1## 2## Copyright (c) 2017 The WebM project authors. All Rights Reserved. 3## 4## Use of this source code is governed by a BSD-style license 5## that can be found in the LICENSE file in the root of the source 6## tree. An additional intellectual property rights grant can be found 7## in the file PATENTS. All contributing project authors may 8## be found in the AUTHORS file in the root of the source tree. 9## 10 11sub vp8_common_forward_decls() { 12print <<EOF 13/* 14 * VP8 15 */ 16 17struct blockd; 18struct macroblockd; 19struct loop_filter_info; 20 21/* Encoder forward decls */ 22struct block; 23struct macroblock; 24struct variance_vtable; 25union int_mv; 26struct yv12_buffer_config; 27EOF 28} 29forward_decls qw/vp8_common_forward_decls/; 30 31# 32# Dequant 33# 34add_proto qw/void vp8_dequantize_b/, "struct blockd*, short *DQC"; 35specialize qw/vp8_dequantize_b mmx neon msa mmi/; 36 37add_proto qw/void vp8_dequant_idct_add/, "short *input, short *dq, unsigned char *dest, int stride"; 38specialize qw/vp8_dequant_idct_add mmx neon dspr2 msa mmi/; 39 40add_proto qw/void vp8_dequant_idct_add_y_block/, "short *q, short *dq, unsigned char *dst, int stride, char *eobs"; 41specialize qw/vp8_dequant_idct_add_y_block sse2 neon dspr2 msa mmi/; 42 43add_proto qw/void vp8_dequant_idct_add_uv_block/, "short *q, short *dq, unsigned char *dst_u, unsigned char *dst_v, int stride, char *eobs"; 44specialize qw/vp8_dequant_idct_add_uv_block sse2 neon dspr2 msa mmi/; 45 46# 47# Loopfilter 48# 49add_proto qw/void vp8_loop_filter_mbv/, "unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr, int y_stride, int uv_stride, struct loop_filter_info *lfi"; 50specialize qw/vp8_loop_filter_mbv sse2 neon dspr2 msa mmi/; 51 52add_proto qw/void vp8_loop_filter_bv/, "unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr, int y_stride, int uv_stride, struct loop_filter_info *lfi"; 53specialize qw/vp8_loop_filter_bv sse2 neon dspr2 msa mmi/; 54 55add_proto qw/void vp8_loop_filter_mbh/, "unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr, int y_stride, int uv_stride, struct loop_filter_info *lfi"; 56specialize qw/vp8_loop_filter_mbh sse2 neon dspr2 msa mmi/; 57 58add_proto qw/void vp8_loop_filter_bh/, "unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr, int y_stride, int uv_stride, struct loop_filter_info *lfi"; 59specialize qw/vp8_loop_filter_bh sse2 neon dspr2 msa mmi/; 60 61 62add_proto qw/void vp8_loop_filter_simple_mbv/, "unsigned char *y_ptr, int y_stride, const unsigned char *blimit"; 63specialize qw/vp8_loop_filter_simple_mbv sse2 neon msa mmi/; 64$vp8_loop_filter_simple_mbv_c=vp8_loop_filter_simple_vertical_edge_c; 65$vp8_loop_filter_simple_mbv_sse2=vp8_loop_filter_simple_vertical_edge_sse2; 66$vp8_loop_filter_simple_mbv_neon=vp8_loop_filter_mbvs_neon; 67$vp8_loop_filter_simple_mbv_msa=vp8_loop_filter_simple_vertical_edge_msa; 68$vp8_loop_filter_simple_mbv_mmi=vp8_loop_filter_simple_vertical_edge_mmi; 69 70add_proto qw/void vp8_loop_filter_simple_mbh/, "unsigned char *y_ptr, int y_stride, const unsigned char *blimit"; 71specialize qw/vp8_loop_filter_simple_mbh sse2 neon msa mmi/; 