FFmpeg: libavcodec/arm/dsputil

00001 /*
00002  * ARM NEON optimised DSP functions
00003  * Copyright (c) 2008 Mans Rullgard <mans@mansr.com>
00004  *
00005  * This file is part of FFmpeg.
00006  *
00007  * FFmpeg is free software; you can redistribute it and/or
00008  * modify it under the terms of the GNU Lesser General Public
00009  * License as published by the Free Software Foundation; either
00010  * version 2.1 of the License, or (at your option) any later version.
00011  *
00012  * FFmpeg is distributed in the hope that it will be useful,
00013  * but WITHOUT ANY WARRANTY; without even the implied warranty of
00014  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
00015  * Lesser General Public License for more details.
00016  *
00017  * You should have received a copy of the GNU Lesser General Public
00018  * License along with FFmpeg; if not, write to the Free Software
00019  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
00020  */
00021 
00022 #include <stdint.h>
00023 
00024 #include "libavcodec/avcodec.h"
00025 #include "libavcodec/dsputil.h"
00026 
00027 void ff_put_pixels16_neon(uint8_t *, const uint8_t *, int, int);
00028 void ff_put_pixels16_x2_neon(uint8_t *, const uint8_t *, int, int);
00029 void ff_put_pixels16_y2_neon(uint8_t *, const uint8_t *, int, int);
00030 void ff_put_pixels16_xy2_neon(uint8_t *, const uint8_t *, int, int);
00031 void ff_put_pixels8_neon(uint8_t *, const uint8_t *, int, int);
00032 void ff_put_pixels8_x2_neon(uint8_t *, const uint8_t *, int, int);
00033 void ff_put_pixels8_y2_neon(uint8_t *, const uint8_t *, int, int);
00034 void ff_put_pixels8_xy2_neon(uint8_t *, const uint8_t *, int, int);
00035 void ff_put_pixels16_x2_no_rnd_neon(uint8_t *, const uint8_t *, int, int);
00036 void ff_put_pixels16_y2_no_rnd_neon(uint8_t *, const uint8_t *, int, int);
00037 void ff_put_pixels16_xy2_no_rnd_neon(uint8_t *, const uint8_t *, int, int);
00038 void ff_put_pixels8_x2_no_rnd_neon(uint8_t *, const uint8_t *, int, int);
00039 void ff_put_pixels8_y2_no_rnd_neon(uint8_t *, const uint8_t *, int, int);
00040 void ff_put_pixels8_xy2_no_rnd_neon(uint8_t *, const uint8_t *, int, int);
00041 
00042 void ff_avg_pixels16_neon(uint8_t *, const uint8_t *, int, int);
00043 
00044 void ff_put_h264_qpel16_mc00_neon(uint8_t *, uint8_t *, int);
00045 void ff_put_h264_qpel16_mc10_neon(uint8_t *, uint8_t *, int);
00046 void ff_put_h264_qpel16_mc20_neon(uint8_t *, uint8_t *, int);
00047 void ff_put_h264_qpel16_mc30_neon(uint8_t *, uint8_t *, int);
00048 void ff_put_h264_qpel16_mc01_neon(uint8_t *, uint8_t *, int);
00049 void ff_put_h264_qpel16_mc11_neon(uint8_t *, uint8_t *, int);
00050 void ff_put_h264_qpel16_mc21_neon(uint8_t *, uint8_t *, int);
00051 void ff_put_h264_qpel16_mc31_neon(uint8_t *, uint8_t *, int);
00052 void ff_put_h264_qpel16_mc02_neon(uint8_t *, uint8_t *, int);
00053 void ff_put_h264_qpel16_mc12_neon(uint8_t *, uint8_t *, int);
00054 void ff_put_h264_qpel16_mc22_neon(uint8_t *, uint8_t *, int);
00055 void ff_put_h264_qpel16_mc32_neon(uint8_t *, uint8_t *, int);
00056 void ff_put_h264_qpel16_mc03_neon(uint8_t *, uint8_t *, int);
00057 void ff_put_h264_qpel16_mc13_neon(uint8_t *, uint8_t *, int);
