av1/common/av1_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 av1_common_forward_decls() {
  12 print <<EOF
  13 /*
  14  * AV1
  15  */
  16
  17 #include "aom/aom_integer.h"
  18 #include "aom_dsp/txfm_common.h"
  19 #include "av1/common/common.h"
  20 #include "av1/common/enums.h"
  21 #include "av1/common/quant_common.h"
  22 #include "av1/common/filter.h"
  23 #include "av1/common/convolve.h"
  24 #include "av1/common/av1_txfm.h"
  25 #include "av1/common/odintrin.h"
  26
  27 #if CONFIG_LOOP_RESTORATION
  28 #include "av1/common/restoration.h"
  29 #endif
  30
  31 #if CONFIG_DAALA_TX
  32 #include "av1/common/daala_inv_txfm.h"
  33 #endif
  34
  35 struct macroblockd;
  36
  37 /* Encoder forward decls */
  38 struct macroblock;
  39 struct txfm_param;
  40 struct aom_variance_vtable;
  41 struct search_site_config;
  42 struct mv;
  43 union int_mv;
  44 struct yv12_buffer_config;
  45 EOF
  46 }
  47 forward_decls qw/av1_common_forward_decls/;
  48
  49 # functions that are 64 bit only.
  50 $mmx_x86_64 = $sse2_x86_64 = $ssse3_x86_64 = $avx_x86_64 = $avx2_x86_64 = '';
  51 if ($opts{arch} eq "x86_64") {
  52   $mmx_x86_64 = 'mmx';
  53   $sse2_x86_64 = 'sse2';
  54   $ssse3_x86_64 = 'ssse3';
  55   $avx_x86_64 = 'avx';
  56   $avx2_x86_64 = 'avx2';
  57 }
  58
  59 #
  60 # 10/12-tap convolution filters
  61 #
  62 add_proto qw/void av1_lowbd_convolve_init/, "void";
  63 specialize qw/av1_lowbd_convolve_init ssse3/;
  64
  65 add_proto qw/void av1_convolve_horiz/, "const uint8_t *src, int src_stride, uint8_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, ConvolveParams *conv_params";
  66 specialize qw/av1_convolve_horiz ssse3/;
  67
  68 add_proto qw/void av1_convolve_vert/, "const uint8_t *src, int src_stride, uint8_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, ConvolveParams *conv_params";
  69 specialize qw/av1_convolve_vert ssse3/;
  70
  71 if (aom_config("CONFIG_HORZONLY_FRAME_SUPERRES") eq "yes") {
  72   add_proto qw/void av1_convolve_horiz_rs/, "const uint8_t *src, int src_stride, uint8_t *dst, int dst_stride, int w, int h, const int16_t *x_filters, int interp_taps, const int x0_qn, const int x_step_qn";
  73
  74     add_proto qw/void av1_highbd_convolve_horiz_rs/, "const uint16_t *src, int src_stride, uint16_t *dst, int dst_stride, int w, int h, const int16_t *x_filters, int interp_taps, const int x0_qn, const int x_step_qn, int bd";
  75 }
  76
  77
  78   add_proto qw/void av1_highbd_convolve_init/, "void";
  79   specialize qw/av1_highbd_convolve_init sse4_1/;
  80   add_proto qw/void av1_highbd_convolve_horiz/, "const uint16_t *src, int src_stride, uint16_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, int avg, int bd";
  81   specialize qw/av1_highbd_convolve_horiz sse4_1/;
  82   add_proto qw/void av1_highbd_convolve_vert/, "const uint16_t *src, int src_stride, uint16_t *dst, int dst_stride, int w, int h, const InterpFilterParams fp, const int subpel_x_q4, int x_step_q4, int avg, int bd";
  83   specialize qw/av1_highbd_convolve_vert sse4_1/;
  84
  85
  86 #
  87 # txb
  88 #
  89 if (aom_config("CONFIG_LV_MAP") eq "yes") {
  90   add_proto qw/void av1_get_br_level_counts/, "const uint8_t *const levels, const int width, const int height, uint8_t *const level_counts";
  91   specialize qw/av1_get_br_level_counts sse2/;
  92 }
  93
  94 #
  95 # Inverse dct
  96 #
  97 add_proto qw/void av1_iht4x4_16_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
  98 if (aom_config("CONFIG_DAALA_TX4") ne "yes") {
  99     specialize qw/av1_iht4x4_16_add sse2/;
 100 }
 101
 102 if (aom_config("CONFIG_DAALA_TX") ne "yes") {
 103   add_proto qw/void av1_iht4x8_32_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
 104   specialize qw/av1_iht4x8_32_add sse2/;
 105
 106   add_proto qw/void av1_iht8x4_32_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
 107   specialize qw/av1_iht8x4_32_add sse2/;
 108
 109   add_proto qw/void av1_iht8x16_128_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
 110   specialize qw/av1_iht8x16_128_add sse2/;
 111
 112   add_proto qw/void av1_iht16x8_128_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
 113   specialize qw/av1_iht16x8_128_add sse2/;
 114
 115   add_proto qw/void av1_iht16x32_512_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
 116   specialize qw/av1_iht16x32_512_add sse2/;
 117
 118   add_proto qw/void av1_iht32x16_512_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
 119   specialize qw/av1_iht32x16_512_add sse2/;
 120
 121   add_proto qw/void av1_iht4x16_64_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
 122
 123   add_proto qw/void av1_iht16x4_64_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
 124
 125   add_proto qw/void av1_iht8x32_256_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
 126
 127   add_proto qw/void av1_iht32x8_256_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