72$vp8_loop_filter_simple_mbh_c=vp8_loop_filter_simple_horizontal_edge_c; 73$vp8_loop_filter_simple_mbh_sse2=vp8_loop_filter_simple_horizontal_edge_sse2; 74$vp8_loop_filter_simple_mbh_neon=vp8_loop_filter_mbhs_neon; 75$vp8_loop_filter_simple_mbh_msa=vp8_loop_filter_simple_horizontal_edge_msa; 76$vp8_loop_filter_simple_mbh_mmi=vp8_loop_filter_simple_horizontal_edge_mmi; 77 78add_proto qw/void vp8_loop_filter_simple_bv/, "unsigned char *y_ptr, int y_stride, const unsigned char *blimit"; 79specialize qw/vp8_loop_filter_simple_bv sse2 neon msa mmi/; 80$vp8_loop_filter_simple_bv_c=vp8_loop_filter_bvs_c; 81$vp8_loop_filter_simple_bv_sse2=vp8_loop_filter_bvs_sse2; 82$vp8_loop_filter_simple_bv_neon=vp8_loop_filter_bvs_neon; 83$vp8_loop_filter_simple_bv_msa=vp8_loop_filter_bvs_msa; 84$vp8_loop_filter_simple_bv_mmi=vp8_loop_filter_bvs_mmi; 85 86add_proto qw/void vp8_loop_filter_simple_bh/, "unsigned char *y_ptr, int y_stride, const unsigned char *blimit"; 87specialize qw/vp8_loop_filter_simple_bh sse2 neon msa mmi/; 88$vp8_loop_filter_simple_bh_c=vp8_loop_filter_bhs_c; 89$vp8_loop_filter_simple_bh_sse2=vp8_loop_filter_bhs_sse2; 90$vp8_loop_filter_simple_bh_neon=vp8_loop_filter_bhs_neon; 91$vp8_loop_filter_simple_bh_msa=vp8_loop_filter_bhs_msa; 92$vp8_loop_filter_simple_bh_mmi=vp8_loop_filter_bhs_mmi; 93 94# 95# IDCT 96# 97#idct16 98add_proto qw/void vp8_short_idct4x4llm/, "short *input, unsigned char *pred_ptr, int pred_stride, unsigned char *dst_ptr, int dst_stride"; 99specialize qw/vp8_short_idct4x4llm mmx neon dspr2 msa mmi/; 100 101#iwalsh1 102add_proto qw/void vp8_short_inv_walsh4x4_1/, "short *input, short *mb_dqcoeff"; 103specialize qw/vp8_short_inv_walsh4x4_1 dspr2/; 104 105#iwalsh16 106add_proto qw/void vp8_short_inv_walsh4x4/, "short *input, short *mb_dqcoeff"; 107specialize qw/vp8_short_inv_walsh4x4 sse2 neon dspr2 msa mmi/; 108 109#idct1_scalar_add 110add_proto qw/void vp8_dc_only_idct_add/, "short input_dc, unsigned char *pred_ptr, int pred_stride, unsigned char *dst_ptr, int dst_stride"; 111specialize qw/vp8_dc_only_idct_add mmx neon dspr2 msa mmi/; 112 113# 114# RECON 115# 116add_proto qw/void vp8_copy_mem16x16/, "unsigned char *src, int src_stride, unsigned char *dst, int dst_stride"; 117specialize qw/vp8_copy_mem16x16 sse2 neon dspr2 msa mmi/; 118 119add_proto qw/void vp8_copy_mem8x8/, "unsigned char *src, int src_stride, unsigned char *dst, int dst_stride"; 120specialize qw/vp8_copy_mem8x8 mmx neon dspr2 msa mmi/; 121 122add_proto qw/void vp8_copy_mem8x4/, "unsigned char *src, int src_stride, unsigned char *dst, int dst_stride"; 123specialize qw/vp8_copy_mem8x4 mmx neon dspr2 msa mmi/; 124 125# 126# Postproc 127# 128if (vpx_config("CONFIG_POSTPROC") eq "yes") { 129 130 add_proto qw/void vp8_blend_mb_inner/, "unsigned char *y, unsigned char *u, unsigned char *v, int y_1, int u_1, int v_1, int alpha, int stride"; 131 132 add_proto qw/void vp8_blend_mb_outer/, "unsigned char *y, unsigned char *u, unsigned char *v, int y_1, int u_1, int v_1, int alpha, int stride"; 133 134 add_proto qw/void vp8_blend_b/, "unsigned char *y, unsigned char *u, unsigned char *v, int y_1, int u_1, int v_1, int alpha, int