vpx_dsp_rtcd_defs.pl 120 KB
Newer Older
Johann's avatar
Johann committed
1 2 3 4 5 6 7
sub vpx_dsp_forward_decls() {
print <<EOF
/*
 * DSP
 */

#include "vpx/vpx_integer.h"
8
#include "vpx_dsp/vpx_dsp_common.h"
Johann's avatar
Johann committed
9 10 11 12 13

EOF
}
forward_decls qw/vpx_dsp_forward_decls/;

14 15 16 17 18 19
# x86inc.asm had specific constraints. break it out so it's easy to disable.
# zero all the variables to avoid tricky else conditions.
$mmx_x86inc = $sse_x86inc = $sse2_x86inc = $ssse3_x86inc = $avx_x86inc =
  $avx2_x86inc = '';
$mmx_x86_64_x86inc = $sse_x86_64_x86inc = $sse2_x86_64_x86inc =
  $ssse3_x86_64_x86inc = $avx_x86_64_x86inc = $avx2_x86_64_x86inc = '';
Johann's avatar
Johann committed
20 21 22 23 24 25 26
if (vpx_config("CONFIG_USE_X86INC") eq "yes") {
  $mmx_x86inc = 'mmx';
  $sse_x86inc = 'sse';
  $sse2_x86inc = 'sse2';
  $ssse3_x86inc = 'ssse3';
  $avx_x86inc = 'avx';
  $avx2_x86inc = 'avx2';
27 28 29 30 31 32 33 34
  if ($opts{arch} eq "x86_64") {
    $mmx_x86_64_x86inc = 'mmx';
    $sse_x86_64_x86inc = 'sse';
    $sse2_x86_64_x86inc = 'sse2';
    $ssse3_x86_64_x86inc = 'ssse3';
    $avx_x86_64_x86inc = 'avx';
    $avx2_x86_64_x86inc = 'avx2';
  }
Johann's avatar
Johann committed
35 36
}

Zoe Liu's avatar
Zoe Liu committed
37 38 39 40 41 42
# optimizations which depend on multiple features
$avx2_ssse3 = '';
if ((vpx_config("HAVE_AVX2") eq "yes") && (vpx_config("HAVE_SSSE3") eq "yes")) {
  $avx2_ssse3 = 'avx2';
}

43 44
# functions that are 64 bit only.
$mmx_x86_64 = $sse2_x86_64 = $ssse3_x86_64 = $avx_x86_64 = $avx2_x86_64 = '';
Johann's avatar
Johann committed
45 46 47 48 49 50 51 52
if ($opts{arch} eq "x86_64") {
  $mmx_x86_64 = 'mmx';
  $sse2_x86_64 = 'sse2';
  $ssse3_x86_64 = 'ssse3';
  $avx_x86_64 = 'avx';
  $avx2_x86_64 = 'avx2';
}

53 54 55 56
#
# Intra prediction
#

57 58
add_proto qw/void vpx_d207_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_d207_predictor_4x4/, "$ssse3_x86inc";
59

60 61 62
add_proto qw/void vpx_d207e_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_d207e_predictor_4x4/;

63 64
add_proto qw/void vpx_d45_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_d45_predictor_4x4 neon/, "$ssse3_x86inc";
65

66 67 68
add_proto qw/void vpx_d45e_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_d45e_predictor_4x4/;

69 70
add_proto qw/void vpx_d63_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_d63_predictor_4x4/, "$ssse3_x86inc";
71

72 73 74
add_proto qw/void vpx_d63e_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_d63e_predictor_4x4/;

75 76 77
add_proto qw/void vpx_d63f_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_d63f_predictor_4x4/;

78
add_proto qw/void vpx_h_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
79
specialize qw/vpx_h_predictor_4x4 neon dspr2 msa/, "$sse2_x86inc";
80

81 82 83
add_proto qw/void vpx_he_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_he_predictor_4x4/;

84 85
add_proto qw/void vpx_d117_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_d117_predictor_4x4/;
86

87 88
add_proto qw/void vpx_d135_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_d135_predictor_4x4 neon/;
89

90 91
add_proto qw/void vpx_d153_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_d153_predictor_4x4/, "$ssse3_x86inc";
92

93 94
add_proto qw/void vpx_v_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_v_predictor_4x4 neon msa/, "$sse_x86inc";
95

96 97 98
add_proto qw/void vpx_ve_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_ve_predictor_4x4/;

99
add_proto qw/void vpx_tm_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
Jian Zhou's avatar
Jian Zhou committed
100
specialize qw/vpx_tm_predictor_4x4 neon dspr2 msa/, "$sse2_x86inc";
101

102 103
add_proto qw/void vpx_dc_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_dc_predictor_4x4 dspr2 msa neon/, "$sse_x86inc";
104

105 106
add_proto qw/void vpx_dc_top_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_dc_top_predictor_4x4 msa neon/, "$sse_x86inc";
107

108 109
add_proto qw/void vpx_dc_left_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_dc_left_predictor_4x4 msa neon/, "$sse_x86inc";
110

111 112
add_proto qw/void vpx_dc_128_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_dc_128_predictor_4x4 msa neon/, "$sse_x86inc";
113

114 115
add_proto qw/void vpx_d207_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_d207_predictor_8x8/, "$ssse3_x86inc";
116

117 118 119
add_proto qw/void vpx_d207e_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_d207e_predictor_8x8/;

120 121
add_proto qw/void vpx_d45_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_d45_predictor_8x8 neon/, "$ssse3_x86inc";
122

123 124 125
add_proto qw/void vpx_d45e_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_d45e_predictor_8x8/;

126 127
add_proto qw/void vpx_d63_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_d63_predictor_8x8/, "$ssse3_x86inc";
128

129 130 131
add_proto qw/void vpx_d63e_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_d63e_predictor_8x8/;

132 133
add_proto qw/void vpx_h_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_h_predictor_8x8 neon dspr2 msa/, "$ssse3_x86inc";
134

135 136
add_proto qw/void vpx_d117_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_d117_predictor_8x8/;
137

138 139
add_proto qw/void vpx_d135_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_d135_predictor_8x8/;
140

141 142
add_proto qw/void vpx_d153_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_d153_predictor_8x8/, "$ssse3_x86inc";
143

144 145
add_proto qw/void vpx_v_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_v_predictor_8x8 neon msa/, "$sse_x86inc";
146

147 148
add_proto qw/void vpx_tm_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_tm_predictor_8x8 neon dspr2 msa/, "$sse2_x86inc";
149

150 151
add_proto qw/void vpx_dc_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_dc_predictor_8x8 dspr2 neon msa/, "$sse_x86inc";
152

153 154
add_proto qw/void vpx_dc_top_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_dc_top_predictor_8x8 neon msa/, "$sse_x86inc";
155

