw/aom_highbd_8_sub_pixel_avg_variance16x8 sse2/;
1553
1554     add_proto qw/uint32_t aom_highbd_8_sub_pixel_avg_variance8x16/, "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";
1555     specialize qw/aom_highbd_8_sub_pixel_avg_variance8x16 sse2/;
1556
1557     add_proto qw/uint32_t aom_highbd_8_sub_pixel_avg_variance8x8/, "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";
1558     specialize qw/aom_highbd_8_sub_pixel_avg_variance8x8 sse2/;
1559
1560     add_proto qw/uint32_t aom_highbd_8_sub_pixel_avg_variance8x4/, "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";
1561     specialize qw/aom_highbd_8_sub_pixel_avg_variance8x4 sse2/;
1562
1563     add_proto qw/uint32_t aom_highbd_8_sub_pixel_avg_variance4x8/, "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";
1564     add_proto qw/uint32_t aom_highbd_8_sub_pixel_avg_variance4x4/, "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";
1565
1566
1567
1568   add_proto qw/void aom_comp_mask_pred/, "uint8_t *comp_pred, const uint8_t *pred, int width, int height, const uint8_t *ref, int ref_stride, const uint8_t *mask, int mask_stride, int invert_mask";
1569   specialize qw/aom_comp_mask_pred ssse3 avx2/;
1570
1571   add_proto qw/void aom_highbd_comp_mask_pred/, "uint16_t *comp_pred, const uint8_t *pred8, int width, int height, const uint8_t *ref8, int ref_stride, const uint8_t *mask, int mask_stride, int invert_mask";
1572   add_proto qw/void aom_highbd_comp_mask_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, const uint8_t *mask, int mask_stride, int invert_mask, int bd";
1573
1574
1575 }  # CONFIG_AV1_ENCODER
1576
1577 1;