00058 void ff_put_h264_qpel16_mc23_neon(uint8_t *, uint8_t *, int);
00059 void ff_put_h264_qpel16_mc33_neon(uint8_t *, uint8_t *, int);
00060 
00061 void ff_put_h264_qpel8_mc00_neon(uint8_t *, uint8_t *, int);
00062 void ff_put_h264_qpel8_mc10_neon(uint8_t *, uint8_t *, int);
00063 void ff_put_h264_qpel8_mc20_neon(uint8_t *, uint8_t *, int);
00064 void ff_put_h264_qpel8_mc30_neon(uint8_t *, uint8_t *, int);
00065 void ff_put_h264_qpel8_mc01_neon(uint8_t *, uint8_t *, int);
00066 void ff_put_h264_qpel8_mc11_neon(uint8_t *, uint8_t *, int);
00067 void ff_put_h264_qpel8_mc21_neon(uint8_t *, uint8_t *, int);
00068 void ff_put_h264_qpel8_mc31_neon(uint8_t *, uint8_t *, int);
00069 void ff_put_h264_qpel8_mc02_neon(uint8_t *, uint8_t *, int);
00070 void ff_put_h264_qpel8_mc12_neon(uint8_t *, uint8_t *, int);
00071 void ff_put_h264_qpel8_mc22_neon(uint8_t *, uint8_t *, int);
00072 void ff_put_h264_qpel8_mc32_neon(uint8_t *, uint8_t *, int);
00073 void ff_put_h264_qpel8_mc03_neon(uint8_t *, uint8_t *, int);
00074 void ff_put_h264_qpel8_mc13_neon(uint8_t *, uint8_t *, int);
00075 void ff_put_h264_qpel8_mc23_neon(uint8_t *, uint8_t *, int);
00076 void ff_put_h264_qpel8_mc33_neon(uint8_t *, uint8_t *, int);
00077 
00078 void ff_avg_h264_qpel16_mc00_neon(uint8_t *, uint8_t *, int);
00079 
00080 void ff_put_h264_chroma_mc8_neon(uint8_t *, uint8_t *, int, int, int, int);
00081 void ff_put_h264_chroma_mc4_neon(uint8_t *, uint8_t *, int, int, int, int);
00082 
00083 void ff_avg_h264_chroma_mc8_neon(uint8_t *, uint8_t *, int, int, int, int);
00084 void ff_avg_h264_chroma_mc4_neon(uint8_t *, uint8_t *, int, int, int, int);
00085 
00086 void ff_h264_v_loop_filter_luma_neon(uint8_t *pix, int stride, int alpha,
00087                                      int beta, int8_t *tc0);
00088 void ff_h264_h_loop_filter_luma_neon(uint8_t *pix, int stride, int alpha,
00089                                      int beta, int8_t *tc0);
00090 void ff_h264_v_loop_filter_chroma_neon(uint8_t *pix, int stride, int alpha,
00091                                        int beta, int8_t *tc0);
00092 void ff_h264_h_loop_filter_chroma_neon(uint8_t *pix, int stride, int alpha,
00093                                        int beta, int8_t *tc0);
00094 
00095 void ff_weight_h264_pixels_16x16_neon(uint8_t *ds, int stride, int log2_den,
00096                                       int weight, int offset);
00097 void ff_weight_h264_pixels_16x8_neon(uint8_t *ds, int stride, int log2_den,
00098                                      int weight, int offset);
00099 void ff_weight_h264_pixels_8x16_neon(uint8_t *ds, int stride, int log2_den,
00100                                      int weight, int offset);
00101 void ff_weight_h264_pixels_8x8_neon(uint8_t *ds, int stride, int log2_den,
00102                                     int weight, int offset);
00103 void ff_weight_h264_pixels_8x4_neon(uint8_t *ds, int stride, int log2_den,
00104                                     int weight, int offset);
00105 void ff_weight_h264_pixels_4x8_neon(uint8_t *ds, int stride, int log2_den,
00106                                     int weight, int offset);