156 157
add_proto qw/void vpx_dc_left_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_dc_left_predictor_8x8 neon msa/, "$sse_x86inc";
158

159 160
add_proto qw/void vpx_dc_128_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_dc_128_predictor_8x8 neon msa/, "$sse_x86inc";
161

162 163
add_proto qw/void vpx_d207_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_d207_predictor_16x16/, "$ssse3_x86inc";
164

165 166 167
add_proto qw/void vpx_d207e_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_d207e_predictor_16x16/;

168 169
add_proto qw/void vpx_d45_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_d45_predictor_16x16 neon/, "$ssse3_x86inc";
170

171 172 173
add_proto qw/void vpx_d45e_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_d45e_predictor_16x16/;

174 175
add_proto qw/void vpx_d63_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_d63_predictor_16x16/, "$ssse3_x86inc";
176

177 178 179
add_proto qw/void vpx_d63e_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_d63e_predictor_16x16/;

180 181
add_proto qw/void vpx_h_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_h_predictor_16x16 neon dspr2 msa/, "$ssse3_x86inc";
182

183 184
add_proto qw/void vpx_d117_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_d117_predictor_16x16/;
185

186 187
add_proto qw/void vpx_d135_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_d135_predictor_16x16/;
188

189 190
add_proto qw/void vpx_d153_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_d153_predictor_16x16/, "$ssse3_x86inc";
191

192 193
add_proto qw/void vpx_v_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_v_predictor_16x16 neon msa/, "$sse2_x86inc";
194

195 196
add_proto qw/void vpx_tm_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_tm_predictor_16x16 neon msa/, "$sse2_x86inc";
197

198 199
add_proto qw/void vpx_dc_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_dc_predictor_16x16 dspr2 neon msa/, "$sse2_x86inc";
200

201 202
add_proto qw/void vpx_dc_top_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_dc_top_predictor_16x16 neon msa/, "$sse2_x86inc";
203

204 205
add_proto qw/void vpx_dc_left_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_dc_left_predictor_16x16 neon msa/, "$sse2_x86inc";
206

207 208
add_proto qw/void vpx_dc_128_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_dc_128_predictor_16x16 neon msa/, "$sse2_x86inc";
209

210 211
add_proto qw/void vpx_d207_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_d207_predictor_32x32/, "$ssse3_x86inc";
212

213 214 215
add_proto qw/void vpx_d207e_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_d207e_predictor_32x32/;

216 217
add_proto qw/void vpx_d45_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_d45_predictor_32x32/, "$ssse3_x86inc";
218

219 220 221
add_proto qw/void vpx_d45e_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_d45e_predictor_32x32/;

222 223
add_proto qw/void vpx_d63_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_d63_predictor_32x32/, "$ssse3_x86inc";
224

225 226 227
add_proto qw/void vpx_d63e_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_d63e_predictor_32x32/;

228 229
add_proto qw/void vpx_h_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_h_predictor_32x32 neon msa/, "$ssse3_x86inc";
230

231 232
add_proto qw/void vpx_d117_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_d117_predictor_32x32/;
233

234 235
add_proto qw/void vpx_d135_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_d135_predictor_32x32/;
236

237 238
add_proto qw/void vpx_d153_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_d153_predictor_32x32/, "$ssse3_x86inc";
239

240 241
add_proto qw/void vpx_v_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_v_predictor_32x32 neon msa/, "$sse2_x86inc";
242

243 244
add_proto qw/void vpx_tm_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_tm_predictor_32x32 neon msa/, "$sse2_x86_64_x86inc";
245

246 247
add_proto qw/void vpx_dc_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_dc_predictor_32x32 msa neon/, "$sse2_x86inc";
248

249 250
add_proto qw/void vpx_dc_top_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_dc_top_predictor_32x32 msa neon/, "$sse2_x86inc";
251

252 253
add_proto qw/void vpx_dc_left_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_dc_left_predictor_32x32 msa neon/, "$sse2_x86inc";
254

255 256
add_proto qw/void vpx_dc_128_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
specialize qw/vpx_dc_128_predictor_32x32 msa neon/, "$sse2_x86inc";
257 258

# High bitdepth functions
259 260 261
if (vpx_config("CONFIG_VP9_HIGHBITDEPTH") eq "yes") {
  add_proto qw/void vpx_highbd_d207_predictor_4x4/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_d207_predictor_4x4/;
262

263 264 265
  add_proto qw/void vpx_highbd_d207e_predictor_4x4/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_d207e_predictor_4x4/;

266 267
  add_proto qw/void vpx_highbd_d45_predictor_4x4/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_d45_predictor_4x4/;
268

269 270 271
  add_proto qw/void vpx_highbd_d45e_predictor_4x4/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_d45e_predictor_4x4/;

272 273
  add_proto qw/void vpx_highbd_d63_predictor_4x4/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_d63_predictor_4x4/;
274

275 276 277
  add_proto qw/void vpx_highbd_d63e_predictor_4x4/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_d63e_predictor_4x4/;

278 279
  add_proto qw/void vpx_highbd_h_predictor_4x4/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_h_predictor_4x4/;
280

281 282
  add_proto qw/void vpx_highbd_d117_predictor_4x4/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_d117_predictor_4x4/;
283

284 285
  add_proto qw/void vpx_highbd_d135_predictor_4x4/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_d135_predictor_4x4/;
286

287 288
  add_proto qw/void vpx_highbd_d153_predictor_4x4/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_d153_predictor_4x4/;
289

290 291
  add_proto qw/void vpx_highbd_v_predictor_4x4/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_v_predictor_4x4/, "$sse_x86inc";
292

293 294
  add_proto qw/void vpx_highbd_tm_predictor_4x4/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_tm_predictor_4x4/, "$sse_x86inc";
295

296 297
  add_proto qw/void vpx_highbd_dc_predictor_4x4/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_dc_predictor_4x4/, "$sse_x86inc";
298

299 300
  add_proto qw/void vpx_highbd_dc_top_predictor_4x4/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_dc_top_predictor_4x4/;
301

302 303
  add_proto qw/void vpx_highbd_dc_left_predictor_4x4/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_dc_left_predictor_4x4/;
304

305 306
  add_proto qw/void vpx_highbd_dc_128_predictor_4x4/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_dc_128_predictor_4x4/;
307

308 309
  add_proto qw/void vpx_highbd_d207_predictor_8x8/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_d207_predictor_8x8/;
310

311 312 313
  add_proto qw/void vpx_highbd_d207e_predictor_8x8/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_d207e_predictor_8x8/;

314 315
  add_proto qw/void vpx_highbd_d45_predictor_8x8/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_d45_predictor_8x8/;
316

