diff options
Diffstat (limited to 'third_party/aom/av1/common/av1_rtcd_defs.pl')
| -rwxr-xr-x | third_party/aom/av1/common/av1_rtcd_defs.pl | 46 |
1 files changed, 33 insertions, 13 deletions
diff --git a/third_party/aom/av1/common/av1_rtcd_defs.pl b/third_party/aom/av1/common/av1_rtcd_defs.pl index fa8b34981..dee1f1c79 100755 --- a/third_party/aom/av1/common/av1_rtcd_defs.pl +++ b/third_party/aom/av1/common/av1_rtcd_defs.pl @@ -76,12 +76,12 @@ specialize qw/av1_wiener_convolve_add_src sse2 avx2 neon/; specialize qw/av1_highbd_wiener_convolve_add_src ssse3/; specialize qw/av1_highbd_wiener_convolve_add_src avx2/; + # directional intra predictor functions add_proto qw/void av1_dr_prediction_z1/, "uint8_t *dst, ptrdiff_t stride, int bw, int bh, const uint8_t *above, const uint8_t *left, int upsample_above, int dx, int dy"; add_proto qw/void av1_dr_prediction_z2/, "uint8_t *dst, ptrdiff_t stride, int bw, int bh, const uint8_t *above, const uint8_t *left, int upsample_above, int upsample_left, int dx, int dy"; add_proto qw/void av1_dr_prediction_z3/, "uint8_t *dst, ptrdiff_t stride, int bw, int bh, const uint8_t *above, const uint8_t *left, int upsample_left, int dx, int dy"; - # FILTER_INTRA predictor functions add_proto qw/void av1_filter_intra_predictor/, "uint8_t *dst, ptrdiff_t stride, TX_SIZE tx_size, const uint8_t *above, const uint8_t *left, int mode"; specialize qw/av1_filter_intra_predictor sse4_1/; @@ -108,6 +108,22 @@ specialize qw/av1_highbd_convolve8_vert/, "$sse2_x86_64"; add_proto qw/void av1_inv_txfm_add/, "const tran_low_t *dqcoeff, uint8_t *dst, int stride, const TxfmParam *txfm_param"; specialize qw/av1_inv_txfm_add ssse3 avx2 neon/; +add_proto qw/void av1_highbd_inv_txfm_add/, "const tran_low_t *dqcoeff, uint8_t *dst, int stride, const TxfmParam *txfm_param"; +specialize qw/av1_highbd_inv_txfm_add sse4_1 avx2/; + +add_proto qw/void av1_highbd_inv_txfm_add_4x4/, "const tran_low_t *dqcoeff, uint8_t *dst, int stride, const TxfmParam *txfm_param"; +specialize qw/av1_highbd_inv_txfm_add_4x4 sse4_1/; +add_proto qw/void av1_highbd_inv_txfm_add_8x8/, "const tran_low_t *dqcoeff, uint8_t *dst, int stride, const TxfmParam *txfm_param"; +specialize qw/av1_highbd_inv_txfm_add_8x8 sse4_1/; +add_proto qw/void av1_highbd_inv_txfm_add_16x8/, "const tran_low_t *dqcoeff, uint8_t *dst, int stride, const TxfmParam *txfm_param"; +specialize qw/av1_highbd_inv_txfm_add_16x8 sse4_1/; +add_proto qw/void av1_highbd_inv_txfm_add_8x16/, "const tran_low_t *dqcoeff, uint8_t *dst, int stride, const TxfmParam *txfm_param"; +specialize qw/av1_highbd_inv_txfm_add_8x16 sse4_1/; +add_proto qw/void av1_highbd_inv_txfm_add_16x16/, "const tran_low_t *dqcoeff, uint8_t *dst, int stride, const TxfmParam *txfm_param"; +specialize qw/av1_highbd_inv_txfm_add_16x16 sse4_1/; +add_proto qw/void av1_highbd_inv_txfm_add_32x32/, "const tran_low_t *dqcoeff, uint8_t *dst, int stride, const TxfmParam *txfm_param"; +specialize qw/av1_highbd_inv_txfm_add_32x32 sse4_1 avx2/; + add_proto qw/void av1_highbd_iwht4x4_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int bd"; add_proto qw/void av1_highbd_iwht4x4_16_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int bd"; @@ -122,9 +138,7 @@ 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"; 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"; -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"; -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, 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"; @@ -132,8 +146,6 @@ add_proto qw/void av1_inv_txfm2d_add_64x32/, "const int32_t *input, uint16_t *ou add_proto qw/void av1_inv_txfm2d_add_16x64/, "const int32_t *input, uint16_t *output, int stride, TX_TYPE tx_type, int bd"; add_proto qw/void av1_inv_txfm2d_add_64x16/, "const int32_t *input, uint16_t *output, int stride, TX_TYPE tx_type, int bd"; -specialize qw/av1_inv_txfm2d_add_64x64 sse4_1/; - add_proto qw/void av1_inv_txfm2d_add_4x16/, "const int32_t *input, uint16_t *output, int stride, TX_TYPE tx_type, int bd"; add_proto qw/void av1_inv_txfm2d_add_16x4/, "const int32_t *input, uint16_t *output, int stride, TX_TYPE tx_type, int bd"; add_proto qw/void av1_inv_txfm2d_add_8x32/, "const int32_t *input, uint16_t *output, int stride, TX_TYPE tx_type, int bd"; @@ -146,13 +158,13 @@ add_proto qw/void av1_highbd_dr_prediction_z3/, "uint16_t *dst, ptrdiff_t stride # build compound seg mask functions add_proto