Update aom to commit id e87fb2378f01103d5d6e477a4ef6892dc714e614

1 files changed, 292 insertions, 283 deletions

diff --git a/third_party/aom/av1/common/av1_rtcd_defs.pl b/third_party/aom/av1/common/av1_rtcd_defs.pl
index 88e9ea4d8d..203426e59b 100755
--- a/third_party/aom/av1/common/av1_rtcd_defs.pl
+++ b/third_party/aom/av1/common/av1_rtcd_defs.pl
@@ -24,7 +24,6 @@ struct search_site_config;
 struct mv;
 union int_mv;
 struct yv12_buffer_config;
-typedef uint16_t od_dering_in;
 EOF
 }
 forward_decls qw/av1_common_forward_decls/;
@@ -64,86 +63,94 @@ if (aom_config("CONFIG_HIGHBITDEPTH") eq "yes") {
 # Inverse dct
 #
 if (aom_config("CONFIG_HIGHBITDEPTH") eq "yes") {
-  {
-    add_proto qw/void av1_iht4x4_16_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
-      specialize qw/av1_iht4x4_16_add sse2/;
+  add_proto qw/void av1_iht4x4_16_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
+  specialize qw/av1_iht4x4_16_add sse2/;
 
-    add_proto qw/void av1_iht4x8_32_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
-      specialize qw/av1_iht4x8_32_add sse2/;
+  add_proto qw/void av1_iht4x8_32_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
+  specialize qw/av1_iht4x8_32_add sse2/;
 
-    add_proto qw/void av1_iht8x4_32_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
-      specialize qw/av1_iht8x4_32_add sse2/;
+  add_proto qw/void av1_iht8x4_32_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
+  specialize qw/av1_iht8x4_32_add sse2/;
 
-    add_proto qw/void av1_iht8x16_128_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
-      specialize qw/av1_iht8x16_128_add sse2/;
+  add_proto qw/void av1_iht8x16_128_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
+  specialize qw/av1_iht8x16_128_add sse2/;
 
-    add_proto qw/void av1_iht16x8_128_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
-      specialize qw/av1_iht16x8_128_add sse2/;
+  add_proto qw/void av1_iht16x8_128_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
+  specialize qw/av1_iht16x8_128_add sse2/;
 
-    add_proto qw/void av1_iht16x32_512_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
-      specialize qw/av1_iht16x32_512_add sse2/;
+  add_proto qw/void av1_iht16x32_512_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
+  specialize qw/av1_iht16x32_512_add sse2/;
 
-    add_proto qw/void av1_iht32x16_512_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
-      specialize qw/av1_iht32x16_512_add sse2/;
+  add_proto qw/void av1_iht32x16_512_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
+  specialize qw/av1_iht32x16_512_add sse2/;
 
-    add_proto qw/void av1_iht4x16_64_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
+  add_proto qw/void av1_iht4x16_64_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
 
-    add_proto qw/void av1_iht16x4_64_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
+  add_proto qw/void av1_iht16x4_64_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
 
-    add_proto qw/void av1_iht8x32_256_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
+  add_proto qw/void av1_iht8x32_256_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
 
-    add_proto qw/void av1_iht32x8_256_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
+  add_proto qw/void av1_iht32x8_256_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
 
-    add_proto qw/void av1_iht8x8_64_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
-      specialize qw/av1_iht8x8_64_add sse2/;
+  add_proto qw/void av1_iht8x8_64_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
+  specialize qw/av1_iht8x8_64_add sse2/;
 
-    add_proto qw/void av1_iht16x16_256_add/, "const tran_low_t *input, uint8_t *output, int pitch, const struct txfm_param *param";
-      specialize qw/av1_iht16x16_256_add sse2 avx2/;
+  add_proto qw/void av1_iht16x16_256_add/, "const tran_low_t *input, uint8_t *output, int pitch, const struct txfm_param *param";
+  specialize qw/av1_iht16x16_256_add sse2 avx2/;
 
-    add_proto qw/void av1_iht32x32_1024_add/, "const tran_low_t *input, uint8_t *output, int pitch, const struct txfm_param *param";
-  }
+  add_proto qw/void av1_iht32x32_1024_add/, "const tran_low_t *input, uint8_t *output, int pitch, const struct txfm_param *param";
 } else {
-  {
-    add_proto qw/void av1_iht4x4_16_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
-    specialize qw/av1_iht4x4_16_add sse2 neon dspr2/;
+  add_proto qw/void av1_iht4x4_16_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
+  if (aom_config("CONFIG_DAALA_DCT4") ne "yes") {
+    specialize qw/av1_iht4x4_16_add sse2 neon/;
+  }
 
-    add_proto qw/void av1_iht4x8_32_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
-      specialize qw/av1_iht4x8_32_add sse2/;
+  add_proto qw/void av1_iht4x8_32_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
+  specialize qw/av1_iht4x8_32_add sse2/;
 
-    add_proto qw/void av1_iht8x4_32_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
-      specialize qw/av1_iht8x4_32_add sse2/;
+  add_proto qw/void av1_iht8x4_32_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
+  specialize qw/av1_iht8x4_32_add sse2/;
 
-    add_proto qw/void av1_iht8x16_128_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
-      specialize qw/av1_iht8x16_128_add sse2/;
+  add_proto qw/void av1_iht8x16_128_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
+  specialize qw/av1_iht8x16_128_add sse2/;
 
-    add_proto qw/void av1_iht16x8_128_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
-      specialize qw/av1_iht16x8_128_add sse2/;
+  add_proto qw/void av1_iht16x8_128_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
+  specialize qw/av1_iht16x8_128_add sse2/;
 
-    add_proto qw/void av1_iht16x32_512_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
-      specialize qw/av1_iht16x32_512_add sse2/;
+  add_proto qw/void av1_iht16x32_512_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
+  specialize qw/av1_iht16x32_512_add sse2/;
 