stride"; 135 136 add_proto qw/void vp8_filter_by_weight16x16/, "unsigned char *src, int src_stride, unsigned char *dst, int dst_stride, int src_weight"; 137 specialize qw/vp8_filter_by_weight16x16 sse2 msa/; 138 139 add_proto qw/void vp8_filter_by_weight8x8/, "unsigned char *src, int src_stride, unsigned char *dst, int dst_stride, int src_weight"; 140 specialize qw/vp8_filter_by_weight8x8 sse2 msa/; 141 142 add_proto qw/void vp8_filter_by_weight4x4/, "unsigned char *src, int src_stride, unsigned char *dst, int dst_stride, int src_weight"; 143} 144 145# 146# Subpixel 147# 148add_proto qw/void vp8_sixtap_predict16x16/, "unsigned char *src_ptr, int src_pixels_per_line, int xoffset, int yoffset, unsigned char *dst_ptr, int dst_pitch"; 149specialize qw/vp8_sixtap_predict16x16 sse2 ssse3 neon dspr2 msa mmi/; 150 151add_proto qw/void vp8_sixtap_predict8x8/, "unsigned char *src_ptr, int src_pixels_per_line, int xoffset, int yoffset, unsigned char *dst_ptr, int dst_pitch"; 152specialize qw/vp8_sixtap_predict8x8 sse2 ssse3 neon dspr2 msa mmi/; 153 154add_proto qw/void vp8_sixtap_predict8x4/, "unsigned char *src_ptr, int src_pixels_per_line, int xoffset, int yoffset, unsigned char *dst_ptr, int dst_pitch"; 155specialize qw/vp8_sixtap_predict8x4 sse2 ssse3 neon dspr2 msa mmi/; 156 157add_proto qw/void vp8_sixtap_predict4x4/, "unsigned char *src_ptr, int src_pixels_per_line, int xoffset, int yoffset, unsigned char *dst_ptr, int dst_pitch"; 158specialize qw/vp8_sixtap_predict4x4 mmx ssse3 neon dspr2 msa mmi/; 159 160add_proto qw/void vp8_bilinear_predict16x16/, "unsigned char *src_ptr, int src_pixels_per_line, int xoffset, int yoffset, unsigned char *dst_ptr, int dst_pitch"; 161specialize qw/vp8_bilinear_predict16x16 sse2 ssse3 neon msa/; 162 163add_proto qw/void vp8_bilinear_predict8x8/, "unsigned char *src_ptr, int src_pixels_per_line, int xoffset, int yoffset, unsigned char *dst_ptr, int dst_pitch"; 164specialize qw/vp8_bilinear_predict8x8 sse2 ssse3 neon msa/; 165 166add_proto qw/void vp8_bilinear_predict8x4/, "unsigned char *src_ptr, int src_pixels_per_line, int xoffset, int yoffset, unsigned char *dst_ptr, int dst_pitch"; 167specialize qw/vp8_bilinear_predict8x4 sse2 neon msa/; 168 169add_proto qw/void vp8_bilinear_predict4x4/, "unsigned char *src_ptr, int src_pixels_per_line, int xoffset, int yoffset, unsigned char *dst_ptr, int dst_pitch"; 170specialize qw/vp8_bilinear_predict4x4 sse2 neon msa/; 171 172# 173# Encoder functions below this point. 174# 175if (vpx_config("CONFIG_VP8_ENCODER") eq "yes") { 176 177# 178# Block copy 179# 180add_proto qw/void vp8_copy32xn/, "const unsigned char *src_ptr, int src_stride, unsigned char *dst_ptr, int dst_stride, int height"; 181specialize qw/vp8_copy32xn sse2 sse3/; 182 183# 184# Forward DCT 185# 186add_proto qw/void vp8_short_fdct4x4/, "short *input, short *output, int pitch"; 187specialize qw/vp8_short_fdct4x4 sse2 neon msa mmi/; 188 189add_proto qw/void vp8_short_fdct8x4/, "short *input, short *output, int pitch"; 190specialize qw/vp8_short_fdct8x4 sse2 neon msa mmi/; 191 192add_proto qw/void vp8_short_walsh4x4/, "short *input, short *output, int pitch"; 