 128
 129   add_proto qw/void av1_iht8x8_64_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
 130   if (aom_config("CONFIG_DAALA_TX8") ne "yes") {
 131       specialize qw/av1_iht8x8_64_add sse2/;
 132   }
 133
 134   add_proto qw/void av1_iht16x16_256_add/, "const tran_low_t *input, uint8_t *output, int pitch, const struct txfm_param *param";
 135
 136   if (aom_config("CONFIG_DAALA_TX16") ne "yes") {
 137       specialize qw/av1_iht16x16_256_add sse2 avx2/;
 138   }
 139
 140   add_proto qw/void av1_iht32x32_1024_add/, "const tran_low_t *input, uint8_t *output, int pitch, const struct txfm_param *param";
 141
 142   add_proto qw/void av1_iht32x32_1024_add/, "const tran_low_t *input, uint8_t *output, int pitch, const struct txfm_param *param";
 143
 144   if (aom_config("CONFIG_TX64X64") eq "yes") {
 145     add_proto qw/void av1_iht64x64_4096_add/, "const tran_low_t *input, uint8_t *output, int pitch, const struct txfm_param *param";
 146     add_proto qw/void av1_iht32x64_2048_add/, "const tran_low_t *input, uint8_t *output, int pitch, const struct txfm_param *param";
 147     add_proto qw/void av1_iht64x32_2048_add/, "const tran_low_t *input, uint8_t *output, int pitch, const struct txfm_param *param";
 148     add_proto qw/void av1_iht16x64_1024_add/, "const tran_low_t *input, uint8_t *output, int pitch, const struct txfm_param *param";
 149     add_proto qw/void av1_iht64x16_1024_add/, "const tran_low_t *input, uint8_t *output, int pitch, const struct txfm_param *param";
 150   }
 151 }
 152
 153 if (aom_config("CONFIG_NEW_QUANT") eq "yes") {
 154   add_proto qw/void quantize_nuq/, "const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, const int16_t *dequant_ptr, const cuml_bins_type_nuq *cuml_bins_ptr, const dequant_val_type_nuq *dequant_val, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, uint16_t *eob_ptr, const int16_t *scan";
 155
 156   add_proto qw/void quantize_fp_nuq/, "const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *quant_ptr, const int16_t *dequant_ptr, const cuml_bins_type_nuq *cuml_bins_ptr, const dequant_val_type_nuq *dequant_val, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, uint16_t *eob_ptr, const int16_t *scan";
 157
 158   add_proto qw/void quantize_32x32_nuq/, "const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, const int16_t *dequant_ptr, const cuml_bins_type_nuq *cuml_bins_ptr, const dequant_val_type_nuq *dequant_val, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, uint16_t *eob_ptr, const int16_t *scan";
 159
 160   add_proto qw/void quantize_32x32_fp_nuq/, "const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *quant_ptr, const int16_t *dequant_ptr, const cuml_bins_type_nuq *cuml_bins_ptr, const dequant_val_type_nuq *dequant_val, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, uint16_t *eob_ptr, const int16_t *scan";
 161
 162   if (aom_config("CONFIG_TX64X64") eq "yes") {
 163     add_proto qw/void quantize_64x64_nuq/, "const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, const int16_t *dequant_ptr, const cuml_bins_type_nuq *cuml_bins_ptr, const dequant_val_type_nuq *dequant_val, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, uint16_t *eob_ptr, const int16_t *scan";
 164
 165     add_proto qw/void quantize_64x64_fp_nuq/, "const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *quant_ptr, const int16_t *dequant_ptr, const cuml_bins_type_nuq *cuml_bins_ptr, const dequant_val_type_nuq *dequant_val, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, uint16_t *eob_ptr, const int16_t *scan";
 166   }
 167 }
 168
 169 # FILTER_INTRA predictor functions
 170 if (aom_config("CONFIG_FILTER_INTRA") eq "yes") {
 171   add_proto qw/void av1_dc_filter_predictor/, "uint8_t *dst, ptrdiff_t stride, TX_SIZE tx_size, const uint8_t *above, const uint8_t *left";
 172   add_proto qw/void av1_v_filter_predictor/, "uint8_t *dst, ptrdiff_t stride, TX_SIZE tx_size, const uint8_t *above, const uint8_t *left";
 173   add_proto qw/void av1_h_filter_predictor/, "uint8_t *dst, ptrdiff_t stride, TX_SIZE tx_size, const uint8_t *above, const uint8_t *left";
 174   add_proto qw/void av1_d117_filter_predictor/, "uint8_t *dst, ptrdiff_t stride, TX_SIZE tx_size, const uint8_t *above, const uint8_t *left";
 175   add_proto qw/void av1_d153_filter_predictor/, "uint8_t *dst, ptrdiff_t stride, TX_SIZE tx_size, const uint8_t *above, const uint8_t *left";
 176   add_proto qw/void av1_paeth_filter_predictor/, "uint8_t *dst, ptrdiff_t stride, TX_SIZE tx_size, const uint8_t *above, const uint8_t *left";
 177   # High bitdepth functions
 178     add_proto qw/void av1_highbd_dc_filter_predictor/, "uint16_t *dst, ptrdiff_t stride, TX_SIZE tx_size, const uint16_t *above, const uint16_t *left, int bd";
 179     add_proto qw/void av1_highbd_v_filter_predictor/, "uint16_t *dst, ptrdiff_t stride, TX_SIZE tx_size, const uint16_t *above, const uint16_t *left, int bd";