-    add_proto qw/void av1_iht32x16_512_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
-      specialize qw/av1_iht32x16_512_add sse2/;
+  add_proto qw/void av1_iht32x16_512_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
+  specialize qw/av1_iht32x16_512_add sse2/;
 
-    add_proto qw/void av1_iht4x16_64_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
+  add_proto qw/void av1_iht4x16_64_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
 
-    add_proto qw/void av1_iht16x4_64_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
+  add_proto qw/void av1_iht16x4_64_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
 
-    add_proto qw/void av1_iht8x32_256_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
+  add_proto qw/void av1_iht8x32_256_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
 
-    add_proto qw/void av1_iht32x8_256_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
+  add_proto qw/void av1_iht32x8_256_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
 
-    add_proto qw/void av1_iht8x8_64_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
-      specialize qw/av1_iht8x8_64_add sse2 neon dspr2/;
+  add_proto qw/void av1_iht8x8_64_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, const struct txfm_param *param";
+  if (aom_config("CONFIG_DAALA_DCT8") ne "yes") {
+    specialize qw/av1_iht8x8_64_add sse2 neon/;
+  }
 
-    add_proto qw/void av1_iht16x16_256_add/, "const tran_low_t *input, uint8_t *output, int pitch, const struct txfm_param *param";
-      specialize qw/av1_iht16x16_256_add sse2 avx2 dspr2/;
+  add_proto qw/void av1_iht16x16_256_add/, "const tran_low_t *input, uint8_t *output, int pitch, const struct txfm_param *param";
+  if (aom_config("CONFIG_DAALA_DCT16") ne "yes") {
+    specialize qw/av1_iht16x16_256_add sse2 avx2/;
+  }
 
-    add_proto qw/void av1_iht32x32_1024_add/, "const tran_low_t *input, uint8_t *output, int pitch, const struct txfm_param *param";
+  add_proto qw/void av1_iht32x32_1024_add/, "const tran_low_t *input, uint8_t *output, int pitch, const struct txfm_param *param";
 
-    if (aom_config("CONFIG_EXT_TX") ne "yes") {
+  if (aom_config("CONFIG_EXT_TX") ne "yes") {
+    if (aom_config("CONFIG_DAALA_DCT4") ne "yes") {
       specialize qw/av1_iht4x4_16_add msa/;
+    }
+    if (aom_config("CONFIG_DAALA_DCT8") ne "yes") {
       specialize qw/av1_iht8x8_64_add msa/;
+    }
+    if (aom_config("CONFIG_DAALA_DCT16") ne "yes") {
       specialize qw/av1_iht16x16_256_add msa/;
     }
   }
@@ -153,6 +160,8 @@ add_proto qw/void av1_iht32x32_1024_add/, "const tran_low_t *input, uint8_t *out
 
 if (aom_config("CONFIG_TX64X64") eq "yes") {
   add_proto qw/void av1_iht64x64_4096_add/, "const tran_low_t *input, uint8_t *output, int pitch, const struct txfm_param *param";
+  add_proto qw/void av1_iht32x64_2048_add/, "const tran_low_t *input, uint8_t *output, int pitch, const struct txfm_param *param";
+  add_proto qw/void av1_iht64x32_2048_add/, "const tran_low_t *input, uint8_t *output, int pitch, const struct txfm_param *param";
 }
 
 if (aom_config("CONFIG_NEW_QUANT") eq "yes") {
@@ -256,63 +265,41 @@ if (aom_config("CONFIG_HIGHBITDEPTH") eq "yes") {
 }
 
 #inv txfm
-add_proto qw/void av1_inv_txfm2d_add_4x8/, "const int32_t *input, uint16_t *output, int stride, int tx_type, int bd";
-add_proto qw/void av1_inv_txfm2d_add_8x4/, "const int32_t *input, uint16_t *output, int stride, int tx_type, int bd";
-add_proto qw/void av1_inv_txfm2d_add_8x16/, "const int32_t *input, uint16_t *output, int stride, int tx_type, int bd";
-add_proto qw/void av1_inv_txfm2d_add_16x8/, "const int32_t *input, uint16_t *output, int stride, int tx_type, int bd";
-add_proto qw/void av1_inv_txfm2d_add_16x32/, "const int32_t *input, uint16_t *output, int stride, int tx_type, int bd";
-add_proto qw/void av1_inv_txfm2d_add_32x16/, "const int32_t *input, uint16_t *output, int stride, int tx_type, int bd";
-add_proto qw/void av1_inv_txfm2d_add_4x4/, "const int32_t *input, uint16_t *output, int stride, int tx_type, int bd";
-specialize qw/av1_inv_txfm2d_add_4x4 sse4_1/;
-add_proto qw/void av1_inv_txfm2d_add_8x8/, "const int32_t *input, uint16_t *output, int stride, int tx_type, int bd";
-specialize qw/av1_inv_txfm2d_add_8x8 sse4_1/;
-add_proto qw/void av1_inv_txfm2d_add_16x16/, "const int32_t *input, uint16_t *output, int stride, int tx_type, int bd";
-specialize qw/av1_inv_txfm2d_add_16x16 sse4_1/;
-add_proto qw/void av1_inv_txfm2d_add_32x32/, "const int32_t *input, uint16_t *output, int stride, int tx_type, int bd";
-specialize qw/av1_inv_txfm2d_add_32x32 avx2/;
-add_proto qw/void av1_inv_txfm2d_add_64x64/, "const int32_t *input, uint16_t *output, int stride, int tx_type, int bd";
+add_proto qw/void av1_inv_txfm2d_add_4x8/, "const int32_t *input, uint16_t *output, int stride, TX_TYPE tx_type, int bd";
+add_proto qw/void av1_inv_txfm2d_add_8x4/, "const int32_t *input, uint16_t *output, int stride, TX_TYPE tx_type, int bd";
+add_proto qw/void av1_inv_txfm2d_add_8x16/, "const int32_t *input, uint16_t *output, int stride, TX_TYPE tx_type, int bd";
+add_proto qw/void av1_inv_txfm2d_add_16x8/, "const int32_t *input, uint16_t *output, int stride, TX_TYPE tx_type, int bd";
+add_proto qw/void av1_inv_txfm2d_add_16x32/, "const int32_t *input, uint16_t *output, int stride, TX_TYPE tx_type, int bd";
+add_proto qw/void av1_inv_txfm2d_add_32x16/, "const int32_t *input, uint16_t *output, int stride, TX_TYPE tx_type, int bd";
+add_proto qw/void av1_inv_txfm2d_add_4x4/, "const int32_t *input, uint16_t *output, int stride, TX_TYPE tx_type, int bd";
+if (aom_config("CONFIG_DAALA_DCT4") ne "yes") {
+  specialize qw/av1_inv_txfm2d_add_4x4 sse4_1/;
+}
+add_proto qw/void av1_inv_txfm2d_add_8x8/, "const int32_t *input, uint16_t *output, int stride, TX_TYPE tx_type, int bd";
+if (aom_config("CONFIG_DAALA_DCT8") ne "yes") {
+  specialize qw/av1_inv_txfm2d_add_8x8 sse4_1/;
+}
+add_proto qw/void av1_inv_txfm2d_add_16x16/, "const int32_t *input, uint16_t *output, int stride, TX_TYPE tx_type, int bd";
+if (aom_config("CONFIG_DAALA_DCT16") ne "yes") {
+  specialize qw/av1_inv_txfm2d_add_16x16 sse4_1/;
+}
+add_proto qw/void av1_inv_txfm2d_add_32x32/, "const int32_t *input, uint16_t *output, int stride, TX_TYPE tx_type, int bd";
+if (aom_config("CONFIG_DAALA_DCT32") ne "yes") {
+  specialize qw/av1_inv_txfm2d_add_32x32 avx2/;
+}
+if (aom_config("CONFIG_TX64X64") eq "yes") {
+  add_proto qw/void av1_inv_txfm2d_add_64x64/, "const int32_t *input, uint16_t *output, int stride, TX_TYPE tx_type, int bd";
+  add_proto qw/void av1_inv_txfm2d_add_64x32/, "const int32_t *input, uint16_t *output, int stride, TX_TYPE tx_type, int bd";
+  add_proto qw/void av1_inv_txfm2d_add_32x64/, "const int32_t *input, uint16_t *output, int stride, TX_TYPE tx_type, int bd";
+}
 