317 318 319
  add_proto qw/void vpx_highbd_d45e_predictor_8x8/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_d45e_predictor_8x8/;

320 321
  add_proto qw/void vpx_highbd_d63_predictor_8x8/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_d63_predictor_8x8/;
322

323 324 325
  add_proto qw/void vpx_highbd_d63e_predictor_8x8/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_d63e_predictor_8x8/;

326 327
  add_proto qw/void vpx_highbd_h_predictor_8x8/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_h_predictor_8x8/;
328

329 330
  add_proto qw/void vpx_highbd_d117_predictor_8x8/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_d117_predictor_8x8/;
331

332 333
  add_proto qw/void vpx_highbd_d135_predictor_8x8/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_d135_predictor_8x8/;
334

335 336
  add_proto qw/void vpx_highbd_d153_predictor_8x8/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_d153_predictor_8x8/;
337

338 339
  add_proto qw/void vpx_highbd_v_predictor_8x8/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_v_predictor_8x8/, "$sse2_x86inc";
340

341 342
  add_proto qw/void vpx_highbd_tm_predictor_8x8/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_tm_predictor_8x8/, "$sse2_x86inc";
343

344 345
  add_proto qw/void vpx_highbd_dc_predictor_8x8/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_dc_predictor_8x8/, "$sse2_x86inc";;
346

347 348
  add_proto qw/void vpx_highbd_dc_top_predictor_8x8/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_dc_top_predictor_8x8/;
349

350 351
  add_proto qw/void vpx_highbd_dc_left_predictor_8x8/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_dc_left_predictor_8x8/;
352

353 354
  add_proto qw/void vpx_highbd_dc_128_predictor_8x8/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_dc_128_predictor_8x8/;
355

356 357
  add_proto qw/void vpx_highbd_d207_predictor_16x16/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_d207_predictor_16x16/;
358

359 360 361
  add_proto qw/void vpx_highbd_d207e_predictor_16x16/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_d207e_predictor_16x16/;

362 363
  add_proto qw/void vpx_highbd_d45_predictor_16x16/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_d45_predictor_16x16/;
364

365 366 367
  add_proto qw/void vpx_highbd_d45e_predictor_16x16/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_d45e_predictor_16x16/;

368 369
  add_proto qw/void vpx_highbd_d63_predictor_16x16/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_d63_predictor_16x16/;
370

371 372 373
  add_proto qw/void vpx_highbd_d63e_predictor_16x16/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_d63e_predictor_16x16/;

374 375
  add_proto qw/void vpx_highbd_h_predictor_16x16/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_h_predictor_16x16/;
376

377 378
  add_proto qw/void vpx_highbd_d117_predictor_16x16/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_d117_predictor_16x16/;
379

380 381
  add_proto qw/void vpx_highbd_d135_predictor_16x16/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_d135_predictor_16x16/;
382

383 384
  add_proto qw/void vpx_highbd_d153_predictor_16x16/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_d153_predictor_16x16/;
385

386 387
  add_proto qw/void vpx_highbd_v_predictor_16x16/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_v_predictor_16x16/, "$sse2_x86inc";
388

389 390
  add_proto qw/void vpx_highbd_tm_predictor_16x16/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_tm_predictor_16x16/, "$sse2_x86_64_x86inc";
391

392 393
  add_proto qw/void vpx_highbd_dc_predictor_16x16/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_dc_predictor_16x16/, "$sse2_x86inc";
394

395 396
  add_proto qw/void vpx_highbd_dc_top_predictor_16x16/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_dc_top_predictor_16x16/;
397

398 399
  add_proto qw/void vpx_highbd_dc_left_predictor_16x16/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_dc_left_predictor_16x16/;
400

401 402
  add_proto qw/void vpx_highbd_dc_128_predictor_16x16/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_dc_128_predictor_16x16/;
403

404 405
  add_proto qw/void vpx_highbd_d207_predictor_32x32/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_d207_predictor_32x32/;
406

407 408 409
  add_proto qw/void vpx_highbd_d207e_predictor_32x32/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_d207e_predictor_32x32/;

410 411
  add_proto qw/void vpx_highbd_d45_predictor_32x32/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_d45_predictor_32x32/;
412

413 414 415
  add_proto qw/void vpx_highbd_d45e_predictor_32x32/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_d45e_predictor_32x32/;

416 417
  add_proto qw/void vpx_highbd_d63_predictor_32x32/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_d63_predictor_32x32/;
418

419 420 421
  add_proto qw/void vpx_highbd_d63e_predictor_32x32/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_d63e_predictor_32x32/;

422 423
  add_proto qw/void vpx_highbd_h_predictor_32x32/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_h_predictor_32x32/;
424

425 426
  add_proto qw/void vpx_highbd_d117_predictor_32x32/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_d117_predictor_32x32/;
427

428 429
  add_proto qw/void vpx_highbd_d135_predictor_32x32/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_d135_predictor_32x32/;
430

431 432
  add_proto qw/void vpx_highbd_d153_predictor_32x32/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_d153_predictor_32x32/;
433

434 435
  add_proto qw/void vpx_highbd_v_predictor_32x32/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_v_predictor_32x32/, "$sse2_x86inc";
436

437 438
  add_proto qw/void vpx_highbd_tm_predictor_32x32/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_tm_predictor_32x32/, "$sse2_x86_64_x86inc";
439

440 441
  add_proto qw/void vpx_highbd_dc_predictor_32x32/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_dc_predictor_32x32/, "$sse2_x86_64_x86inc";
442

443 444
  add_proto qw/void vpx_highbd_dc_top_predictor_32x32/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_dc_top_predictor_32x32/;
445

446 447
  add_proto qw/void vpx_highbd_dc_left_predictor_32x32/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_dc_left_predictor_32x32/;
448

449 450 451
  add_proto qw/void vpx_highbd_dc_128_predictor_32x32/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
  specialize qw/vpx_highbd_dc_128_predictor_32x32/;
}  # CONFIG_VP9_HIGHBITDEPTH
452