 180     add_proto qw/void av1_highbd_h_filter_predictor/, "uint16_t *dst, ptrdiff_t stride, TX_SIZE tx_size, const uint16_t *above, const uint16_t *left, int bd";
 181     add_proto qw/void av1_highbd_d117_filter_predictor/, "uint16_t *dst, ptrdiff_t stride, TX_SIZE tx_size, const uint16_t *above, const uint16_t *left, int bd";
 182     add_proto qw/void av1_highbd_d153_filter_predictor/, "uint16_t *dst, ptrdiff_t stride, TX_SIZE tx_size, const uint16_t *above, const uint16_t *left, int bd";
 183     add_proto qw/void av1_highbd_paeth_filter_predictor/, "uint16_t *dst, ptrdiff_t stride, TX_SIZE tx_size, const uint16_t *above, const uint16_t *left, int bd";
 184
 185 }
 186
 187 # High bitdepth functions
 188   #
 189   # Sub Pixel Filters
 190   #
 191   add_proto qw/void av1_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";
 192
 193   add_proto qw/void av1_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";
 194
 195   add_proto qw/void av1_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";
 196   specialize qw/av1_highbd_convolve8/, "$sse2_x86_64";
 197
 198   add_proto qw/void av1_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";
 199   specialize qw/av1_highbd_convolve8_horiz/, "$sse2_x86_64";
 200
 201   add_proto qw/void av1_highbd_convolve8_vert/, "const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps";
 202   specialize qw/av1_highbd_convolve8_vert/, "$sse2_x86_64";
 203
 204   add_proto qw/void av1_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";
 205   specialize qw/av1_highbd_convolve8_avg/, "$sse2_x86_64";
 206
 207   add_proto qw/void av1_highbd_convolve8_avg_horiz/, "const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h, int bps";
 208   specialize qw/av1_highbd_convolve8_avg_horiz/, "$sse2_x86_64";
 209
 210   add_proto qw/void av1_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";
 211   specialize qw/av1_highbd_convolve8_avg_vert/, "$sse2_x86_64";
 212
 213   #
 214   # dct
 215   #
 216   add_proto qw/void av1_highbd_iht4x4_16_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
 217
 218   add_proto qw/void av1_highbd_iht4x8_32_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
 219
 220   add_proto qw/void av1_highbd_iht8x4_32_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
 221
 222   add_proto qw/void av1_highbd_iht8x16_128_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
 223
 224   add_proto qw/void av1_highbd_iht16x8_128_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
 225
 226   add_proto qw/void av1_highbd_iht16x32_512_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
 227
 228   add_proto qw/void av1_highbd_iht32x16_512_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
 229
 230   add_proto qw/void av1_highbd_iht4x16_64_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
 231
 232   add_proto qw/void av1_highbd_iht16x4_64_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
 233
 234   add_proto qw/void av1_highbd_iht8x32_256_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
 235
 236   add_proto qw/void av1_highbd_iht32x8_256_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
 237
 238   add_proto qw/void av1_highbd_iht8x8_64_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
 239
 240   add_proto qw/void av1_highbd_iht16x16_256_add/, "const tran_low_t *input, uint8_t *output, int pitch, const struct txfm_param *param";
 241
 242 #inv txfm
 243 add_proto qw/void av1_inv_txfm2d_add_4x8/, "const int32_t *input, uint16_t *output, int stride, TX_TYPE tx_type, int bd";
 244 add_proto qw/void av1_inv_txfm2d_add_8x4/, "const int32_t *input, uint16_t *output, int stride, TX_TYPE tx_type, int bd";
 245 add_proto qw/void av1_inv_txfm2d_add_8x16/, "const int32_t *input, uint16_t *output, int stride, TX_TYPE tx_type, int bd";
 246 add_proto qw/void av1_inv_txfm2d_add_16x8/, "const int32_t *input, uint16_t *output, int stride, TX_TYPE tx_type, int bd";
 247 add_proto qw/void av1_inv_txfm2d_add_16x32/, "const int32_t *input, uint16_t *output, int stride, TX_TYPE tx_type, int bd";
 248 add_proto qw/void av1_inv_txfm2d_add_32x16/, "const int32_t *input, uint16_t *output, int stride, TX_TYPE tx_type, int bd";
 249 add_proto qw/void av1_inv_txfm2d_add_4x4/, "const int32_t *input, uint16_t *output, int stride, TX_TYPE tx_type, int bd";
 250 if (aom_config("CONFIG_DAALA_TX4") ne "yes") {
 251   specialize qw/av1_inv_txfm2d_add_4x4 sse4_1/;
 252 }
 253 add_proto qw/void av1_inv_txfm2d_add_8x8/, "const int32_t *input, uint16_t *output, int stride, TX_TYPE tx_type, int bd";
 254 if (aom_config("CONFIG_DAALA_TX8") ne "yes") {
 255   specialize qw/av1_inv_txfm2d_add_8x8 sse4_1/;
 256 }
 257 add_proto qw/void av1_inv_txfm2d_add_16x16/, "const int32_t *input, uint16_t *output, int stride, TX_TYPE tx_type, int bd";
 258 if (aom_config("CONFIG_DAALA_TX16") ne "yes") {