 #
 # Encoder functions below this point.
 #
 if (aom_config("CONFIG_AV1_ENCODER") eq "yes") {
 
-# ENCODEMB INVOKE
-
-if (aom_config("CONFIG_AOM_QM") eq "yes") {
-  if (aom_config("CONFIG_HIGHBITDEPTH") eq "yes") {
-    # the transform coefficients are held in 32-bit
-    # values, so the assembler code for  av1_block_error can no longer be used.
-    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";
-    specialize qw/av1_block_error avx2/;
-
-    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, const qm_val_t * qm_ptr, const qm_val_t *iqm_ptr";
-
-    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, const qm_val_t * qm_ptr, const qm_val_t *iqm_ptr";
-
-    if (aom_config("CONFIG_TX64X64") eq "yes") {
-      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, const qm_val_t * qm_ptr, const qm_val_t *iqm_ptr";
-    }
-
-    add_proto qw/void av1_fdct8x8_quant/, "const int16_t *input, int stride, 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";
-  } else {
-    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";
-    specialize qw/av1_block_error avx2 msa/, "$sse2_x86inc";
-
-    add_proto qw/int64_t av1_block_error_fp/, "const int16_t *coeff, const int16_t *dqcoeff, int block_size";
-    specialize qw/av1_block_error_fp neon/, "$sse2_x86inc";
-
-    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, const qm_val_t * qm_ptr, const qm_val_t *iqm_ptr";
-
-    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, const qm_val_t * qm_ptr, const qm_val_t *iqm_ptr";
-
-    if (aom_config("CONFIG_TX64X64") eq "yes") {
-      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, const qm_val_t * qm_ptr, const qm_val_t *iqm_ptr";
-    }
+  # ENCODEMB INVOKE
 
-    add_proto qw/void av1_fdct8x8_quant/, "const int16_t *input, int stride, 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";
-  }
-} else {
   if (aom_config("CONFIG_HIGHBITDEPTH") eq "yes") {
     # the transform coefficients are held in 32-bit
     # values, so the assembler code for  av1_block_error can no longer be used.
@@ -328,8 +315,6 @@ if (aom_config("CONFIG_AOM_QM") eq "yes") {
     if (aom_config("CONFIG_TX64X64") eq "yes") {
       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";
     }
-
-    add_proto qw/void av1_fdct8x8_quant/, "const int16_t *input, int stride, 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";
   } else {
     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";
     specialize qw/av1_block_error sse2 avx2 msa/;
@@ -347,249 +332,257 @@ if (aom_config("CONFIG_AOM_QM") eq "yes") {
       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";
     }
 
-    add_proto qw/void av1_fdct8x8_quant/, "const int16_t *input, int stride, 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/av1_fdct8x8_quant sse2 ssse3 neon/;
   }
 
-}
-
-# fdct functions
+  # fdct functions
 
-add_proto qw/void av1_fht4x4/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
-specialize qw/av1_fht4x4 sse2/;
+  add_proto qw/void av1_fht4x4/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
+  if (aom_config("CONFIG_DAALA_DCT4") ne "yes") {
+    specialize qw/av1_fht4x4 sse2/;
+  }
 
-add_proto qw/void av1_fwht4x4/, "const int16_t *input, tran_low_t *output, int stride";
+  add_proto qw/void av1_fwht4x4/, "const int16_t *input, tran_low_t *output, int stride";
 
-add_proto qw/void av1_fht8x8/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
-specialize qw/av1_fht8x8 sse2/;
+  add_proto qw/void av1_fht8x8/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
+  if (aom_config("CONFIG_DAALA_DCT8") ne "yes") {
+    specialize qw/av1_fht8x8 sse2/;
+  }
 