Zoe Liu's avatar
Zoe Liu committed
453 454 455 456
#
# Sub Pixel Filters
#
add_proto qw/void vpx_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";
457
specialize qw/vpx_convolve_copy neon dspr2 msa/, "$sse2_x86inc";
Zoe Liu's avatar
Zoe Liu committed
458 459

add_proto qw/void vpx_convolve_avg/, "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";
460
specialize qw/vpx_convolve_avg neon dspr2 msa/, "$sse2_x86inc";
Zoe Liu's avatar
Zoe Liu committed
461 462

add_proto qw/void vpx_convolve8/, "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";
463
specialize qw/vpx_convolve8 sse2 ssse3 neon dspr2 msa/, "$avx2_ssse3";
Zoe Liu's avatar
Zoe Liu committed
464 465

add_proto qw/void vpx_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";
466
specialize qw/vpx_convolve8_horiz sse2 ssse3 neon dspr2 msa/, "$avx2_ssse3";
Zoe Liu's avatar
Zoe Liu committed
467 468

add_proto qw/void vpx_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";
469
specialize qw/vpx_convolve8_vert sse2 ssse3 neon dspr2 msa/, "$avx2_ssse3";
Zoe Liu's avatar
Zoe Liu committed
470 471

add_proto qw/void vpx_convolve8_avg/, "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";
472
specialize qw/vpx_convolve8_avg sse2 ssse3 neon dspr2 msa/;
Zoe Liu's avatar
Zoe Liu committed
473 474

add_proto qw/void vpx_convolve8_avg_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";
475
specialize qw/vpx_convolve8_avg_horiz sse2 ssse3 neon dspr2 msa/;
Zoe Liu's avatar
Zoe Liu committed
476 477

add_proto qw/void vpx_convolve8_avg_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";
478
specialize qw/vpx_convolve8_avg_vert sse2 ssse3 neon dspr2 msa/;
Zoe Liu's avatar
Zoe Liu committed
479

480
add_proto qw/void vpx_scaled_2d/, "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";
481
specialize qw/vpx_scaled_2d ssse3/;
482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497

add_proto qw/void vpx_scaled_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";
specialize qw/vpx_scaled_horiz/;

add_proto qw/void vpx_scaled_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";
specialize qw/vpx_scaled_vert/;

add_proto qw/void vpx_scaled_avg_2d/, "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";
specialize qw/vpx_scaled_avg_2d/;

add_proto qw/void vpx_scaled_avg_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";
specialize qw/vpx_scaled_avg_horiz/;

add_proto qw/void vpx_scaled_avg_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";
specialize qw/vpx_scaled_avg_vert/;

Zoe Liu's avatar
Zoe Liu committed
498 499 500 501 502
if (vpx_config("CONFIG_VP9_HIGHBITDEPTH") eq "yes") {
  #
  # Sub Pixel Filters
  #
  add_proto qw/void vpx_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";
503
  specialize qw/vpx_highbd_convolve_copy/, "$sse2_x86inc";
Zoe Liu's avatar
Zoe Liu committed
504 505

  add_proto qw/void vpx_highbd_convolve_avg/, "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";
506
  specialize qw/vpx_highbd_convolve_avg/, "$sse2_x86inc";
Zoe Liu's avatar
Zoe Liu committed
507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526

  add_proto qw/void vpx_highbd_convolve8/, "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";
  specialize qw/vpx_highbd_convolve8/, "$sse2_x86_64";

  add_proto qw/void vpx_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";
  specialize qw/vpx_highbd_convolve8_horiz/, "$sse2_x86_64";

  add_proto qw/void vpx_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";
  specialize qw/vpx_highbd_convolve8_vert/, "$sse2_x86_64";

  add_proto qw/void vpx_highbd_convolve8_avg/, "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";
  specialize qw/vpx_highbd_convolve8_avg/, "$sse2_x86_64";

  add_proto qw/void vpx_highbd_convolve8_avg_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";
  specialize qw/vpx_highbd_convolve8_avg_horiz/, "$sse2_x86_64";

  add_proto qw/void vpx_highbd_convolve8_avg_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";
  specialize qw/vpx_highbd_convolve8_avg_vert/, "$sse2_x86_64";
}  # CONFIG_VP9_HIGHBITDEPTH

527 528 529
#
# Loopfilter
#
530
add_proto qw/void vpx_lpf_vertical_16/, "uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh";
531
specialize qw/vpx_lpf_vertical_16 sse2 neon_asm dspr2 msa/;
532
$vpx_lpf_vertical_16_neon_asm=vpx_lpf_vertical_16_neon;
533

534
add_proto qw/void vpx_lpf_vertical_16_dual/, "uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh";
535
specialize qw/vpx_lpf_vertical_16_dual sse2 neon_asm dspr2 msa/;
536
$vpx_lpf_vertical_16_dual_neon_asm=vpx_lpf_vertical_16_dual_neon;
537

538
add_proto qw/void vpx_lpf_vertical_8/, "uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh, int count";
539
specialize qw/vpx_lpf_vertical_8 sse2 neon dspr2 msa/;
540

541
add_proto qw/void vpx_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";
542
specialize qw/vpx_lpf_vertical_8_dual sse2 neon_asm dspr2 msa/;
543
$vpx_lpf_vertical_8_dual_neon_asm=vpx_lpf_vertical_8_dual_neon;
544

545
add_proto qw/void vpx_lpf_vertical_4/, "uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh, int count";
546
specialize qw/vpx_lpf_vertical_4 mmx neon dspr2 msa/;
547

548
add_proto qw/void vpx_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";
549
specialize qw/vpx_lpf_vertical_4_dual sse2 neon dspr2 msa/;
550

551
add_proto qw/void vpx_lpf_horizontal_16/, "uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh, int count";
552
specialize qw/vpx_lpf_horizontal_16 sse2 avx2 neon_asm dspr2 msa/;
553
$vpx_lpf_horizontal_16_neon_asm=vpx_lpf_horizontal_16_neon;
554

555
add_proto qw/void vpx_lpf_horizontal_8/, "uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh, int count";
556
specialize qw/vpx_lpf_horizontal_8 sse2 neon dspr2 msa/;
557

558
add_proto qw/void vpx_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";
559
specialize qw/vpx_lpf_horizontal_8_dual sse2 neon_asm dspr2 msa/;
560
$vpx_lpf_horizontal_8_dual_neon_asm=vpx_lpf_horizontal_8_dual_neon;
561

562
add_proto qw/void vpx_lpf_horizontal_4/, "uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh, int count";
563
specialize qw/vpx_lpf_horizontal_4 mmx neon dspr2 msa/;
564

565
add_proto qw/void vpx_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";
566
specialize qw/vpx_lpf_horizontal_4_dual sse2 neon dspr2 msa/;
567 568

if (vpx_config("CONFIG_VP9_HIGHBITDEPTH") eq "yes") {
569 570
  add_proto qw/void vpx_highbd_lpf_vertical_16/, "uint16_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh, int bd";
  specialize qw/vpx_highbd_lpf_vertical_16 sse2/;
571

572 573
  add_proto qw/void vpx_highbd_lpf_vertical_16_dual/, "uint16_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh, int bd";
  specialize qw/vpx_highbd_lpf_vertical_16_dual sse2/;
574

575 576
  add_proto qw/void vpx_highbd_lpf_vertical_8/, "uint16_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh, int count, int bd";
  specialize qw/vpx_highbd_lpf_vertical_8 sse2/;
577