 259   specialize qw/av1_inv_txfm2d_add_16x16 sse4_1/;
 260 }
 261 add_proto qw/void av1_inv_txfm2d_add_32x32/, "const int32_t *input, uint16_t *output, int stride, TX_TYPE tx_type, int bd";
 262 if (aom_config("CONFIG_DAALA_TX32") ne "yes") {
 263   specialize qw/av1_inv_txfm2d_add_32x32 avx2/;
 264 }
 265 if (aom_config("CONFIG_TX64X64") eq "yes") {
 266   add_proto qw/void av1_inv_txfm2d_add_64x64/, "const int32_t *input, uint16_t *output, int stride, TX_TYPE tx_type, int bd";
 267   add_proto qw/void av1_inv_txfm2d_add_32x64/, "const int32_t *input, uint16_t *output, int stride, TX_TYPE tx_type, int bd";
 268   add_proto qw/void av1_inv_txfm2d_add_64x32/, "const int32_t *input, uint16_t *output, int stride, TX_TYPE tx_type, int bd";
 269   add_proto qw/void av1_inv_txfm2d_add_16x64/, "const int32_t *input, uint16_t *output, int stride, TX_TYPE tx_type, int bd";
 270   add_proto qw/void av1_inv_txfm2d_add_64x16/, "const int32_t *input, uint16_t *output, int stride, TX_TYPE tx_type, int bd";
 271 }
 272 add_proto qw/void av1_inv_txfm2d_add_4x16/, "const int32_t *input, uint16_t *output, int stride, TX_TYPE tx_type, int bd";
 273 add_proto qw/void av1_inv_txfm2d_add_16x4/, "const int32_t *input, uint16_t *output, int stride, TX_TYPE tx_type, int bd";
 274 add_proto qw/void av1_inv_txfm2d_add_8x32/, "const int32_t *input, uint16_t *output, int stride, TX_TYPE tx_type, int bd";
 275 add_proto qw/void av1_inv_txfm2d_add_32x8/, "const int32_t *input, uint16_t *output, int stride, TX_TYPE tx_type, int bd";
 276
 277 #
 278 # Encoder functions below this point.
 279 #
 280 if (aom_config("CONFIG_AV1_ENCODER") eq "yes") {
 281
 282   # ENCODEMB INVOKE
 283
 284     # the transform coefficients are held in 32-bit
 285     # values, so the assembler code for  av1_block_error can no longer be used.
 286     add_proto qw/int64_t av1_block_error/, "const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz";
 287     specialize qw/av1_block_error avx2/;
 288
 289     add_proto qw/void av1_quantize_fp/, "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";
 290     specialize qw/av1_quantize_fp sse2 avx2/;
 291
 292     add_proto qw/void av1_quantize_fp_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";
 293     specialize qw/av1_quantize_fp_32x32 avx2/;
 294
 295     if (aom_config("CONFIG_TX64X64") eq "yes") {
 296       add_proto qw/void av1_quantize_fp_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";
 297     }
 298   # fdct functions
 299
 300   add_proto qw/void av1_fht4x4/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
 301   if (aom_config("CONFIG_DAALA_TX4") ne "yes") {
 302     specialize qw/av1_fht4x4 sse2/;
 303   }
 304
 305   add_proto qw/void av1_fwht4x4/, "const int16_t *input, tran_low_t *output, int stride";
 306
 307   add_proto qw/void av1_fht8x8/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
 308   if (aom_config("CONFIG_DAALA_TX8") ne "yes") {
 309     specialize qw/av1_fht8x8 sse2/;
 310   }
 311
 312   add_proto qw/void av1_fht16x16/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
 313   if (aom_config("CONFIG_DAALA_TX16") ne "yes") {
 314     specialize qw/av1_fht16x16 sse2 avx2/;
 315   }
 316
 317   add_proto qw/void av1_fht32x32/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
 318   if (aom_config("CONFIG_DAALA_TX32") ne "yes") {
 319     specialize qw/av1_fht32x32 sse2 avx2/;
 320   }
 321
 322   if (aom_config("CONFIG_TX64X64") eq "yes") {
 323     add_proto qw/void av1_fht64x64/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
 324     add_proto qw/void av1_fht32x64/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
 325     add_proto qw/void av1_fht64x32/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
 326     add_proto qw/void av1_fht16x64/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
 327     add_proto qw/void av1_fht64x16/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
 328   }
 329
 330   add_proto qw/void av1_fht4x8/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
 331   specialize qw/av1_fht4x8 sse2/;
 332
 333   add_proto qw/void av1_fht8x4/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
 334   specialize qw/av1_fht8x4 sse2/;
 335
 336   add_proto qw/void av1_fht8x16/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
 337   specialize qw/av1_fht8x16 sse2/;
 338
 339   add_proto qw/void av1_fht16x8/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
 340   specialize qw/av1_fht16x8 sse2/;
 341
 342   add_proto qw/void av1_fht16x32/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
 343   specialize qw/av1_fht16x32 sse2/;
 344
 345   add_proto qw/void av1_fht32x16/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
 346   specialize qw/av1_fht32x16 sse2/;
 347
 348   add_proto qw/void av1_fht4x16/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
 349