-add_proto qw/void av1_fht16x16/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
-specialize qw/av1_fht16x16 sse2 avx2/;
+  add_proto qw/void av1_fht16x16/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
+  if (aom_config("CONFIG_DAALA_DCT16") ne "yes") {
+    specialize qw/av1_fht16x16 sse2 avx2/;
+  }
 
-add_proto qw/void av1_fht32x32/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
-specialize qw/av1_fht32x32 sse2 avx2/;
+  add_proto qw/void av1_fht32x32/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
+  if (aom_config("CONFIG_DAALA_DCT32") ne "yes") {
+    specialize qw/av1_fht32x32 sse2 avx2/;
+  }
 
-if (aom_config("CONFIG_TX64X64") eq "yes") {
-  add_proto qw/void av1_fht64x64/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
-}
+  if (aom_config("CONFIG_TX64X64") eq "yes") {
+    add_proto qw/void av1_fht64x64/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
+    add_proto qw/void av1_fht64x32/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
+    add_proto qw/void av1_fht32x64/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
+  }
 
-add_proto qw/void av1_fht4x8/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
-specialize qw/av1_fht4x8 sse2/;
+  add_proto qw/void av1_fht4x8/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
+  specialize qw/av1_fht4x8 sse2/;
 
-add_proto qw/void av1_fht8x4/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
-specialize qw/av1_fht8x4 sse2/;
+  add_proto qw/void av1_fht8x4/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
+  specialize qw/av1_fht8x4 sse2/;
 
-add_proto qw/void av1_fht8x16/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
-specialize qw/av1_fht8x16 sse2/;
+  add_proto qw/void av1_fht8x16/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
+  specialize qw/av1_fht8x16 sse2/;
 
-add_proto qw/void av1_fht16x8/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
-specialize qw/av1_fht16x8 sse2/;
+  add_proto qw/void av1_fht16x8/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
+  specialize qw/av1_fht16x8 sse2/;
 
-add_proto qw/void av1_fht16x32/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
-specialize qw/av1_fht16x32 sse2/;
+  add_proto qw/void av1_fht16x32/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
+  specialize qw/av1_fht16x32 sse2/;
 
-add_proto qw/void av1_fht32x16/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
-specialize qw/av1_fht32x16 sse2/;
+  add_proto qw/void av1_fht32x16/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
+  specialize qw/av1_fht32x16 sse2/;
 
-add_proto qw/void av1_fht4x16/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
+  add_proto qw/void av1_fht4x16/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
 
-add_proto qw/void av1_fht16x4/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
+  add_proto qw/void av1_fht16x4/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
 
-add_proto qw/void av1_fht8x32/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
+  add_proto qw/void av1_fht8x32/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
 
-add_proto qw/void av1_fht32x8/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
+  add_proto qw/void av1_fht32x8/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
 
-if (aom_config("CONFIG_HIGHBITDEPTH") ne "yes") {
-  if (aom_config("CONFIG_EXT_TX") ne "yes") {
-    specialize qw/av1_fht4x4 msa/;
-    specialize qw/av1_fht8x8 msa/;
-    specialize qw/av1_fht16x16 msa/;
+  if (aom_config("CONFIG_HIGHBITDEPTH") ne "yes") {
+    if (aom_config("CONFIG_EXT_TX") ne "yes") {
+      if (aom_config("CONFIG_DAALA_DCT4") ne "yes") {
+        specialize qw/av1_fht4x4 msa/;
+      }
+      if (aom_config("CONFIG_DAALA_DCT8") ne "yes") {
+        specialize qw/av1_fht8x8 msa/;
+      }
+      if (aom_config("CONFIG_DAALA_DCT16") ne "yes") {
+        specialize qw/av1_fht16x16 msa/;
+      }
+    }
   }
-}
 
-add_proto qw/void av1_fwd_idtx/, "const int16_t *src_diff, tran_low_t *coeff, int stride, int bs, int tx_type";
-
-if (aom_config("CONFIG_DPCM_INTRA") eq "yes") {
-  @sizes = (4, 8, 16, 32);
-  foreach $size (@sizes) {
-    if (aom_config("CONFIG_HIGHBITDEPTH") eq "yes") {
-      add_proto "void", "av1_hbd_dpcm_ft$size", "const int16_t *input, int stride, TX_TYPE_1D tx_type, tran_low_t *output, int dir";
-    }
-    add_proto "void", "av1_dpcm_ft$size", "const int16_t *input, int stride, TX_TYPE_1D tx_type, tran_low_t *output";
+  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";
+
+  #fwd txfm
+  add_proto qw/void av1_fwd_txfm2d_4x8/, "const int16_t *input, int32_t *output, int stride, TX_TYPE tx_type, int bd";
+  add_proto qw/void av1_fwd_txfm2d_8x4/, "const int16_t *input, int32_t *output, int stride, TX_TYPE tx_type, int bd";
+  add_proto qw/void av1_fwd_txfm2d_8x16/, "const int16_t *input, int32_t *output, int stride, TX_TYPE tx_type, int bd";
+  add_proto qw/void av1_fwd_txfm2d_16x8/, "const int16_t *input, int32_t *output, int stride, TX_TYPE tx_type, int bd";
+  add_proto qw/void av1_fwd_txfm2d_16x32/, "const int16_t *input, int32_t *output, int stride, TX_TYPE tx_type, int bd";
+  add_proto qw/void av1_fwd_txfm2d_32x16/, "const int16_t *input, int32_t *output, int stride, TX_TYPE tx_type, int bd";
+  add_proto qw/void av1_fwd_txfm2d_4x4/, "const int16_t *input, int32_t *output, int stride, TX_TYPE tx_type, int bd";
+  if (aom_config("CONFIG_DAALA_DCT4") ne "yes") {
+    specialize qw/av1_fwd_txfm2d_4x4 sse4_1/;
+  }
+  add_proto qw/void av1_fwd_txfm2d_8x8/, "const int16_t *input, int32_t *output, int stride, TX_TYPE tx_type, int bd";
+  if (aom_config("CONFIG_DAALA_DCT8") ne "yes") {
+    specialize qw/av1_fwd_txfm2d_8x8 sse4_1/;
+  }
+  add_proto qw/void av1_fwd_txfm2d_16x16/, "const int16_t *input, int32_t *output, int stride, TX_TYPE tx_type, int bd";
+  if (aom_config("CONFIG_DAALA_DCT16") ne "yes") {
+    specialize qw/av1_fwd_txfm2d_16x16 sse4_1/;
+  }
+  add_proto qw/void av1_fwd_txfm2d_32x32/, "const int16_t *input, int32_t *output, int stride, TX_TYPE tx_type, int bd";
+  if (aom_config("CONFIG_DAALA_DCT32") ne "yes") {
+    specialize qw/av1_fwd_txfm2d_32x32 sse4_1/;
   }
-}
 