578 579
  add_proto qw/void vpx_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";
  specialize qw/vpx_highbd_lpf_vertical_8_dual sse2/;
580

581 582
  add_proto qw/void vpx_highbd_lpf_vertical_4/, "uint16_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh, int count, int bd";
  specialize qw/vpx_highbd_lpf_vertical_4 sse2/;
583

584 585
  add_proto qw/void vpx_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";
  specialize qw/vpx_highbd_lpf_vertical_4_dual sse2/;
586

587 588
  add_proto qw/void vpx_highbd_lpf_horizontal_16/, "uint16_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh, int count, int bd";
  specialize qw/vpx_highbd_lpf_horizontal_16 sse2/;
589

590 591
  add_proto qw/void vpx_highbd_lpf_horizontal_8/, "uint16_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh, int count, int bd";
  specialize qw/vpx_highbd_lpf_horizontal_8 sse2/;
592

593 594
  add_proto qw/void vpx_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";
  specialize qw/vpx_highbd_lpf_horizontal_8_dual sse2/;
595

596 597
  add_proto qw/void vpx_highbd_lpf_horizontal_4/, "uint16_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh, int count, int bd";
  specialize qw/vpx_highbd_lpf_horizontal_4 sse2/;
598

599 600
  add_proto qw/void vpx_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";
  specialize qw/vpx_highbd_lpf_horizontal_4_dual sse2/;
601 602
}  # CONFIG_VP9_HIGHBITDEPTH

603 604 605
#
# Encoder functions.
#
606 607 608 609

#
# Forward transform
#
610
if ((vpx_config("CONFIG_VP9_ENCODER") eq "yes") || (vpx_config("CONFIG_VP10_ENCODER") eq "yes")) {
611
if (vpx_config("CONFIG_VP9_HIGHBITDEPTH") eq "yes") {
612 613
  add_proto qw/void vpx_fdct4x4/, "const int16_t *input, tran_low_t *output, int stride";
  specialize qw/vpx_fdct4x4 sse2/;
614

615 616
  add_proto qw/void vpx_fdct4x4_1/, "const int16_t *input, tran_low_t *output, int stride";
  specialize qw/vpx_fdct4x4_1 sse2/;
617

618 619
  add_proto qw/void vpx_fdct8x8/, "const int16_t *input, tran_low_t *output, int stride";
  specialize qw/vpx_fdct8x8 sse2/;
620

621 622
  add_proto qw/void vpx_fdct8x8_1/, "const int16_t *input, tran_low_t *output, int stride";
  specialize qw/vpx_fdct8x8_1 sse2/;
623

624 625
  add_proto qw/void vpx_fdct16x16/, "const int16_t *input, tran_low_t *output, int stride";
  specialize qw/vpx_fdct16x16 sse2/;
626

627 628
  add_proto qw/void vpx_fdct16x16_1/, "const int16_t *input, tran_low_t *output, int stride";
  specialize qw/vpx_fdct16x16_1 sse2/;
629

630 631
  add_proto qw/void vpx_fdct32x32/, "const int16_t *input, tran_low_t *output, int stride";
  specialize qw/vpx_fdct32x32 sse2/;
632

633 634
  add_proto qw/void vpx_fdct32x32_rd/, "const int16_t *input, tran_low_t *output, int stride";
  specialize qw/vpx_fdct32x32_rd sse2/;
635

636 637
  add_proto qw/void vpx_fdct32x32_1/, "const int16_t *input, tran_low_t *output, int stride";
  specialize qw/vpx_fdct32x32_1 sse2/;
638

639 640
  add_proto qw/void vpx_highbd_fdct4x4/, "const int16_t *input, tran_low_t *output, int stride";
  specialize qw/vpx_highbd_fdct4x4 sse2/;
641

642 643
  add_proto qw/void vpx_highbd_fdct8x8/, "const int16_t *input, tran_low_t *output, int stride";
  specialize qw/vpx_highbd_fdct8x8 sse2/;
644

645 646
  add_proto qw/void vpx_highbd_fdct8x8_1/, "const int16_t *input, tran_low_t *output, int stride";
  specialize qw/vpx_highbd_fdct8x8_1/;
647

648 649
  add_proto qw/void vpx_highbd_fdct16x16/, "const int16_t *input, tran_low_t *output, int stride";
  specialize qw/vpx_highbd_fdct16x16 sse2/;
650

651 652
  add_proto qw/void vpx_highbd_fdct16x16_1/, "const int16_t *input, tran_low_t *output, int stride";
  specialize qw/vpx_highbd_fdct16x16_1/;
653

654 655
  add_proto qw/void vpx_highbd_fdct32x32/, "const int16_t *input, tran_low_t *output, int stride";
  specialize qw/vpx_highbd_fdct32x32 sse2/;
656

657 658
  add_proto qw/void vpx_highbd_fdct32x32_rd/, "const int16_t *input, tran_low_t *output, int stride";
  specialize qw/vpx_highbd_fdct32x32_rd sse2/;
659

660 661
  add_proto qw/void vpx_highbd_fdct32x32_1/, "const int16_t *input, tran_low_t *output, int stride";
  specialize qw/vpx_highbd_fdct32x32_1/;
662
} else {
663 664
  add_proto qw/void vpx_fdct4x4/, "const int16_t *input, tran_low_t *output, int stride";
  specialize qw/vpx_fdct4x4 sse2 msa/;
665

666 667
  add_proto qw/void vpx_fdct4x4_1/, "const int16_t *input, tran_low_t *output, int stride";
  specialize qw/vpx_fdct4x4_1 sse2/;
668

669 670
  add_proto qw/void vpx_fdct8x8/, "const int16_t *input, tran_low_t *output, int stride";
  specialize qw/vpx_fdct8x8 sse2 neon msa/, "$ssse3_x86_64_x86inc";
671

672 673
  add_proto qw/void vpx_fdct8x8_1/, "const int16_t *input, tran_low_t *output, int stride";
  specialize qw/vpx_fdct8x8_1 sse2 neon msa/;
674

675 676
  add_proto qw/void vpx_fdct16x16/, "const int16_t *input, tran_low_t *output, int stride";
  specialize qw/vpx_fdct16x16 sse2 msa/;
677

678 679
  add_proto qw/void vpx_fdct16x16_1/, "const int16_t *input, tran_low_t *output, int stride";
  specialize qw/vpx_fdct16x16_1 sse2 msa/;
680

681 682
  add_proto qw/void vpx_fdct32x32/, "const int16_t *input, tran_low_t *output, int stride";
  specialize qw/vpx_fdct32x32 sse2 avx2 msa/;
683