 350   add_proto qw/void av1_fht16x4/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
 351
 352   add_proto qw/void av1_fht8x32/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
 353
 354   add_proto qw/void av1_fht32x8/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
 355
 356
 357   add_proto qw/void av1_fwd_idtx/, "const int16_t *src_diff, tran_low_t *coeff, int stride, int bsx, int bsy, TX_TYPE tx_type";
 358
 359   #fwd txfm
 360   add_proto qw/void av1_fwd_txfm2d_4x8/, "const int16_t *input, int32_t *output, int stride, TX_TYPE tx_type, int bd";
 361   add_proto qw/void av1_fwd_txfm2d_8x4/, "const int16_t *input, int32_t *output, int stride, TX_TYPE tx_type, int bd";
 362   add_proto qw/void av1_fwd_txfm2d_8x16/, "const int16_t *input, int32_t *output, int stride, TX_TYPE tx_type, int bd";
 363   add_proto qw/void av1_fwd_txfm2d_16x8/, "const int16_t *input, int32_t *output, int stride, TX_TYPE tx_type, int bd";
 364   add_proto qw/void av1_fwd_txfm2d_16x32/, "const int16_t *input, int32_t *output, int stride, TX_TYPE tx_type, int bd";
 365   add_proto qw/void av1_fwd_txfm2d_32x16/, "const int16_t *input, int32_t *output, int stride, TX_TYPE tx_type, int bd";
 366   add_proto qw/void av1_fwd_txfm2d_4x16/, "const int16_t *input, int32_t *output, int stride, TX_TYPE tx_type, int bd";
 367   add_proto qw/void av1_fwd_txfm2d_16x4/, "const int16_t *input, int32_t *output, int stride, TX_TYPE tx_type, int bd";
 368   add_proto qw/void av1_fwd_txfm2d_8x32/, "const int16_t *input, int32_t *output, int stride, TX_TYPE tx_type, int bd";
 369   add_proto qw/void av1_fwd_txfm2d_32x8/, "const int16_t *input, int32_t *output, int stride, TX_TYPE tx_type, int bd";
 370   add_proto qw/void av1_fwd_txfm2d_4x4/, "const int16_t *input, int32_t *output, int stride, TX_TYPE tx_type, int bd";
 371   if (aom_config("CONFIG_DAALA_TX4") ne "yes") {
 372     specialize qw/av1_fwd_txfm2d_4x4 sse4_1/;
 373   }
 374   add_proto qw/void av1_fwd_txfm2d_8x8/, "const int16_t *input, int32_t *output, int stride, TX_TYPE tx_type, int bd";
 375   if (aom_config("CONFIG_DAALA_TX8") ne "yes") {
 376     specialize qw/av1_fwd_txfm2d_8x8 sse4_1/;
 377   }
 378   add_proto qw/void av1_fwd_txfm2d_16x16/, "const int16_t *input, int32_t *output, int stride, TX_TYPE tx_type, int bd";
 379   if (aom_config("CONFIG_DAALA_TX16") ne "yes") {
 380     specialize qw/av1_fwd_txfm2d_16x16 sse4_1/;
 381   }
 382   add_proto qw/void av1_fwd_txfm2d_32x32/, "const int16_t *input, int32_t *output, int stride, TX_TYPE tx_type, int bd";
 383   if (aom_config("CONFIG_DAALA_TX32") ne "yes") {
 384     specialize qw/av1_fwd_txfm2d_32x32 sse4_1/;
 385   }
 386
 387   if (aom_config("CONFIG_TX64X64") eq "yes") {
 388     add_proto qw/void av1_fwd_txfm2d_64x64/, "const int16_t *input, int32_t *output, int stride, TX_TYPE tx_type, int bd";
 389     add_proto qw/void av1_fwd_txfm2d_32x64/, "const int16_t *input, int32_t *output, int stride, TX_TYPE tx_type, int bd";
 390     add_proto qw/void av1_fwd_txfm2d_64x32/, "const int16_t *input, int32_t *output, int stride, TX_TYPE tx_type, int bd";
 391     add_proto qw/void av1_fwd_txfm2d_16x64/, "const int16_t *input, int32_t *output, int stride, TX_TYPE tx_type, int bd";
 392     add_proto qw/void av1_fwd_txfm2d_64x16/, "const int16_t *input, int32_t *output, int stride, TX_TYPE tx_type, int bd";
 393   }
 394   #
 395   # Motion search
 396   #
 397   add_proto qw/int av1_full_search_sad/, "const struct macroblock *x, const struct mv *ref_mv, int sad_per_bit, int distance, const struct aom_variance_vtable *fn_ptr, const struct mv *center_mv, struct mv *best_mv";
 398   specialize qw/av1_full_search_sad sse3 sse4_1/;
 399   $av1_full_search_sad_sse3=av1_full_search_sadx3;
 400   $av1_full_search_sad_sse4_1=av1_full_search_sadx8;
 401
 402   add_proto qw/int av1_diamond_search_sad/, "struct macroblock *x, const struct search_site_config *cfg,  struct mv *ref_mv, struct mv *best_mv, int search_param, int sad_per_bit, int *num00, const struct aom_variance_vtable *fn_ptr, const struct mv *center_mv";
 403
 404   add_proto qw/int av1_full_range_search/, "const struct macroblock *x, const struct search_site_config *cfg, struct mv *ref_mv, struct mv *best_mv, int search_param, int sad_per_bit, int *num00, const struct aom_variance_vtable *fn_ptr, const struct mv *center_mv";
 405
 406   add_proto qw/void av1_temporal_filter_apply/, "uint8_t *frame1, unsigned int stride, uint8_t *frame2, unsigned int block_width, unsigned int block_height, int strength, int filter_weight, unsigned int *accumulator, uint16_t *count";
 407   specialize qw/av1_temporal_filter_apply sse2 msa/;
 408
 409   if (aom_config("CONFIG_AOM_QM") eq "yes") {
 410     add_proto qw/void av1_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, const qm_val_t * qm_ptr, const qm_val_t * iqm_ptr, int log_scale";
 411   } else {