00107 void ff_weight_h264_pixels_4x4_neon(uint8_t *ds, int stride, int log2_den,
00108                                     int weight, int offset);
00109 void ff_weight_h264_pixels_4x2_neon(uint8_t *ds, int stride, int log2_den,
00110                                     int weight, int offset);
00111 
00112 void ff_biweight_h264_pixels_16x16_neon(uint8_t *dst, uint8_t *src, int stride,
00113                                         int log2_den, int weightd, int weights,
00114                                         int offset);
00115 void ff_biweight_h264_pixels_16x8_neon(uint8_t *dst, uint8_t *src, int stride,
00116                                        int log2_den, int weightd, int weights,
00117                                        int offset);
00118 void ff_biweight_h264_pixels_8x16_neon(uint8_t *dst, uint8_t *src, int stride,
00119                                        int log2_den, int weightd, int weights,
00120                                        int offset);
00121 void ff_biweight_h264_pixels_8x8_neon(uint8_t *dst, uint8_t *src, int stride,
00122                                       int log2_den, int weightd, int weights,
00123                                       int offset);
00124 void ff_biweight_h264_pixels_8x4_neon(uint8_t *dst, uint8_t *src, int stride,
00125                                       int log2_den, int weightd, int weights,
00126                                       int offset);
00127 void ff_biweight_h264_pixels_4x8_neon(uint8_t *dst, uint8_t *src, int stride,
00128                                       int log2_den, int weightd, int weights,
00129                                       int offset);
00130 void ff_biweight_h264_pixels_4x4_neon(uint8_t *dst, uint8_t *src, int stride,
00131                                       int log2_den, int weightd, int weights,
00132                                       int offset);
00133 void ff_biweight_h264_pixels_4x2_neon(uint8_t *dst, uint8_t *src, int stride,
00134                                       int log2_den, int weightd, int weights,
00135                                       int offset);
00136 
00137 void ff_h264_idct_add_neon(uint8_t *dst, DCTELEM *block, int stride);
00138 void ff_h264_idct_dc_add_neon(uint8_t *dst, DCTELEM *block, int stride);
00139 void ff_h264_idct_add16_neon(uint8_t *dst, const int *block_offset,
00140                              DCTELEM *block, int stride,
00141                              const uint8_t nnzc[6*8]);
00142 void ff_h264_idct_add16intra_neon(uint8_t *dst, const int *block_offset,
00143                                   DCTELEM *block, int stride,
00144                                   const uint8_t nnzc[6*8]);
00145 void ff_h264_idct_add8_neon(uint8_t **dest, const int *block_offset,
00146                             DCTELEM *block, int stride,
00147                             const uint8_t nnzc[6*8]);
00148 
00149 void ff_vector_fmul_neon(float *dst, const float *src, int len);
00150 void ff_vector_fmul_window_neon(float *dst, const float *src0,
00151                                 const float *src1, const float *win,
00152                                 float add_bias, int len);
00153 
00154 void ff_float_to_int16_neon(int16_t *, const float *, long);