-#fwd txfm
-add_proto qw/void av1_fwd_txfm2d_4x8/, "const int16_t *input, int32_t *output, int stride, int tx_type, int bd";
-add_proto qw/void av1_fwd_txfm2d_8x4/, "const int16_t *input, int32_t *output, int stride, int tx_type, int bd";
-add_proto qw/void av1_fwd_txfm2d_8x16/, "const int16_t *input, int32_t *output, int stride, int tx_type, int bd";
-add_proto qw/void av1_fwd_txfm2d_16x8/, "const int16_t *input, int32_t *output, int stride, int tx_type, int bd";
-add_proto qw/void av1_fwd_txfm2d_16x32/, "const int16_t *input, int32_t *output, int stride, int tx_type, int bd";
-add_proto qw/void av1_fwd_txfm2d_32x16/, "const int16_t *input, int32_t *output, int stride, int tx_type, int bd";
-add_proto qw/void av1_fwd_txfm2d_4x4/, "const int16_t *input, int32_t *output, int stride, int tx_type, int bd";
-specialize qw/av1_fwd_txfm2d_4x4 sse4_1/;
-add_proto qw/void av1_fwd_txfm2d_8x8/, "const int16_t *input, int32_t *output, int stride, int tx_type, int bd";
-specialize qw/av1_fwd_txfm2d_8x8 sse4_1/;
-add_proto qw/void av1_fwd_txfm2d_16x16/, "const int16_t *input, int32_t *output, int stride, int tx_type, int bd";
-specialize qw/av1_fwd_txfm2d_16x16 sse4_1/;
-add_proto qw/void av1_fwd_txfm2d_32x32/, "const int16_t *input, int32_t *output, int stride, int tx_type, int bd";
-specialize qw/av1_fwd_txfm2d_32x32 sse4_1/;
-add_proto qw/void av1_fwd_txfm2d_64x64/, "const int16_t *input, int32_t *output, int stride, int tx_type, int bd";
-specialize qw/av1_fwd_txfm2d_64x64 sse4_1/;
+  if (aom_config("CONFIG_TX64X64") eq "yes") {
+    add_proto qw/void av1_fwd_txfm2d_32x64/, "const int16_t *input, int32_t *output, int stride, TX_TYPE tx_type, int bd";
+    add_proto qw/void av1_fwd_txfm2d_64x32/, "const int16_t *input, int32_t *output, int stride, TX_TYPE tx_type, int bd";
+    add_proto qw/void av1_fwd_txfm2d_64x64/, "const int16_t *input, int32_t *output, int stride, TX_TYPE tx_type, int bd";
+  }
+  #
+  # Motion search
+  #
+  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";
+  specialize qw/av1_full_search_sad sse3 sse4_1/;
+  $av1_full_search_sad_sse3=av1_full_search_sadx3;
+  $av1_full_search_sad_sse4_1=av1_full_search_sadx8;
 
-#
-# Motion search
-#
-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";
-specialize qw/av1_full_search_sad sse3 sse4_1/;
-$av1_full_search_sad_sse3=av1_full_search_sadx3;
-$av1_full_search_sad_sse4_1=av1_full_search_sadx8;
+  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";
 
-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";
+  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";
 
-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";
+  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";
+  specialize qw/av1_temporal_filter_apply sse2 msa/;
 
-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";
-specialize qw/av1_temporal_filter_apply sse2 msa/;
+  if (aom_config("CONFIG_AOM_QM") eq "yes") {
+    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";
+  } else {
+    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";
+  }
 
-if (aom_config("CONFIG_AOM_QM") eq "yes") {
-  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";
-} else {
-  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";
-}
+  if (aom_config("CONFIG_LGT_FROM_PRED") eq "yes") {
+    add_proto qw/void flgt2d_from_pred/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
+  }
 
-if (aom_config("CONFIG_HIGHBITDEPTH") eq "yes") {
+  if (aom_config("CONFIG_HIGHBITDEPTH") eq "yes") {
 
-  # ENCODEMB INVOKE
-  if (aom_config("CONFIG_NEW_QUANT") eq "yes") {
-    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, const uint8_t *band";
+    # ENCODEMB INVOKE
+    if (aom_config("CONFIG_NEW_QUANT") eq "yes") {
+      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, const uint8_t *band";
 
-    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, const uint8_t *band";
+      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, const uint8_t *band";
 
-    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, const uint8_t *band";
+      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, const uint8_t *band";
 
-    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, const uint8_t *band";
+      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, const uint8_t *band";
 
-    if (aom_config("CONFIG_TX64X64") eq "yes") {
-      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, const uint8_t *band";
+      if (aom_config("CONFIG_TX64X64") eq "yes") {
+        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, const uint8_t *band";
 
-      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, const uint8_t *band";
+        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, const uint8_t *band";
+      }
     }
-  }
-
-  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";
-  specialize qw/av1_highbd_block_error sse2/;
 
