libavcodec/arm/rv40dsp_init_arm.c

   1 /*
   2  * Copyright (c) 2011 Janne Grunau <janne-libav@jannau.net>
   3  *
   4  * This file is part of FFmpeg.
   5  *
   6  * FFmpeg is free software; you can redistribute it and/or
   7  * modify it under the terms of the GNU Lesser General Public
   8  * License as published by the Free Software Foundation; either
   9  * version 2.1 of the License, or (at your option) any later version.
  10  *
  11  * FFmpeg is distributed in the hope that it will be useful,
  12  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  13  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  14  * Lesser General Public License for more details.
  15  *
  16  * You should have received a copy of the GNU Lesser General Public
  17  * License along with FFmpeg; if not, write to the Free Software
  18  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  19  */
  20
  21 #include <stdint.h>
  22
  23 #include "libavutil/attributes.h"
  24 #include "libavcodec/avcodec.h"
  25 #include "libavcodec/rv34dsp.h"
  26 #include "libavutil/arm/cpu.h"
  27
  28 #define DECL_QPEL3(type, w, pos) \
  29 void ff_ ## type ## _rv40_qpel ## w ## _mc ## pos ## _neon(uint8_t *dst,       \
  30                                                            const uint8_t *src, \
  31                                                            ptrdiff_t stride)
  32
  33 #define DECL_QPEL2(w, pos)                      \
  34     DECL_QPEL3(put, w, pos);                    \
  35     DECL_QPEL3(avg, w, pos)
  36
  37 #define DECL_QPEL_XY(x, y)                      \
  38     DECL_QPEL2(16, x ## y);                     \
  39     DECL_QPEL2(8,  x ## y)
  40
  41 #define DECL_QPEL_Y(y)                          \
  42     DECL_QPEL_XY(0, y);                         \
  43     DECL_QPEL_XY(1, y);                         \
  44     DECL_QPEL_XY(2, y);                         \
  45     DECL_QPEL_XY(3, y);                         \
  46
  47 DECL_QPEL_Y(0);
  48 DECL_QPEL_Y(1);
  49 DECL_QPEL_Y(2);
  50 DECL_QPEL_Y(3);
  51
  52 void ff_put_rv40_chroma_mc8_neon(uint8_t *, uint8_t *, int, int, int, int);
  53 void ff_put_rv40_chroma_mc4_neon(uint8_t *, uint8_t *, int, int, int, int);
  54
  55 void ff_avg_rv40_chroma_mc8_neon(uint8_t *, uint8_t *, int, int, int, int);
  56 void ff_avg_rv40_chroma_mc4_neon(uint8_t *, uint8_t *, int, int, int, int);
  57
  58 void ff_rv40_weight_func_16_neon(uint8_t *, uint8_t *, uint8_t *, int, int, ptrdiff_t);
  59 void ff_rv40_weight_func_8_neon(uint8_t *, uint8_t *, uint8_t *, int, int, ptrdiff_t);
  60
  61 int ff_rv40_h_loop_filter_strength_neon(uint8_t *src, ptrdiff_t stride,
  62                                         int beta, int beta2, int edge,
  63                                         int *p1, int *q1);
  64 int ff_rv40_v_loop_filter_strength_neon(uint8_t *src, ptrdiff_t stride,
  65                                         int beta, int beta2, int edge,
  66                                         int *p1, int *q1);
  67
  68 void ff_rv40_h_weak_loop_filter_neon(uint8_t *src, ptrdiff_t stride, int filter_p1,
  69                                      int filter_q1, int alpha, int beta,
  70                                      int lim_p0q0, int lim_q1, int lim_p1);
  71 void ff_rv40_v_weak_loop_filter_neon(uint8_t *src, ptrdiff_t stride, int filter_p1,
  72                                      int filter_q1, int alpha, int beta,
  73                                      int lim_p0q0, int lim_q1, int lim_p1);
  74
  75 static av_cold void rv40dsp_init_neon(RV34DSPContext *c)
  76 {
  77     c->put_pixels_tab[0][ 1] = ff_put_rv40_qpel16_mc10_neon;
  78     c->put_pixels_tab[0][ 3] = ff_put_rv40_qpel16_mc30_neon;
  79     c->put_pixels_tab[0][ 4] = ff_put_rv40_qpel16_mc01_neon;
  80     c->put_pixels_tab[0][ 5] = ff_put_rv40_qpel16_mc11_neon;
  81     c->put_pixels_tab[0][ 6] = ff_put_rv40_qpel16_mc21_neon;
  82     c->put_pixels_tab[0][ 7] = ff_put_rv40_qpel16_mc31_neon;