00155 void ff_float_to_int16_interleave_neon(int16_t *, const float **, long, int);
00156 
00157 void ff_dsputil_init_neon(DSPContext *c, AVCodecContext *avctx)
00158 {
00159     c->put_pixels_tab[0][0] = ff_put_pixels16_neon;
00160     c->put_pixels_tab[0][1] = ff_put_pixels16_x2_neon;
00161     c->put_pixels_tab[0][2] = ff_put_pixels16_y2_neon;
00162     c->put_pixels_tab[0][3] = ff_put_pixels16_xy2_neon;
00163     c->put_pixels_tab[1][0] = ff_put_pixels8_neon;
00164     c->put_pixels_tab[1][1] = ff_put_pixels8_x2_neon;
00165     c->put_pixels_tab[1][2] = ff_put_pixels8_y2_neon;
00166     c->put_pixels_tab[1][3] = ff_put_pixels8_xy2_neon;
00167 
00168     c->put_no_rnd_pixels_tab[0][0] = ff_put_pixels16_neon;
00169     c->put_no_rnd_pixels_tab[0][1] = ff_put_pixels16_x2_no_rnd_neon;
00170     c->put_no_rnd_pixels_tab[0][2] = ff_put_pixels16_y2_no_rnd_neon;
00171     c->put_no_rnd_pixels_tab[0][3] = ff_put_pixels16_xy2_no_rnd_neon;
00172     c->put_no_rnd_pixels_tab[1][0] = ff_put_pixels8_neon;
00173     c->put_no_rnd_pixels_tab[1][1] = ff_put_pixels8_x2_no_rnd_neon;
00174     c->put_no_rnd_pixels_tab[1][2] = ff_put_pixels8_y2_no_rnd_neon;
00175     c->put_no_rnd_pixels_tab[1][3] = ff_put_pixels8_xy2_no_rnd_neon;
00176 
00177     c->avg_pixels_tab[0][0] = ff_avg_pixels16_neon;
00178 
00179     c->put_h264_chroma_pixels_tab[0] = ff_put_h264_chroma_mc8_neon;
00180     c->put_h264_chroma_pixels_tab[1] = ff_put_h264_chroma_mc4_neon;
00181 
00182     c->avg_h264_chroma_pixels_tab[0] = ff_avg_h264_chroma_mc8_neon;
00183     c->avg_h264_chroma_pixels_tab[1] = ff_avg_h264_chroma_mc4_neon;
00184 
00185     c->put_h264_qpel_pixels_tab[0][ 0] = ff_put_h264_qpel16_mc00_neon;
00186     c->put_h264_qpel_pixels_tab[0][ 1] = ff_put_h264_qpel16_mc10_neon;
00187     c->put_h264_qpel_pixels_tab[0][ 2] = ff_put_h264_qpel16_mc20_neon;
00188     c->put_h264_qpel_pixels_tab[0][ 3] = ff_put_h264_qpel16_mc30_neon;
00189     c->put_h264_qpel_pixels_tab[0][ 4] = ff_put_h264_qpel16_mc01_neon;
00190     c->put_h264_qpel_pixels_tab[0][ 5] = ff_put_h264_qpel16_mc11_neon;
00191     c->put_h264_qpel_pixels_tab[0][ 6] = ff_put_h264_qpel16_mc21_neon;
00192     c->put_h264_qpel_pixels_tab[0][ 7] = ff_put_h264_qpel16_mc31_neon;
00193     c->put_h264_qpel_pixels_tab[0][ 8] = ff_put_h264_qpel16_mc02_neon;
00194     c->put_h264_qpel_pixels_tab[0][ 9] = ff_put_h264_qpel16_mc12_neon;
00195     c->put_h264_qpel_pixels_tab[0][10] = ff_put_h264_qpel16_mc22_neon;
00196     c->put_h264_qpel_pixels_tab[0][11] = ff_put_h264_qpel16_mc32_neon;
00197     c->put_h264_qpel_pixels_tab[0][12] = ff_put_h264_qpel16_mc03_neon;
00198     c->put_h264_qpel_pixels_tab[0][13] = ff_put_h264_qpel16_mc13_neon;
00199     c->put_h264_qpel_pixels_tab[0][14] = ff_put_h264_qpel16_mc23_neon;
00200     c->put_h264_qpel_pixels_tab[0][15] = ff_put_h264_qpel16_mc33_neon;
00201 
00202     c->put_h264_qpel_pixels_tab[1][ 0] = ff_put_h264_qpel8_mc00_neon;
00203     c->put_h264_qpel_pixels_tab[1][ 1] = ff_put_h264_qpel8_mc10_neon;
00204     c->put_h264_qpel_pixels_tab[1][ 2] = ff_put_h264_qpel8_mc20_neon;
00205     c->put_h264_qpel_pixels_tab[1][ 3] = ff_put_h264_qpel8_mc30_neon;