-  # fdct functions
-  if (aom_config("CONFIG_TX64X64") eq "yes") {
-    add_proto qw/void av1_highbd_fht64x64/, "const int16_t *input, tran_low_t *output, int stride, int tx_type";
-  }
+    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";
+    specialize qw/av1_highbd_block_error sse2/;
 
-  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";
+    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";
 
-}
-
-if (aom_config("CONFIG_AOM_QM") eq "yes") {
-  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, const qm_val_t * qm_ptr, const qm_val_t * iqm_ptr, int log_scale";
-
-  add_proto qw/void av1_highbd_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, const qm_val_t * qm_ptr, const qm_val_t * iqm_ptr, int log_scale";
-
-  if (aom_config("CONFIG_TX64X64") eq "yes") {
-    add_proto qw/void av1_highbd_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, const qm_val_t * qm_ptr, const qm_val_t * iqm_ptr, int log_scale";
   }
 
-  add_proto qw/void av1_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, const qm_val_t * qm_ptr, const qm_val_t * iqm_ptr, int log_scale";
-} else {
   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";
   specialize qw/av1_highbd_quantize_fp sse4_1 avx2/;
 
-  add_proto qw/void av1_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, int log_scale";
-}
-
-add_proto qw/void av1_highbd_fwht4x4/, "const int16_t *input, tran_low_t *output, int stride";
+  add_proto qw/void av1_highbd_fwht4x4/, "const int16_t *input, tran_low_t *output, int stride";
 
-# End av1_high encoder functions
+  # End av1_high encoder functions
 
-if (aom_config("CONFIG_EXT_INTER") eq "yes") {
   add_proto qw/uint64_t av1_wedge_sse_from_residuals/, "const int16_t *r1, const int16_t *d, const uint8_t *m, int N";
   specialize qw/av1_wedge_sse_from_residuals sse2/;
   add_proto qw/int av1_wedge_sign_from_residuals/, "const int16_t *ds, const uint8_t *m, int N, int64_t limit";
   specialize qw/av1_wedge_sign_from_residuals sse2/;
   add_proto qw/void av1_wedge_compute_delta_squares/, "int16_t *d, const int16_t *a, const int16_t *b, int N";
   specialize qw/av1_wedge_compute_delta_squares sse2/;
-}
 
 }
 # end encoder functions
 
 # If PVQ is enabled, fwd transforms are required by decoder
 if (aom_config("CONFIG_PVQ") eq "yes") {
-# fdct functions
+  # fdct functions
 
-if (aom_config("CONFIG_HIGHBITDEPTH") eq "yes") {
-  add_proto qw/void av1_fht4x4/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
-  specialize qw/av1_fht4x4 sse2/;
+  if (aom_config("CONFIG_HIGHBITDEPTH") eq "yes") {
+    add_proto qw/void av1_fht4x4/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
+    specialize qw/av1_fht4x4 sse2/;
 
-  add_proto qw/void av1_fht8x8/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
-  specialize qw/av1_fht8x8 sse2/;
+    add_proto qw/void av1_fht8x8/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
+    specialize qw/av1_fht8x8 sse2/;
 
-  add_proto qw/void av1_fht16x16/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
-  specialize qw/av1_fht16x16 sse2/;
+    add_proto qw/void av1_fht16x16/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
+    specialize qw/av1_fht16x16 sse2/;
 
-  add_proto qw/void av1_fwht4x4/, "const int16_t *input, tran_low_t *output, int stride";
-  specialize qw/av1_fwht4x4 sse2/;
-} else {
-  add_proto qw/void av1_fht4x4/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
-  specialize qw/av1_fht4x4 sse2 msa/;
+    add_proto qw/void av1_fwht4x4/, "const int16_t *input, tran_low_t *output, int stride";
+    specialize qw/av1_fwht4x4 sse2/;
+  } else {
+    add_proto qw/void av1_fht4x4/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
+    specialize qw/av1_fht4x4 sse2 msa/;
 
-  add_proto qw/void av1_fht8x8/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
-  specialize qw/av1_fht8x8 sse2 msa/;
+    add_proto qw/void av1_fht8x8/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
+    specialize qw/av1_fht8x8 sse2 msa/;
 
-  add_proto qw/void av1_fht16x16/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
-  specialize qw/av1_fht16x16 sse2 msa/;
+    add_proto qw/void av1_fht16x16/, "const int16_t *input, tran_low_t *output, int stride, struct txfm_param *param";
+    specialize qw/av1_fht16x16 sse2 msa/;
 
-  add_proto qw/void av1_fwht4x4/, "const int16_t *input, tran_low_t *output, int stride";
-  specialize qw/av1_fwht4x4 msa sse2/;
-}
+    add_proto qw/void av1_fwht4x4/, "const int16_t *input, tran_low_t *output, int stride";
+    specialize qw/av1_fwht4x4 msa sse2/;
+  }
 