  83     c->put_pixels_tab[0][ 9] = ff_put_rv40_qpel16_mc12_neon;
  84     c->put_pixels_tab[0][10] = ff_put_rv40_qpel16_mc22_neon;
  85     c->put_pixels_tab[0][11] = ff_put_rv40_qpel16_mc32_neon;
  86     c->put_pixels_tab[0][12] = ff_put_rv40_qpel16_mc03_neon;
  87     c->put_pixels_tab[0][13] = ff_put_rv40_qpel16_mc13_neon;
  88     c->put_pixels_tab[0][14] = ff_put_rv40_qpel16_mc23_neon;
  89     c->put_pixels_tab[0][15] = ff_put_rv40_qpel16_mc33_neon;
  90     c->avg_pixels_tab[0][ 1] = ff_avg_rv40_qpel16_mc10_neon;
  91     c->avg_pixels_tab[0][ 3] = ff_avg_rv40_qpel16_mc30_neon;
  92     c->avg_pixels_tab[0][ 4] = ff_avg_rv40_qpel16_mc01_neon;
  93     c->avg_pixels_tab[0][ 5] = ff_avg_rv40_qpel16_mc11_neon;
  94     c->avg_pixels_tab[0][ 6] = ff_avg_rv40_qpel16_mc21_neon;
  95     c->avg_pixels_tab[0][ 7] = ff_avg_rv40_qpel16_mc31_neon;
  96     c->avg_pixels_tab[0][ 9] = ff_avg_rv40_qpel16_mc12_neon;
  97     c->avg_pixels_tab[0][10] = ff_avg_rv40_qpel16_mc22_neon;
  98     c->avg_pixels_tab[0][11] = ff_avg_rv40_qpel16_mc32_neon;
  99     c->avg_pixels_tab[0][12] = ff_avg_rv40_qpel16_mc03_neon;
 100     c->avg_pixels_tab[0][13] = ff_avg_rv40_qpel16_mc13_neon;
 101     c->avg_pixels_tab[0][14] = ff_avg_rv40_qpel16_mc23_neon;
 102     c->avg_pixels_tab[0][15] = ff_avg_rv40_qpel16_mc33_neon;
 103     c->put_pixels_tab[1][ 1] = ff_put_rv40_qpel8_mc10_neon;
 104     c->put_pixels_tab[1][ 3] = ff_put_rv40_qpel8_mc30_neon;
 105     c->put_pixels_tab[1][ 4] = ff_put_rv40_qpel8_mc01_neon;
 106     c->put_pixels_tab[1][ 5] = ff_put_rv40_qpel8_mc11_neon;
 107     c->put_pixels_tab[1][ 6] = ff_put_rv40_qpel8_mc21_neon;
 108     c->put_pixels_tab[1][ 7] = ff_put_rv40_qpel8_mc31_neon;
 109     c->put_pixels_tab[1][ 9] = ff_put_rv40_qpel8_mc12_neon;
 110     c->put_pixels_tab[1][10] = ff_put_rv40_qpel8_mc22_neon;
 111     c->put_pixels_tab[1][11] = ff_put_rv40_qpel8_mc32_neon;
 112     c->put_pixels_tab[1][12] = ff_put_rv40_qpel8_mc03_neon;
 113     c->put_pixels_tab[1][13] = ff_put_rv40_qpel8_mc13_neon;
 114     c->put_pixels_tab[1][14] = ff_put_rv40_qpel8_mc23_neon;
 115     c->put_pixels_tab[1][15] = ff_put_rv40_qpel8_mc33_neon;
 116     c->avg_pixels_tab[1][ 1] = ff_avg_rv40_qpel8_mc10_neon;
 117     c->avg_pixels_tab[1][ 3] = ff_avg_rv40_qpel8_mc30_neon;
 118     c->avg_pixels_tab[1][ 4] = ff_avg_rv40_qpel8_mc01_neon;
 119     c->avg_pixels_tab[1][ 5] = ff_avg_rv40_qpel8_mc11_neon;
 120     c->avg_pixels_tab[1][ 6] = ff_avg_rv40_qpel8_mc21_neon;
 121     c->avg_pixels_tab[1][ 7] = ff_avg_rv40_qpel8_mc31_neon;
 122     c->avg_pixels_tab[1][ 9] = ff_avg_rv40_qpel8_mc12_neon;
 123     c->avg_pixels_tab[1][10] = ff_avg_rv40_qpel8_mc22_neon;
 124     c->avg_pixels_tab[1][11] = ff_avg_rv40_qpel8_mc32_neon;
 125     c->avg_pixels_tab[1][12] = ff_avg_rv40_qpel8_mc03_neon;
 126     c->avg_pixels_tab[1][13] = ff_avg_rv40_qpel8_mc13_neon;
 127     c->avg_pixels_tab[1][14] = ff_avg_rv40_qpel8_mc23_neon;
 128     c->avg_pixels_tab[1][15] = ff_avg_rv40_qpel8_mc33_neon;
 129
 130     c->put_chroma_pixels_tab[0] = ff_put_rv40_chroma_mc8_neon;
 131     c->put_chroma_pixels_tab[1] = ff_put_rv40_chroma_mc4_neon;
 132     c->avg_chroma_pixels_tab[0] = ff_avg_rv40_chroma_mc8_neon;
 133     c->avg_chroma_pixels_tab[1] = ff_avg_rv40_chroma_mc4_neon;
 134
 135     c->rv40_weight_pixels_tab[0][0] = ff_rv40_weight_func_16_neon;
 136     c->rv40_weight_pixels_tab[0][1] = ff_rv40_weight_func_8_neon;
 137
 138     c->rv40_loop_filter_strength[0] = ff_rv40_h_loop_filter_strength_neon;
 139     c->rv40_loop_filter_strength[1] = ff_rv40_v_loop_filter_strength_neon;
 140     c->rv40_weak_loop_filter[0]     = ff_rv40_h_weak_loop_filter_neon;
 141     c->rv40_weak_loop_filter[1]     = ff_rv40_v_weak_loop_filter_neon;
 142 }
 143
 144 av_cold void ff_rv40dsp_init_arm(RV34DSPContext *c)
 145 {
 146     int cpu_flags = av_get_cpu_flags();
 147
 148     if (have_neon(cpu_flags))
 149         rv40dsp_init_neon(c);
 150 }