684 685
  add_proto qw/void vpx_fdct32x32_rd/, "const int16_t *input, tran_low_t *output, int stride";
  specialize qw/vpx_fdct32x32_rd sse2 avx2 msa/;
686

687 688
  add_proto qw/void vpx_fdct32x32_1/, "const int16_t *input, tran_low_t *output, int stride";
  specialize qw/vpx_fdct32x32_1 sse2 msa/;
689
}  # CONFIG_VP9_HIGHBITDEPTH
690
}  # CONFIG_VP9_ENCODER || CONFIG_VP10_ENCODER
691

692 693
#
# Inverse transform
694
if ((vpx_config("CONFIG_VP9") eq "yes") || (vpx_config("CONFIG_VP10") eq "yes")) {
695 696 697
if (vpx_config("CONFIG_VP9_HIGHBITDEPTH") eq "yes") {
  # Note as optimized versions of these functions are added we need to add a check to ensure
  # that when CONFIG_EMULATE_HARDWARE is on, it defaults to the C versions only.
698 699
  add_proto qw/void vpx_iwht4x4_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
  specialize qw/vpx_iwht4x4_1_add/;
700

701 702
  add_proto qw/void vpx_iwht4x4_16_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
  specialize qw/vpx_iwht4x4_16_add/;
703

704 705
  add_proto qw/void vpx_highbd_idct4x4_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int bd";
  specialize qw/vpx_highbd_idct4x4_1_add/;
706

707 708
  add_proto qw/void vpx_highbd_idct8x8_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int bd";
  specialize qw/vpx_highbd_idct8x8_1_add/;
709

710 711
  add_proto qw/void vpx_highbd_idct16x16_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int bd";
  specialize qw/vpx_highbd_idct16x16_1_add/;
712

713 714
  add_proto qw/void vpx_highbd_idct32x32_1024_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int bd";
  specialize qw/vpx_highbd_idct32x32_1024_add/;
715

716 717
  add_proto qw/void vpx_highbd_idct32x32_34_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int bd";
  specialize qw/vpx_highbd_idct32x32_34_add/;
718

719 720
  add_proto qw/void vpx_highbd_idct32x32_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int bd";
  specialize qw/vpx_highbd_idct32x32_1_add/;
721

722 723
  add_proto qw/void vpx_highbd_iwht4x4_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int bd";
  specialize qw/vpx_highbd_iwht4x4_1_add/;
724

725 726
  add_proto qw/void vpx_highbd_iwht4x4_16_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int bd";
  specialize qw/vpx_highbd_iwht4x4_16_add/;
727 728 729

  # Force C versions if CONFIG_EMULATE_HARDWARE is 1
  if (vpx_config("CONFIG_EMULATE_HARDWARE") eq "yes") {
730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762
    add_proto qw/void vpx_idct4x4_16_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
    specialize qw/vpx_idct4x4_16_add/;

    add_proto qw/void vpx_idct4x4_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
    specialize qw/vpx_idct4x4_1_add/;

    add_proto qw/void vpx_idct8x8_64_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
    specialize qw/vpx_idct8x8_64_add/;

    add_proto qw/void vpx_idct8x8_12_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
    specialize qw/vpx_idct8x8_12_add/;

    add_proto qw/void vpx_idct8x8_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
    specialize qw/vpx_idct8x8_1_add/;

    add_proto qw/void vpx_idct16x16_256_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
    specialize qw/vpx_idct16x16_256_add/;

    add_proto qw/void vpx_idct16x16_10_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
    specialize qw/vpx_idct16x16_10_add/;

    add_proto qw/void vpx_idct16x16_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
    specialize qw/vpx_idct16x16_1_add/;

    add_proto qw/void vpx_idct32x32_1024_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
    specialize qw/vpx_idct32x32_1024_add/;

    add_proto qw/void vpx_idct32x32_34_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
    specialize qw/vpx_idct32x32_34_add/;

    add_proto qw/void vpx_idct32x32_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
    specialize qw/vpx_idct32x32_1_add/;
    
763 764
    add_proto qw/void vpx_highbd_idct4x4_16_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int bd";
    specialize qw/vpx_highbd_idct4x4_16_add/;
765

766 767
    add_proto qw/void vpx_highbd_idct8x8_64_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int bd";
    specialize qw/vpx_highbd_idct8x8_64_add/;
768

769 770
    add_proto qw/void vpx_highbd_idct8x8_10_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int bd";
    specialize qw/vpx_highbd_idct8x8_10_add/;
771

772 773
    add_proto qw/void vpx_highbd_idct16x16_256_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int bd";
    specialize qw/vpx_highbd_idct16x16_256_add/;
774

775 776
    add_proto qw/void vpx_highbd_idct16x16_10_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int bd";
    specialize qw/vpx_highbd_idct16x16_10_add/;
777
  } else {
778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810
    add_proto qw/void vpx_idct4x4_16_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
    specialize qw/vpx_idct4x4_16_add sse2/;

    add_proto qw/void vpx_idct4x4_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
    specialize qw/vpx_idct4x4_1_add sse2/;

    add_proto qw/void vpx_idct8x8_64_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
    specialize qw/vpx_idct8x8_64_add sse2/;

    add_proto qw/void vpx_idct8x8_12_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
    specialize qw/vpx_idct8x8_12_add sse2/;

    add_proto qw/void vpx_idct8x8_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
    specialize qw/vpx_idct8x8_1_add sse2/;

    add_proto qw/void vpx_idct16x16_256_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
    specialize qw/vpx_idct16x16_256_add sse2/;

    add_proto qw/void vpx_idct16x16_10_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
    specialize qw/vpx_idct16x16_10_add sse2/;

    add_proto qw/void vpx_idct16x16_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
    specialize qw/vpx_idct16x16_1_add sse2/;

    add_proto qw/void vpx_idct32x32_1024_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
    specialize qw/vpx_idct32x32_1024_add sse2/;

    add_proto qw/void vpx_idct32x32_34_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
    specialize qw/vpx_idct32x32_34_add sse2/;

    add_proto qw/void vpx_idct32x32_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
    specialize qw/vpx_idct32x32_1_add sse2/;

811 812
    add_proto qw/void vpx_highbd_idct4x4_16_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int bd";
    specialize qw/vpx_highbd_idct4x4_16_add sse2/;
813

814 815
    add_proto qw/void vpx_highbd_idct8x8_64_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int bd";
    specialize qw/vpx_highbd_idct8x8_64_add sse2/;
816

817 818
    add_proto qw/void vpx_highbd_idct8x8_10_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int bd";
    specialize qw/vpx_highbd_idct8x8_10_add sse2/;
819

820 821
    add_proto qw/void vpx_highbd_idct16x16_256_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int bd";
    specialize qw/vpx_highbd_idct16x16_256_add sse2/;
822