 412     add_proto qw/void av1_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, int log_scale";
 413   }
 414
 415
 416     # ENCODEMB INVOKE
 417     if (aom_config("CONFIG_NEW_QUANT") eq "yes") {
 418       add_proto qw/void highbd_quantize_nuq/, "const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, const int16_t *dequant_ptr, const cuml_bins_type_nuq *cuml_bins_ptr, const dequant_val_type_nuq *dequant_val, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, uint16_t *eob_ptr, const int16_t *scan";
 419
 420       add_proto qw/void highbd_quantize_fp_nuq/, "const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *quant_ptr, const int16_t *dequant_ptr, const cuml_bins_type_nuq *cuml_bins_ptr, const dequant_val_type_nuq *dequant_val, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, uint16_t *eob_ptr, const int16_t *scan";
 421
 422       add_proto qw/void highbd_quantize_32x32_nuq/, "const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, const int16_t *dequant_ptr, const cuml_bins_type_nuq *cuml_bins_ptr, const dequant_val_type_nuq *dequant_val, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, uint16_t *eob_ptr, const int16_t *scan";
 423
 424       add_proto qw/void highbd_quantize_32x32_fp_nuq/, "const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *quant_ptr, const int16_t *dequant_ptr, const cuml_bins_type_nuq *cuml_bins_ptr, const dequant_val_type_nuq *dequant_val, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, uint16_t *eob_ptr, const int16_t *scan";
 425
 426       if (aom_config("CONFIG_TX64X64") eq "yes") {
 427         add_proto qw/void highbd_quantize_64x64_nuq/, "const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, const int16_t *dequant_ptr, const cuml_bins_type_nuq *cuml_bins_ptr, const dequant_val_type_nuq *dequant_val, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, uint16_t *eob_ptr, const int16_t *scan";
 428
 429         add_proto qw/void highbd_quantize_64x64_fp_nuq/, "const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *quant_ptr, const int16_t *dequant_ptr, const cuml_bins_type_nuq *cuml_bins_ptr, const dequant_val_type_nuq *dequant_val, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, uint16_t *eob_ptr, const int16_t *scan";
 430       }
 431     }
 432
 433     add_proto qw/int64_t av1_highbd_block_error/, "const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz, int bd";
 434     if (aom_config("CONFIG_DAALA_TX") ne "yes") {
 435       specialize qw/av1_highbd_block_error sse2/;
 436     }
 437
 438     add_proto qw/void av1_highbd_temporal_filter_apply/, "uint8_t *frame1, unsigned int stride, uint8_t *frame2, unsigned int block_width, unsigned int block_height, int strength, int filter_weight, unsigned int *accumulator, uint16_t *count";
 439
 440
 441   add_proto qw/void av1_highbd_quantize_fp/, "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, int log_scale";
 442   specialize qw/av1_highbd_quantize_fp sse4_1 avx2/;
 443
 444   add_proto qw/void av1_highbd_fwht4x4/, "const int16_t *input, tran_low_t *output, int stride";
 445
 446   # End av1_high encoder functions
 447
 448   # txb
 449   if (aom_config("CONFIG_LV_MAP") eq "yes") {
 450     add_proto qw/void av1_get_nz_map_contexts/, "const uint8_t *const levels, const int16_t *const scan, const uint16_t eob, const TX_SIZE tx_size, const TX_TYPE tx_type, int8_t *const coeff_contexts";
 451     specialize qw/av1_get_nz_map_contexts sse2/;
 452   }
 453
 454   add_proto qw/uint64_t av1_wedge_sse_from_residuals/, "const int16_t *r1, const int16_t *d, const uint8_t *m, int N";
 455   specialize qw/av1_wedge_sse_from_residuals sse2/;
 456   add_proto qw/int av1_wedge_sign_from_residuals/, "const int16_t *ds, const uint8_t *m, int N, int64_t limit";
 457   specialize qw/av1_wedge_sign_from_residuals sse2/;
 458   add_proto qw/void av1_wedge_compute_delta_squares/, "int16_t *d, const int16_t *a, const int16_t *b, int N";
 459   specialize qw/av1_wedge_compute_delta_squares sse2/;
 460
 461 }
 462 # end encoder functions
 463
 464 # Deringing Functions
 465
 466 add_proto qw/int cdef_find_dir/, "const uint16_t *img, int stride, int32_t *var, int coeff_shift";
 467 add_proto qw/void cdef_filter_block/, "uint8_t *dst8, uint16_t *dst16, int dstride, const uint16_t *in, int pri_strength, int sec_strength, int dir, int pri_damping, int sec_damping, int bsize, int max, int coeff_shift";
 468
 469 add_proto qw/void copy_rect8_8bit_to_16bit/, "uint16_t *dst, int dstride, const uint8_t *src, int sstride, int v, int h";
 470 add_proto qw/void copy_rect8_16bit_to_16bit/, "uint16_t *dst, int dstride, const uint16_t *src, int sstride, int v, int h";
 471
 472 # VS compiling for 32 bit targets does not support vector types in
 473 # structs as arguments, which makes the v256 type of the intrinsics
 474 # hard to support, so optimizations for this target are disabled.