 }
 
 # Deringing Functions
 
 if (aom_config("CONFIG_CDEF") eq "yes") {
-  add_proto qw/void aom_clpf_block_hbd/, "uint16_t *dst, const uint16_t *src, int dstride, int sstride, int sizex, int sizey, unsigned int strength, unsigned int bd";
-  add_proto qw/void aom_clpf_hblock_hbd/, "uint16_t *dst, const uint16_t *src, int dstride, int sstride, int sizex, int sizey, unsigned int strength, unsigned int bd";
-  add_proto qw/void aom_clpf_block/, "uint8_t *dst, const uint16_t *src, int dstride, int sstride, int sizex, int sizey, unsigned int strength, unsigned int bd";
-  add_proto qw/void aom_clpf_hblock/, "uint8_t *dst, const uint16_t *src, int dstride, int sstride, int sizex, int sizey, unsigned int strength, unsigned int bd";
-  add_proto qw/int od_dir_find8/, "const od_dering_in *img, int stride, int32_t *var, int coeff_shift";
-  add_proto qw/void od_filter_dering_direction_4x4/, "uint16_t *y, int ystride, const uint16_t *in, int threshold, int dir, int damping";
-  add_proto qw/void od_filter_dering_direction_8x8/, "uint16_t *y, int ystride, const uint16_t *in, int threshold, int dir, int damping";
-
-  add_proto qw/void copy_8x8_16bit_to_8bit/, "uint8_t *dst, int dstride, const uint16_t *src, int sstride";
-  add_proto qw/void copy_4x4_16bit_to_8bit/, "uint8_t *dst, int dstride, const uint16_t *src, int sstride";
-  add_proto qw/void copy_8x8_16bit_to_16bit/, "uint16_t *dst, int dstride, const uint16_t *src, int sstride";
-  add_proto qw/void copy_4x4_16bit_to_16bit/, "uint16_t *dst, int dstride, const uint16_t *src, int sstride";
+  add_proto qw/int cdef_find_dir/, "const uint16_t *img, int stride, int32_t *var, int coeff_shift";
+  if (aom_config("CONFIG_CDEF_SINGLEPASS") ne "yes") {
+    add_proto qw/void aom_clpf_block_hbd/, "uint16_t *dst, const uint16_t *src, int dstride, int sstride, int sizex, int sizey, unsigned int strength, unsigned int bd";
+    add_proto qw/void aom_clpf_hblock_hbd/, "uint16_t *dst, const uint16_t *src, int dstride, int sstride, int sizex, int sizey, unsigned int strength, unsigned int bd";
+    add_proto qw/void aom_clpf_block/, "uint8_t *dst, const uint16_t *src, int dstride, int sstride, int sizex, int sizey, unsigned int strength, unsigned int bd";
+    add_proto qw/void aom_clpf_hblock/, "uint8_t *dst, const uint16_t *src, int dstride, int sstride, int sizex, int sizey, unsigned int strength, unsigned int bd";
+    add_proto qw/void cdef_direction_4x4/, "uint16_t *y, int ystride, const uint16_t *in, int threshold, int dir, int damping";
+    add_proto qw/void cdef_direction_8x8/, "uint16_t *y, int ystride, const uint16_t *in, int threshold, int dir, int damping";
+    add_proto qw/void copy_8x8_16bit_to_8bit/, "uint8_t *dst, int dstride, const uint16_t *src, int sstride";
+    add_proto qw/void copy_4x4_16bit_to_8bit/, "uint8_t *dst, int dstride, const uint16_t *src, int sstride";
+    add_proto qw/void copy_8x8_16bit_to_16bit/, "uint16_t *dst, int dstride, const uint16_t *src, int sstride";
+    add_proto qw/void copy_4x4_16bit_to_16bit/, "uint16_t *dst, int dstride, const uint16_t *src, int sstride";
+  } else {
+    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";
+  }
+
   add_proto qw/void copy_rect8_8bit_to_16bit/, "uint16_t *dst, int dstride, const uint8_t *src, int sstride, int v, int h";
   add_proto qw/void copy_rect8_16bit_to_16bit/, "uint16_t *dst, int dstride, const uint16_t *src, int sstride, int v, int h";
 
-# VS compiling for 32 bit targets does not support vector types in
+  # VS compiling for 32 bit targets does not support vector types in
   # structs as arguments, which makes the v256 type of the intrinsics
   # hard to support, so optimizations for this target are disabled.
   if ($opts{config} !~ /libs-x86-win32-vs.*/) {
-    specialize qw/aom_clpf_block_hbd sse2 ssse3 sse4_1 neon/;
-    specialize qw/aom_clpf_hblock_hbd sse2 ssse3 sse4_1 neon/;
-    specialize qw/aom_clpf_block sse2 ssse3 sse4_1 neon/;
-    specialize qw/aom_clpf_hblock sse2 ssse3 sse4_1 neon/;
-    specialize qw/od_dir_find8 sse2 ssse3 sse4_1 neon/;
-    specialize qw/od_filter_dering_direction_4x4 sse2 ssse3 sse4_1 neon/;
-    specialize qw/od_filter_dering_direction_8x8 sse2 ssse3 sse4_1 neon/;
-
-    specialize qw/copy_8x8_16bit_to_8bit sse2 ssse3 sse4_1 neon/;
-    specialize qw/copy_4x4_16bit_to_8bit sse2 ssse3 sse4_1 neon/;
-    specialize qw/copy_8x8_16bit_to_16bit sse2 ssse3 sse4_1 neon/;
-    specialize qw/copy_4x4_16bit_to_16bit sse2 ssse3 sse4_1 neon/;
-    specialize qw/copy_rect8_8bit_to_16bit sse2 ssse3 sse4_1 neon/;
-    specialize qw/copy_rect8_16bit_to_16bit sse2 ssse3 sse4_1 neon/;
+    if (aom_config("CONFIG_CDEF_SINGLEPASS") eq "yes") {
+      specialize qw/cdef_find_dir sse2 ssse3 sse4_1 avx2 neon/;
+      specialize qw/cdef_filter_block sse2 ssse3 sse4_1 avx2 neon/;
+      specialize qw/copy_rect8_8bit_to_16bit sse2 ssse3 sse4_1 avx2 neon/;
+      specialize qw/copy_rect8_16bit_to_16bit sse2 ssse3 sse4_1 avx2 neon/;
+    } else {
+      specialize qw/cdef_find_dir sse2 ssse3 sse4_1 neon/;
+      specialize qw/aom_clpf_block_hbd sse2 ssse3 sse4_1 neon/;
+      specialize qw/aom_clpf_hblock_hbd sse2 ssse3 sse4_1 neon/;
+      specialize qw/aom_clpf_block sse2 ssse3 sse4_1 neon/;
+      specialize qw/aom_clpf_hblock sse2 ssse3 sse4_1 neon/;
+      specialize qw/cdef_find_dir sse2 ssse3 sse4_1 neon/;
+      specialize qw/cdef_direction_4x4 sse2 ssse3 sse4_1 neon/;
+      specialize qw/cdef_direction_8x8 sse2 ssse3 sse4_1 neon/;
+
+      specialize qw/copy_8x8_16bit_to_8bit sse2 ssse3 sse4_1 neon/;
+      specialize qw/copy_4x4_16bit_to_8bit sse2 ssse3 sse4_1 neon/;
+      specialize qw/copy_8x8_16bit_to_16bit sse2 ssse3 sse4_1 neon/;
+      specialize qw/copy_4x4_16bit_to_16bit sse2 ssse3 sse4_1 neon/;
+      specialize qw/copy_rect8_8bit_to_16bit sse2 ssse3 sse4_1 neon/;
+      specialize qw/copy_rect8_16bit_to_16bit sse2 ssse3 sse4_1 neon/;
+    }
   }
 }
 