823 824
    add_proto qw/void vpx_highbd_idct16x16_10_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int bd";
    specialize qw/vpx_highbd_idct16x16_10_add sse2/;
825 826 827 828
  }  # CONFIG_EMULATE_HARDWARE
} else {
  # Force C versions if CONFIG_EMULATE_HARDWARE is 1
  if (vpx_config("CONFIG_EMULATE_HARDWARE") eq "yes") {
829 830
    add_proto qw/void vpx_idct4x4_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
    specialize qw/vpx_idct4x4_1_add/;
831

832 833
    add_proto qw/void vpx_idct4x4_16_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
    specialize qw/vpx_idct4x4_16_add/;
834

835 836
    add_proto qw/void vpx_idct8x8_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
    specialize qw/vpx_idct8x8_1_add/;
837

838 839
    add_proto qw/void vpx_idct8x8_64_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
    specialize qw/vpx_idct8x8_64_add/;
840

841 842
    add_proto qw/void vpx_idct8x8_12_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
    specialize qw/vpx_idct8x8_12_add/;
843

844 845
    add_proto qw/void vpx_idct16x16_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
    specialize qw/vpx_idct16x16_1_add/;
846

847 848
    add_proto qw/void vpx_idct16x16_256_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
    specialize qw/vpx_idct16x16_256_add/;
849

850 851
    add_proto qw/void vpx_idct16x16_10_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
    specialize qw/vpx_idct16x16_10_add/;
852

853 854
    add_proto qw/void vpx_idct32x32_1024_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
    specialize qw/vpx_idct32x32_1024_add/;
855

856 857
    add_proto qw/void vpx_idct32x32_34_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
    specialize qw/vpx_idct32x32_34_add/;
858

859 860
    add_proto qw/void vpx_idct32x32_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
    specialize qw/vpx_idct32x32_1_add/;
861

862 863
    add_proto qw/void vpx_iwht4x4_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
    specialize qw/vpx_iwht4x4_1_add/;
864

865 866
    add_proto qw/void vpx_iwht4x4_16_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
    specialize qw/vpx_iwht4x4_16_add/;
867
  } else {
868 869
    add_proto qw/void vpx_idct4x4_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
    specialize qw/vpx_idct4x4_1_add sse2 neon dspr2 msa/;
870

871 872
    add_proto qw/void vpx_idct4x4_16_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
    specialize qw/vpx_idct4x4_16_add sse2 neon dspr2 msa/;
873

874 875
    add_proto qw/void vpx_idct8x8_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
    specialize qw/vpx_idct8x8_1_add sse2 neon dspr2 msa/;
876

877 878
    add_proto qw/void vpx_idct8x8_64_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
    specialize qw/vpx_idct8x8_64_add sse2 neon dspr2 msa/, "$ssse3_x86_64_x86inc";
879

880 881
    add_proto qw/void vpx_idct8x8_12_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
    specialize qw/vpx_idct8x8_12_add sse2 neon dspr2 msa/, "$ssse3_x86_64_x86inc";
882

883 884
    add_proto qw/void vpx_idct16x16_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
    specialize qw/vpx_idct16x16_1_add sse2 neon dspr2 msa/;
885

886 887
    add_proto qw/void vpx_idct16x16_256_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
    specialize qw/vpx_idct16x16_256_add sse2 neon dspr2 msa/;
888

889 890
    add_proto qw/void vpx_idct16x16_10_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
    specialize qw/vpx_idct16x16_10_add sse2 neon dspr2 msa/;
891

892
    add_proto qw/void vpx_idct32x32_1024_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
893
    specialize qw/vpx_idct32x32_1024_add sse2 neon dspr2 msa/, "$ssse3_x86_64_x86inc";
894

895
    add_proto qw/void vpx_idct32x32_34_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
896
    specialize qw/vpx_idct32x32_34_add sse2 neon_asm dspr2 msa/, "$ssse3_x86_64_x86inc";
897
    # Need to add 34 eob idct32x32 neon implementation.
898
    $vpx_idct32x32_34_add_neon_asm=vpx_idct32x32_1024_add_neon;
899

900 901
    add_proto qw/void vpx_idct32x32_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
    specialize qw/vpx_idct32x32_1_add sse2 neon dspr2 msa/;
902

903 904
    add_proto qw/void vpx_iwht4x4_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
    specialize qw/vpx_iwht4x4_1_add msa/;
905

906 907
    add_proto qw/void vpx_iwht4x4_16_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
    specialize qw/vpx_iwht4x4_16_add msa/, "$sse2_x86inc";
908 909
  }  # CONFIG_EMULATE_HARDWARE
}  # CONFIG_VP9_HIGHBITDEPTH
910
}  # CONFIG_VP9 || CONFIG_VP10
911

912 913 914
#
# Quantization
#
915
if ((vpx_config("CONFIG_VP9_ENCODER") eq "yes") || (vpx_config("CONFIG_VP10_ENCODER") eq "yes")) {
916
  add_proto qw/void vpx_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";
917
  specialize qw/vpx_quantize_b sse2/, "$ssse3_x86_64_x86inc", "$avx_x86_64_x86inc";
918

919
  add_proto qw/void vpx_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";
920
  specialize qw/vpx_quantize_b_32x32/, "$ssse3_x86_64_x86inc", "$avx_x86_64_x86inc";
921

922 923 924
  if (vpx_config("CONFIG_VP9_HIGHBITDEPTH") eq "yes") {
    add_proto qw/void vpx_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";
    specialize qw/vpx_highbd_quantize_b sse2/;
925

926 927 928
    add_proto qw/void vpx_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";
    specialize qw/vpx_highbd_quantize_b_32x32 sse2/;
  }  # CONFIG_VP9_HIGHBITDEPTH
929
}  # CONFIG_VP9_ENCODER || CONFIG_VP10_ENCODER
930

Johann's avatar
Johann committed
931
if (vpx_config("CONFIG_ENCODERS") eq "yes") {
932 933 934 935 936 937
#
# Block subtraction
#
add_proto qw/void vpx_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";
specialize qw/vpx_subtract_block neon msa/, "$sse2_x86inc";