 475 if ($opts{config} !~ /libs-x86-win32-vs.*/) {
 476   specialize qw/cdef_find_dir sse2 ssse3 sse4_1 avx2 neon/;
 477   specialize qw/cdef_filter_block sse2 ssse3 sse4_1 avx2 neon/;
 478   specialize qw/copy_rect8_8bit_to_16bit sse2 ssse3 sse4_1 avx2 neon/;
 479   specialize qw/copy_rect8_16bit_to_16bit sse2 ssse3 sse4_1 avx2 neon/;
 480 }
 481
 482 # WARPED_MOTION / GLOBAL_MOTION functions
 483
 484 add_proto qw/void av1_warp_affine/, "const int32_t *mat, const uint8_t *ref, int width, int height, int stride, uint8_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, ConvolveParams *conv_params, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta";
 485
 486 if (aom_config("CONFIG_JNT_COMP") eq "yes") {
 487   if (aom_config("CONFIG_JNT_COMP") eq "yes") {
 488     specialize qw/av1_warp_affine sse4_1/;
 489   }
 490 } else {
 491   specialize qw/av1_warp_affine sse2 ssse3/;
 492 }
 493
 494   add_proto qw/void av1_highbd_warp_affine/, "const int32_t *mat, const uint16_t *ref, int width, int height, int stride, uint16_t *pred, int p_col, int p_row, int p_width, int p_height, int p_stride, int subsampling_x, int subsampling_y, int bd, ConvolveParams *conv_params, int16_t alpha, int16_t beta, int16_t gamma, int16_t delta";
 495
 496 if (aom_config("CONFIG_JNT_COMP") eq "yes") {
 497   if (aom_config("CONFIG_JNT_COMP") eq "yes") {
 498     specialize qw/av1_highbd_warp_affine sse4_1/;
 499   }
 500 } else {
 501   specialize qw/av1_highbd_warp_affine ssse3/;
 502 }
 503
 504
 505 if (aom_config("CONFIG_AV1_ENCODER") eq "yes") {
 506   add_proto qw/double compute_cross_correlation/, "unsigned char *im1, int stride1, int x1, int y1, unsigned char *im2, int stride2, int x2, int y2";
 507   specialize qw/compute_cross_correlation sse4_1/;
 508 }
 509
 510 # LOOP_RESTORATION functions
 511
 512 if (aom_config("CONFIG_LOOP_RESTORATION") eq "yes") {
 513   add_proto qw/void apply_selfguided_restoration/, "const uint8_t *dat, int width, int height, int stride, int eps, const int *xqd, uint8_t *dst, int dst_stride, int32_t *tmpbuf, int bit_depth, int highbd";
 514   specialize qw/apply_selfguided_restoration sse4_1/;
 515
 516   add_proto qw/void av1_selfguided_restoration/, "const uint8_t *dgd, int width, int height, int stride, int32_t *flt1, int32_t *flt2, int flt_stride, const sgr_params_type *params, int bit_depth, int highbd";
 517   specialize qw/av1_selfguided_restoration sse4_1/;
 518 }
 519
 520 # CONVOLVE_ROUND/COMPOUND_ROUND functions
 521 add_proto qw/void av1_convolve_2d/, "const uint8_t *src, int src_stride, uint8_t *dst, int dst_stride, int w, int h, InterpFilterParams *filter_params_x, InterpFilterParams *filter_params_y, const int subpel_x_q4, const int subpel_y_q4, ConvolveParams *conv_params";
 522 specialize qw/av1_convolve_2d sse2 avx2/;
 523 add_proto qw/void av1_convolve_2d_sr/, "const uint8_t *src, int src_stride, uint8_t *dst, int dst_stride, int w, int h, InterpFilterParams *filter_params_x, InterpFilterParams *filter_params_y, const int subpel_x_q4, const int subpel_y_q4, ConvolveParams *conv_params";
 524 specialize qw/av1_convolve_2d_sr c/;
 525 add_proto qw/void av1_convolve_rounding/, "const int32_t *src, int src_stride, uint8_t *dst, int dst_stride, int w, int h, int bits";
 526 specialize qw/av1_convolve_rounding avx2/;
 527
 528 add_proto qw/void av1_convolve_2d_copy/, "const uint8_t *src, int src_stride, uint8_t *dst, int dst_stride, int w, int h, InterpFilterParams *filter_params_x, InterpFilterParams *filter_params_y, const int subpel_x_q4, const int subpel_y_q4, ConvolveParams *conv_params";
 529 specialize qw/av1_convolve_2d_copy sse2/;
 530 add_proto qw/void av1_convolve_2d_copy_sr/, "const uint8_t *src, int src_stride, uint8_t *dst, int dst_stride, int w, int h, InterpFilterParams *filter_params_x, InterpFilterParams *filter_params_y, const int subpel_x_q4, const int subpel_y_q4, ConvolveParams *conv_params";
 531 specialize qw/av1_convolve_2d_copy_sr c/;
 532 add_proto qw/void av1_convolve_x/, "const uint8_t *src, int src_stride, uint8_t *dst, int dst_stride, int w, int h, InterpFilterParams *filter_params_x, InterpFilterParams *filter_params_y, const int subpel_x_q4, const int subpel_y_q4, ConvolveParams *conv_params";
 533 specialize qw/av1_convolve_x sse2 avx2/;
 534 add_proto qw/void av1_convolve_y/, "const uint8_t *src, int src_stride, uint8_t *dst, int dst_stride, int w, int h, InterpFilterParams *filter_params_x, InterpFilterParams *filter_params_y, const int subpel_x_q4, const int subpel_y_q4, ConvolveParams *conv_params";