@@ -607,16 +600,9 @@ if ((aom_config("CONFIG_WARPED_MOTION") eq "yes") ||
   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";
   specialize qw/av1_warp_affine sse2 ssse3/;
 
-  if (aom_config("CONFIG_CONVOLVE_ROUND") eq "yes") {
-    add_proto qw/void av1_warp_affine_post_round/, "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";
-  }
-
   if (aom_config("CONFIG_HIGHBITDEPTH") eq "yes") {
     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";
     specialize qw/av1_highbd_warp_affine ssse3/;
-    if (aom_config("CONFIG_CONVOLVE_ROUND") eq "yes") {
-      add_proto qw/void av1_highbd_warp_affine_post_round/, "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";
-    }
   }
 }
 
@@ -632,7 +618,7 @@ if (aom_config("CONFIG_LOOP_RESTORATION") eq "yes") {
   add_proto qw/void apply_selfguided_restoration/, "uint8_t *dat, int width, int height, int stride, int eps, int *xqd, uint8_t *dst, int dst_stride, int32_t *tmpbuf";
   specialize qw/apply_selfguided_restoration sse4_1/;
 
-  add_proto qw/void av1_selfguided_restoration/, "uint8_t *dgd, int width, int height, int stride, int32_t *dst, int dst_stride, int r, int eps, int32_t *tmpbuf";
+  add_proto qw/void av1_selfguided_restoration/, "uint8_t *dgd, int width, int height, int stride, int32_t *dst, int dst_stride, int r, int eps";
   specialize qw/av1_selfguided_restoration sse4_1/;
 
   add_proto qw/void av1_highpass_filter/, "uint8_t *dgd, int width, int height, int stride, int32_t *dst, int dst_stride, int r, int eps";
@@ -642,7 +628,7 @@ if (aom_config("CONFIG_LOOP_RESTORATION") eq "yes") {
     add_proto qw/void apply_selfguided_restoration_highbd/, "uint16_t *dat, int width, int height, int stride, int bit_depth, int eps, int *xqd, uint16_t *dst, int dst_stride, int32_t *tmpbuf";
     specialize qw/apply_selfguided_restoration_highbd sse4_1/;
 
-    add_proto qw/void av1_selfguided_restoration_highbd/, "uint16_t *dgd, int width, int height, int stride, int32_t *dst, int dst_stride, int bit_depth, int r, int eps, int32_t *tmpbuf";
+    add_proto qw/void av1_selfguided_restoration_highbd/, "uint16_t *dgd, int width, int height, int stride, int32_t *dst, int dst_stride, int bit_depth, int r, int eps";
     specialize qw/av1_selfguided_restoration_highbd sse4_1/;
 
     add_proto qw/void av1_highpass_filter_highbd/, "uint16_t *dgd, int width, int height, int stride, int32_t *dst, int dst_stride, int r, int eps";
@@ -653,17 +639,40 @@ if (aom_config("CONFIG_LOOP_RESTORATION") eq "yes") {
 # CONVOLVE_ROUND/COMPOUND_ROUND functions
 
 if (aom_config("CONFIG_CONVOLVE_ROUND") eq "yes") {
-    add_proto qw/void av1_convolve_2d/, "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_q4, const int subpel_y_q4, ConvolveParams *conv_params";
-    specialize qw/av1_convolve_2d sse2/;
-    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";
-    specialize qw/av1_convolve_rounding avx2/;
+  add_proto qw/void av1_convolve_2d/, "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_q4, const int subpel_y_q4, ConvolveParams *conv_params";
+  specialize qw/av1_convolve_2d sse2/;
+  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";
+  specialize qw/av1_convolve_rounding avx2/;
+
+  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";
+  if (aom_config("CONFIG_COMPOUND_ROUND") ne "yes") {
+    specialize qw/av1_convolve_2d_scale sse4_1/;
+  }
 
   if (aom_config("CONFIG_HIGHBITDEPTH") eq "yes") {
     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";
     specialize qw/av1_highbd_convolve_2d ssse3/;
     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";
     specialize qw/av1_highbd_convolve_rounding avx2/;
+
+    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";
+    if (aom_config("CONFIG_COMPOUND_ROUND") ne "yes") {
+        specialize qw/av1_highbd_convolve_2d_scale sse4_1/;
+    }
+  }
+}
+
+# INTRA_EDGE functions
+if (aom_config("CONFIG_INTRA_EDGE") eq "yes") {
+  add_proto qw/void av1_filter_intra_edge/, "uint8_t *p, int sz, int strength";
+  specialize qw/av1_filter_intra_edge sse4_1/;
+  add_proto qw/void av1_upsample_intra_edge/, "uint8_t *p, int sz";
+  specialize qw/av1_upsample_intra_edge sse4_1/;
+  if (aom_config("CONFIG_HIGHBITDEPTH") eq "yes") {
+    add_proto qw/void av1_filter_intra_edge_high/, "uint16_t *p, int sz, int strength";
+    specialize qw/av1_filter_intra_edge_high sse4_1/;
+    add_proto qw/void av1_upsample_intra_edge_high/, "uint16_t *p, int sz, int bd";
+    specialize qw/av1_upsample_intra_edge_high sse4_1/;
   }
 }
 
-1;