193specialize qw/vp8_short_walsh4x4 sse2 neon msa mmi/; 194 195# 196# Quantizer 197# 198add_proto qw/void vp8_regular_quantize_b/, "struct block *, struct blockd *"; 199specialize qw/vp8_regular_quantize_b sse2 sse4_1 msa mmi/; 200 201add_proto qw/void vp8_fast_quantize_b/, "struct block *, struct blockd *"; 202specialize qw/vp8_fast_quantize_b sse2 ssse3 neon msa mmi/; 203 204# 205# Block subtraction 206# 207add_proto qw/int vp8_block_error/, "short *coeff, short *dqcoeff"; 208specialize qw/vp8_block_error sse2 msa/; 209 210add_proto qw/int vp8_mbblock_error/, "struct macroblock *mb, int dc"; 211specialize qw/vp8_mbblock_error sse2 msa/; 212 213add_proto qw/int vp8_mbuverror/, "struct macroblock *mb"; 214specialize qw/vp8_mbuverror sse2 msa/; 215 216# 217# Motion search 218# 219add_proto qw/int vp8_full_search_sad/, "struct macroblock *x, struct block *b, struct blockd *d, union int_mv *ref_mv, int sad_per_bit, int distance, struct variance_vtable *fn_ptr, int *mvcost[2], union int_mv *center_mv"; 220specialize qw/vp8_full_search_sad sse3 sse4_1/; 221$vp8_full_search_sad_sse3=vp8_full_search_sadx3; 222$vp8_full_search_sad_sse4_1=vp8_full_search_sadx8; 223 224add_proto qw/int vp8_refining_search_sad/, "struct macroblock *x, struct block *b, struct blockd *d, union int_mv *ref_mv, int error_per_bit, int search_range, struct variance_vtable *fn_ptr, int *mvcost[2], union int_mv *center_mv"; 225specialize qw/vp8_refining_search_sad sse2 msa/; 226$vp8_refining_search_sad_sse2=vp8_refining_search_sadx4; 227$vp8_refining_search_sad_msa=vp8_refining_search_sadx4; 228 229add_proto qw/int vp8_diamond_search_sad/, "struct macroblock *x, struct block *b, struct blockd *d, union int_mv *ref_mv, union int_mv *best_mv, int search_param, int sad_per_bit, int *num00, struct variance_vtable *fn_ptr, int *mvcost[2], union int_mv *center_mv"; 230specialize qw/vp8_diamond_search_sad sse2 msa/; 231$vp8_diamond_search_sad_sse2=vp8_diamond_search_sadx4; 232$vp8_diamond_search_sad_msa=vp8_diamond_search_sadx4; 233 234# 235# Alt-ref Noise Reduction (ARNR) 236# 237if (vpx_config("CONFIG_REALTIME_ONLY") ne "yes") { 238 add_proto qw/void vp8_temporal_filter_apply/, "unsigned char *frame1, unsigned int stride, unsigned char *frame2, unsigned int block_size, int strength, int filter_weight, unsigned int *accumulator, unsigned short *count"; 239 specialize qw/vp8_temporal_filter_apply sse2 msa/; 240} 241 242# 243# Denoiser filter 244# 245if (vpx_config("CONFIG_TEMPORAL_DENOISING") eq "yes") { 246 add_proto qw/int vp8_denoiser_filter/, "unsigned char *mc_running_avg_y, int mc_avg_y_stride, unsigned char *running_avg_y, int avg_y_stride, unsigned char *sig, int sig_stride, unsigned int motion_magnitude, int increase_denoising"; 247 specialize qw/vp8_denoiser_filter sse2 neon msa/; 248 add_proto qw/int vp8_denoiser_filter_uv/, "unsigned char *mc_running_avg, int mc_avg_stride, unsigned char *running_avg, int avg_stride, unsigned char *sig, int sig_stride, unsigned int motion_magnitude, int increase_denoising"; 249 specialize qw/vp8_denoiser_filter_uv sse2 neon msa/; 250} 251 252# End of encoder only functions 253} 2541; 255