 535 specialize qw/av1_convolve_y sse2 avx2/;
 536 add_proto qw/void av1_convolve_x_sr/, "const uint8_t *src, int src_stride, uint8_t *dst, int dst_stride, int w, int h, InterpFilterParams *filter_params_x, InterpFilterParams *filter_params_y, const int subpel_x_q4, const int subpel_y_q4, ConvolveParams *conv_params";
 537 specialize qw/av1_convolve_x_sr c/;
 538 add_proto qw/void av1_convolve_y_sr/, "const uint8_t *src, int src_stride, uint8_t *dst, int dst_stride, int w, int h, InterpFilterParams *filter_params_x, InterpFilterParams *filter_params_y, const int subpel_x_q4, const int subpel_y_q4, ConvolveParams *conv_params";
 539 specialize qw/av1_convolve_y_sr c/;
 540
 541 add_proto qw/void av1_convolve_2d_scale/, "const uint8_t *src, int src_stride, CONV_BUF_TYPE *dst, int dst_stride, int w, int h, InterpFilterParams *filter_params_x, InterpFilterParams *filter_params_y, const int subpel_x_qn, const int x_step_qn, const int subpel_y_q4, const int y_step_qn, ConvolveParams *conv_params";
 542 specialize qw/av1_convolve_2d_scale sse4_1/;
 543
 544 if (aom_config("CONFIG_JNT_COMP") eq "yes") {
 545   add_proto qw/void av1_jnt_convolve_2d/, "const uint8_t *src, int src_stride, uint8_t *dst, int dst_stride, int w, int h, InterpFilterParams *filter_params_x, InterpFilterParams *filter_params_y, const int subpel_x_q4, const int subpel_y_q4, ConvolveParams *conv_params";
 546   specialize qw/av1_jnt_convolve_2d sse4_1/;
 547
 548   add_proto qw/void av1_jnt_convolve_2d_copy/, "const uint8_t *src, int src_stride, uint8_t *dst, int dst_stride, int w, int h, InterpFilterParams *filter_params_x, InterpFilterParams *filter_params_y, const int subpel_x_q4, const int subpel_y_q4, ConvolveParams *conv_params";
 549   specialize qw/av1_jnt_convolve_2d_copy sse2/;
 550 }
 551
 552   add_proto qw/void av1_highbd_convolve_2d/, "const uint16_t *src, int src_stride, CONV_BUF_TYPE *dst, int dst_stride, int w, int h, InterpFilterParams *filter_params_x, InterpFilterParams *filter_params_y, const int subpel_x_q4, const int subpel_y_q4, ConvolveParams *conv_params, int bd";
 553   specialize qw/av1_highbd_convolve_2d ssse3 avx2/;
 554   add_proto qw/void av1_highbd_convolve_rounding/, "const int32_t *src, int src_stride, uint8_t *dst, int dst_stride, int w, int h, int bits, int bd";
 555   specialize qw/av1_highbd_convolve_rounding avx2/;
 556
 557   add_proto qw/void av1_highbd_convolve_2d_scale/, "const uint16_t *src, int src_stride, CONV_BUF_TYPE *dst, int dst_stride, int w, int h, InterpFilterParams *filter_params_x, InterpFilterParams *filter_params_y, const int subpel_x_q4, const int x_step_qn, const int subpel_y_q4, const int y_step_qn, ConvolveParams *conv_params, int bd";
 558   specialize qw/av1_highbd_convolve_2d_scale sse4_1/;
 559
 560   if (aom_config("CONFIG_JNT_COMP") eq "yes") {
 561     add_proto qw/void av1_highbd_jnt_convolve_2d/, "const uint16_t *src, int src_stride, CONV_BUF_TYPE *dst, int dst_stride, int w, int h, InterpFilterParams *filter_params_x, InterpFilterParams *filter_params_y, const int subpel_x_q4, const int subpel_y_q4, ConvolveParams *conv_params, int bd";
 562     specialize qw/av1_highbd_jnt_convolve_2d sse4_1/;
 563   }
 564
 565
 566 # INTRA_EDGE functions
 567 if (aom_config("CONFIG_INTRA_EDGE") eq "yes") {
 568   add_proto qw/void av1_filter_intra_edge/, "uint8_t *p, int sz, int strength";
 569   specialize qw/av1_filter_intra_edge sse4_1/;
 570   add_proto qw/void av1_upsample_intra_edge/, "uint8_t *p, int sz";
 571   specialize qw/av1_upsample_intra_edge sse4_1/;
 572
 573     add_proto qw/void av1_filter_intra_edge_high/, "uint16_t *p, int sz, int strength";
 574     specialize qw/av1_filter_intra_edge_high sse4_1/;
 575     add_proto qw/void av1_upsample_intra_edge_high/, "uint16_t *p, int sz, int bd";
 576     specialize qw/av1_upsample_intra_edge_high sse4_1/;
 577
 578 }
 579
 580 # DAALA_TX functions
 581 if (aom_config("CONFIG_DAALA_TX") eq "yes") {
 582   add_proto qw/void daala_inv_txfm_add/, "const tran_low_t *input_coeffs, void *output_pixels, int output_stride, TxfmParam *txfm_param";
 583   specialize qw/daala_inv_txfm_add avx2/;
 584 }
 585
 586 # CFL
 587 if (aom_config("CONFIG_CFL") eq "yes") {
 588   add_proto qw/void av1_cfl_subtract/, "int16_t *pred_buf_q3, int width, int height, int16_t avg_q3";
 589   specialize qw/av1_cfl_subtract sse2 avx2/;
 590
 591 }
 592
 593 1;