qw/void av1_build_compound_diffwtd_mask/, "uint8_t *mask, DIFFWTD_MASK_TYPE mask_type, const uint8_t *src0, int src0_stride, const uint8_t *src1, int src1_stride, int h, int w"; -specialize qw/av1_build_compound_diffwtd_mask sse4_1/; +specialize qw/av1_build_compound_diffwtd_mask sse4_1 avx2/; add_proto qw/void av1_build_compound_diffwtd_mask_highbd/, "uint8_t *mask, DIFFWTD_MASK_TYPE mask_type, const uint8_t *src0, int src0_stride, const uint8_t *src1, int src1_stride, int h, int w, int bd"; specialize qw/av1_build_compound_diffwtd_mask_highbd ssse3 avx2/; add_proto qw/void av1_build_compound_diffwtd_mask_d16/, "uint8_t *mask, DIFFWTD_MASK_TYPE mask_type, const CONV_BUF_TYPE *src0, int src0_stride, const CONV_BUF_TYPE *src1, int src1_stride, int h, int w, ConvolveParams *conv_params, int bd"; -specialize qw/av1_build_compound_diffwtd_mask_d16 sse4_1 neon/; +specialize qw/av1_build_compound_diffwtd_mask_d16 sse4_1 avx2 neon/; # # Encoder functions below this point. @@ -186,7 +198,9 @@ if (aom_config("CONFIG_AV1_ENCODER") eq "yes") { 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"; + specialize qw/av1_fwd_txfm2d_8x16 sse4_1/; add_proto qw/void av1_fwd_txfm2d_16x8/, "const int16_t *input, int32_t *output, int stride, TX_TYPE tx_type, int bd"; + specialize qw/av1_fwd_txfm2d_16x8 sse4_1/; 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_4x16/, "const int16_t *input, int32_t *output, int stride, TX_TYPE tx_type, int bd"; @@ -203,6 +217,7 @@ if (aom_config("CONFIG_AV1_ENCODER") eq "yes") { 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, TX_TYPE tx_type, int bd"; + specialize qw/av1_fwd_txfm2d_64x64 sse4_1/; 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_16x64/, "const int16_t *input, int32_t *output, int stride, TX_TYPE tx_type, int bd"; @@ -218,7 +233,7 @@ if (aom_config("CONFIG_AV1_ENCODER") eq "yes") { 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_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"; + add_proto qw/void av1_quantize_b/, "const tran_low_t *coeff_ptr, intptr_t n_coeffs, 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"; # ENCODEMB INVOKE @@ -238,7 +253,7 @@ if (aom_config("CONFIG_AV1_ENCODER") eq "yes") { 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_CLASS tx_class, int8_t *const coeff_contexts"; specialize qw/av1_get_nz_map_contexts sse2/; add_proto qw/void av1_txb_init_levels/, "const tran_low_t *const coeff, const int width, const int height, uint8_t *const levels"; - specialize qw/av1_txb_init_levels sse4_1/; + specialize qw/av1_txb_init_levels sse4_1 avx2/; 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 avx2/; @@ -251,6 +266,11 @@ if (aom_config("CONFIG_AV1_ENCODER") eq "yes") { add_proto qw/uint32_t av1_get_crc32c_value/, "void *crc_calculator, uint8_t *p, int length"; specialize qw/av1_get_crc32c_value sse4_2/; + add_proto qw/void av1_compute_stats/, "int wiener_win, const uint8_t *dgd8, const uint8_t *src8, int h_start, int h_end, int v_start, int v_end, int dgd_stride, int src_stride, double *M, double *H"; + specialize qw/av1_compute_stats sse4_1 avx2/; + + add_proto qw/int64_t av1_lowbd_pixel_proj_error/, " const uint8_t *src8, int width, int height, int src_stride, const uint8_t *dat8, int dat_stride, int32_t *flt0, int flt0_stride, int32_t *flt1, int flt1_stride, int xq[2], const sgr_params_type *params"; + specialize qw/av1_lowbd_pixel_proj_error sse4_1 avx2/; } # end encoder functions @@ -275,7 +295,7 @@ if ($opts{config} !~ /libs-x86-win32-vs.*/) { # WARPED_MOTION / GLOBAL_MOTION functions 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 sse4_1/; +specialize qw/av1_warp_affine sse4_1 neon/; 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 sse4_1/; @@ -290,9 +310,9 @@ if (aom_config("CONFIG_AV1_ENCODER") eq "yes") { 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"; specialize qw/apply_selfguided_restoration sse4_1 avx2 neon/; -add_proto qw/void av1_selfguided_restoration/, "const uint8_t *dgd8, int width, int height, - int dgd_stride, int32_t *flt0, int32_t *flt1, int flt_stride, - int sgr_params_idx, int bit_depth, int highbd"; +add_proto qw/int av1_selfguided_restoration/, "const uint8_t *dgd8, int width, int height, + int dgd_stride, int32_t *flt0, int32_t *flt1, int flt_stride, + int sgr_params_idx, int bit_depth, int highbd"; specialize qw/av1_selfguided_restoration sse4_1 avx2 neon/; # CONVOLVE_ROUND/COMPOUND_ROUND functions |