00206     c->put_h264_qpel_pixels_tab[1][ 4] = ff_put_h264_qpel8_mc01_neon;
00207     c->put_h264_qpel_pixels_tab[1][ 5] = ff_put_h264_qpel8_mc11_neon;
00208     c->put_h264_qpel_pixels_tab[1][ 6] = ff_put_h264_qpel8_mc21_neon;
00209     c->put_h264_qpel_pixels_tab[1][ 7] = ff_put_h264_qpel8_mc31_neon;
00210     c->put_h264_qpel_pixels_tab[1][ 8] = ff_put_h264_qpel8_mc02_neon;
00211     c->put_h264_qpel_pixels_tab[1][ 9] = ff_put_h264_qpel8_mc12_neon;
00212     c->put_h264_qpel_pixels_tab[1][10] = ff_put_h264_qpel8_mc22_neon;
00213     c->put_h264_qpel_pixels_tab[1][11] = ff_put_h264_qpel8_mc32_neon;
00214     c->put_h264_qpel_pixels_tab[1][12] = ff_put_h264_qpel8_mc03_neon;
00215     c->put_h264_qpel_pixels_tab[1][13] = ff_put_h264_qpel8_mc13_neon;
00216     c->put_h264_qpel_pixels_tab[1][14] = ff_put_h264_qpel8_mc23_neon;
00217     c->put_h264_qpel_pixels_tab[1][15] = ff_put_h264_qpel8_mc33_neon;
00218 
00219     c->avg_h264_qpel_pixels_tab[0][ 0] = ff_avg_h264_qpel16_mc00_neon;
00220 
00221     c->h264_v_loop_filter_luma = ff_h264_v_loop_filter_luma_neon;
00222     c->h264_h_loop_filter_luma = ff_h264_h_loop_filter_luma_neon;
00223     c->h264_v_loop_filter_chroma = ff_h264_v_loop_filter_chroma_neon;
00224     c->h264_h_loop_filter_chroma = ff_h264_h_loop_filter_chroma_neon;
00225 
00226     c->weight_h264_pixels_tab[0] = ff_weight_h264_pixels_16x16_neon;
00227     c->weight_h264_pixels_tab[1] = ff_weight_h264_pixels_16x8_neon;
00228     c->weight_h264_pixels_tab[2] = ff_weight_h264_pixels_8x16_neon;
00229     c->weight_h264_pixels_tab[3] = ff_weight_h264_pixels_8x8_neon;
00230     c->weight_h264_pixels_tab[4] = ff_weight_h264_pixels_8x4_neon;
00231     c->weight_h264_pixels_tab[5] = ff_weight_h264_pixels_4x8_neon;
00232     c->weight_h264_pixels_tab[6] = ff_weight_h264_pixels_4x4_neon;
00233     c->weight_h264_pixels_tab[7] = ff_weight_h264_pixels_4x2_neon;
00234 
00235     c->biweight_h264_pixels_tab[0] = ff_biweight_h264_pixels_16x16_neon;
00236     c->biweight_h264_pixels_tab[1] = ff_biweight_h264_pixels_16x8_neon;
00237     c->biweight_h264_pixels_tab[2] = ff_biweight_h264_pixels_8x16_neon;
00238     c->biweight_h264_pixels_tab[3] = ff_biweight_h264_pixels_8x8_neon;
00239     c->biweight_h264_pixels_tab[4] = ff_biweight_h264_pixels_8x4_neon;
00240     c->biweight_h264_pixels_tab[5] = ff_biweight_h264_pixels_4x8_neon;
00241     c->biweight_h264_pixels_tab[6] = ff_biweight_h264_pixels_4x4_neon;
00242     c->biweight_h264_pixels_tab[7] = ff_biweight_h264_pixels_4x2_neon;
00243 
00244     c->h264_idct_add = ff_h264_idct_add_neon;
00245     c->h264_idct_dc_add = ff_h264_idct_dc_add_neon;
00246     c->h264_idct_add16      = ff_h264_idct_add16_neon;
00247     c->h264_idct_add16intra = ff_h264_idct_add16intra_neon;
00248     c->h264_idct_add8       = ff_h264_idct_add8_neon;
00249 
00250     c->vector_fmul = ff_vector_fmul_neon;
00251     c->vector_fmul_window = ff_vector_fmul_window_neon;
00252 
00253     if (!(avctx->flags & CODEC_FLAG_BITEXACT)) {
00254         c->float_to_int16 = ff_float_to_int16_neon;
00255         c->float_to_int16_interleave = ff_float_to_int16_interleave_neon;
00256     }
00257 }
libavcodec/arm/dsputil_neon.c