Johann's avatar
Johann committed
938 939 940 941
#
# Single block SAD
#
add_proto qw/unsigned int vpx_sad64x64/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride";
942
specialize qw/vpx_sad64x64 avx2 neon msa/, "$sse2_x86inc";
Johann's avatar
Johann committed
943 944

add_proto qw/unsigned int vpx_sad64x32/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride";
945
specialize qw/vpx_sad64x32 avx2 msa/, "$sse2_x86inc";
Johann's avatar
Johann committed
946 947

add_proto qw/unsigned int vpx_sad32x64/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride";
948
specialize qw/vpx_sad32x64 avx2 msa/, "$sse2_x86inc";
Johann's avatar
Johann committed
949 950

add_proto qw/unsigned int vpx_sad32x32/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride";
951
specialize qw/vpx_sad32x32 avx2 neon msa/, "$sse2_x86inc";
Johann's avatar
Johann committed
952 953

add_proto qw/unsigned int vpx_sad32x16/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride";
954
specialize qw/vpx_sad32x16 avx2 msa/, "$sse2_x86inc";
Johann's avatar
Johann committed
955 956

add_proto qw/unsigned int vpx_sad16x32/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride";
957
specialize qw/vpx_sad16x32 msa/, "$sse2_x86inc";
Johann's avatar
Johann committed
958 959

add_proto qw/unsigned int vpx_sad16x16/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride";
960
specialize qw/vpx_sad16x16 mmx media neon msa/, "$sse2_x86inc";
Johann's avatar
Johann committed
961 962

add_proto qw/unsigned int vpx_sad16x8/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride";
963
specialize qw/vpx_sad16x8 mmx neon msa/, "$sse2_x86inc";
Johann's avatar
Johann committed
964 965

add_proto qw/unsigned int vpx_sad8x16/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride";
966
specialize qw/vpx_sad8x16 mmx neon msa/, "$sse2_x86inc";
Johann's avatar
Johann committed
967 968

add_proto qw/unsigned int vpx_sad8x8/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride";
969
specialize qw/vpx_sad8x8 mmx neon msa/, "$sse2_x86inc";
Johann's avatar
Johann committed
970 971

add_proto qw/unsigned int vpx_sad8x4/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride";
972
specialize qw/vpx_sad8x4 msa/, "$sse2_x86inc";
Johann's avatar
Johann committed
973 974

add_proto qw/unsigned int vpx_sad4x8/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride";
975
specialize qw/vpx_sad4x8 msa/, "$sse_x86inc";
Johann's avatar
Johann committed
976 977

add_proto qw/unsigned int vpx_sad4x4/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride";
978
specialize qw/vpx_sad4x4 mmx neon msa/, "$sse_x86inc";
Johann's avatar
Johann committed
979 980 981 982 983

#
# Avg
#
add_proto qw/unsigned int vpx_sad64x64_avg/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred";
984
specialize qw/vpx_sad64x64_avg avx2 msa/, "$sse2_x86inc";
Johann's avatar
Johann committed
985 986

add_proto qw/unsigned int vpx_sad64x32_avg/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred";
987
specialize qw/vpx_sad64x32_avg avx2 msa/, "$sse2_x86inc";
Johann's avatar
Johann committed
988 989

add_proto qw/unsigned int vpx_sad32x64_avg/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred";
990
specialize qw/vpx_sad32x64_avg avx2 msa/, "$sse2_x86inc";
Johann's avatar
Johann committed
991 992

add_proto qw/unsigned int vpx_sad32x32_avg/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred";
993
specialize qw/vpx_sad32x32_avg avx2 msa/, "$sse2_x86inc";
Johann's avatar
Johann committed
994 995

add_proto qw/unsigned int vpx_sad32x16_avg/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred";
996
specialize qw/vpx_sad32x16_avg avx2 msa/, "$sse2_x86inc";
Johann's avatar
Johann committed
997 998

add_proto qw/unsigned int vpx_sad16x32_avg/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred";
999
specialize qw/vpx_sad16x32_avg msa/, "$sse2_x86inc";
Johann's avatar
Johann committed
1000 1001

add_proto qw/unsigned int vpx_sad16x16_avg/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred";
1002
specialize qw/vpx_sad16x16_avg msa/, "$sse2_x86inc";
Johann's avatar
Johann committed
1003 1004

add_proto qw/unsigned int vpx_sad16x8_avg/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred";
1005
specialize qw/vpx_sad16x8_avg msa/, "$sse2_x86inc";
Johann's avatar
Johann committed
1006 1007

add_proto qw/unsigned int vpx_sad8x16_avg/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred";
1008
specialize qw/vpx_sad8x16_avg msa/, "$sse2_x86inc";
Johann's avatar
Johann committed
1009 1010

add_proto qw/unsigned int vpx_sad8x8_avg/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred";
1011
specialize qw/vpx_sad8x8_avg msa/, "$sse2_x86inc";
Johann's avatar
Johann committed
1012 1013

add_proto qw/unsigned int vpx_sad8x4_avg/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred";
1014
specialize qw/vpx_sad8x4_avg msa/, "$sse2_x86inc";
Johann's avatar
Johann committed
1015 1016

add_proto qw/unsigned int vpx_sad4x8_avg/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred";
1017
specialize qw/vpx_sad4x8_avg msa/, "$sse_x86inc";
Johann's avatar
Johann committed
1018 1019

add_proto qw/unsigned int vpx_sad4x4_avg/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred";
1020
specialize qw/vpx_sad4x4_avg msa/, "$sse_x86inc";
Johann's avatar
Johann committed
1021 1022 1023 1024 1025 1026

#
# Multi-block SAD, comparing a reference to N blocks 1 pixel apart horizontally
#
# Blocks of 3
add_proto qw/void vpx_sad64x64x3/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, uint32_t *sad_array";
1027
specialize qw/vpx_sad64x64x3 msa/;
Johann's avatar
Johann committed
1028 1029

add_proto qw/void vpx_sad32x32x3/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, uint32_t *sad_array";
1030
specialize qw/vpx_sad32x32x3 msa/;
Johann's avatar
Johann committed
1031 1032

add_proto qw/void vpx_sad16x16x3/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, uint32_t *sad_array";
1033
specialize qw/vpx_sad16x16x3 sse3 ssse3 msa/;
Johann's avatar
Johann committed
1034 1035

add_proto qw/void vpx_sad16x8x3/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, uint32_t *sad_array";
1036
specialize qw/vpx_sad16x8x3 sse3 ssse3 msa/;
Johann's avatar
Johann committed
1037 1038

add_proto qw/void vpx_sad8x16x3/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, uint32_t *sad_array";
1039
specialize qw/vpx_sad8x16x3 sse3 msa/;
Johann's avatar
Johann committed
1040 1041

add_proto qw/void vpx_sad8x8x3/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, uint32_t *sad_array";
1042
specialize qw/vpx_sad8x8x3 sse3 msa/;
Johann's avatar
Johann committed
1043 1044

add_proto qw/void vpx_sad4x4x3/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, uint32_t *sad_array";
1045
specialize qw/vpx_sad4x4x3 sse3 msa/;
Johann's avatar
Johann committed
1046 1047 1048

# Blocks of 8
add_proto qw/void vpx_sad64x64x8/, "const uint8_t *src_ptr, int src_stride, const uint8