LCOV - code coverage report
Current view: top level - src/libavcodec/x86 - idctdsp_init.c (source / functions) Hit Total Coverage
Test: coverage.info Lines: 40 58 69.0 %
Date: 2017-01-21 09:32:20 Functions: 2 2 100.0 %

          Line data    Source code
       1             : /*
       2             :  * This file is part of FFmpeg.
       3             :  *
       4             :  * FFmpeg is free software; you can redistribute it and/or
       5             :  * modify it under the terms of the GNU Lesser General Public
       6             :  * License as published by the Free Software Foundation; either
       7             :  * version 2.1 of the License, or (at your option) any later version.
       8             :  *
       9             :  * FFmpeg is distributed in the hope that it will be useful,
      10             :  * but WITHOUT ANY WARRANTY; without even the implied warranty of
      11             :  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
      12             :  * Lesser General Public License for more details.
      13             :  *
      14             :  * You should have received a copy of the GNU Lesser General Public
      15             :  * License along with FFmpeg; if not, write to the Free Software
      16             :  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
      17             :  */
      18             : 
      19             : #include "config.h"
      20             : #include "libavutil/attributes.h"
      21             : #include "libavutil/cpu.h"
      22             : #include "libavutil/x86/cpu.h"
      23             : #include "libavcodec/avcodec.h"
      24             : #include "libavcodec/idctdsp.h"
      25             : #include "idctdsp.h"
      26             : #include "simple_idct.h"
      27             : 
      28             : /* Input permutation for the simple_idct_mmx */
      29             : static const uint8_t simple_mmx_permutation[64] = {
      30             :     0x00, 0x08, 0x04, 0x09, 0x01, 0x0C, 0x05, 0x0D,
      31             :     0x10, 0x18, 0x14, 0x19, 0x11, 0x1C, 0x15, 0x1D,
      32             :     0x20, 0x28, 0x24, 0x29, 0x21, 0x2C, 0x25, 0x2D,
      33             :     0x12, 0x1A, 0x16, 0x1B, 0x13, 0x1E, 0x17, 0x1F,
      34             :     0x02, 0x0A, 0x06, 0x0B, 0x03, 0x0E, 0x07, 0x0F,
      35             :     0x30, 0x38, 0x34, 0x39, 0x31, 0x3C, 0x35, 0x3D,
      36             :     0x22, 0x2A, 0x26, 0x2B, 0x23, 0x2E, 0x27, 0x2F,
      37             :     0x32, 0x3A, 0x36, 0x3B, 0x33, 0x3E, 0x37, 0x3F,
      38             : };
      39             : 
      40             : static const uint8_t idct_sse2_row_perm[8] = { 0, 4, 1, 5, 2, 6, 3, 7 };
      41             : 
      42        1587 : av_cold int ff_init_scantable_permutation_x86(uint8_t *idct_permutation,
      43             :                                               enum idct_permutation_type perm_type)
      44             : {
      45             :     int i;
      46             : 
      47        1587 :     switch (perm_type) {
      48             :     case FF_IDCT_PERM_SIMPLE:
      49       16900 :         for (i = 0; i < 64; i++)
      50       16640 :             idct_permutation[i] = simple_mmx_permutation[i];
      51         260 :         return 1;
      52             :     case FF_IDCT_PERM_SSE2:
      53         260 :         for (i = 0; i < 64; i++)
      54         256 :             idct_permutation[i] = (i & 0x38) | idct_sse2_row_perm[i & 7];
      55           4 :         return 1;
      56             :     }
      57             : 
      58        1323 :     return 0;
      59             : }
      60             : 
      61        1438 : av_cold void ff_idctdsp_init_x86(IDCTDSPContext *c, AVCodecContext *avctx,
      62             :                                  unsigned high_bit_depth)
      63             : {
      64        1438 :     int cpu_flags = av_get_cpu_flags();
      65             : 
      66        1438 :     if (INLINE_MMX(cpu_flags)) {
      67         526 :         if (!high_bit_depth &&
      68         526 :             avctx->lowres == 0 &&
      69         266 :             (avctx->idct_algo == FF_IDCT_AUTO ||
      70           6 :              avctx->idct_algo == FF_IDCT_SIMPLEAUTO ||
      71           3 :              avctx->idct_algo == FF_IDCT_SIMPLEMMX)) {
      72         260 :                 c->idct_put  = ff_simple_idct_put_mmx;
      73         260 :                 c->idct_add  = ff_simple_idct_add_mmx;
      74         260 :                 c->idct      = ff_simple_idct_mmx;
      75         260 :                 c->perm_type = FF_IDCT_PERM_SIMPLE;
      76             :         }
      77             :     }
      78        1438 :     if (EXTERNAL_MMX(cpu_flags)) {
      79         263 :         c->put_signed_pixels_clamped = ff_put_signed_pixels_clamped_mmx;
      80         263 :         c->put_pixels_clamped        = ff_put_pixels_clamped_mmx;
      81         263 :         c->add_pixels_clamped        = ff_add_pixels_clamped_mmx;
      82             :     }
      83        1438 :     if (EXTERNAL_SSE2(cpu_flags)) {
      84         263 :         c->put_signed_pixels_clamped = ff_put_signed_pixels_clamped_sse2;
      85         263 :         c->put_pixels_clamped        = ff_put_pixels_clamped_sse2;
      86         263 :         c->add_pixels_clamped        = ff_add_pixels_clamped_sse2;
      87             :     }
      88             : 
      89        1438 :     if (ARCH_X86_64 && avctx->lowres == 0) {
      90        1459 :         if (avctx->bits_per_raw_sample == 10 &&
      91          46 :         (avctx->idct_algo == FF_IDCT_AUTO ||
      92          46 :          avctx->idct_algo == FF_IDCT_SIMPLEAUTO ||
      93          23 :          avctx->idct_algo == FF_IDCT_SIMPLE)) {
      94          23 :         if (EXTERNAL_SSE2(cpu_flags)) {
      95           0 :             c->idct_put  = ff_simple_idct10_put_sse2;
      96           0 :             c->idct_add  = NULL;
      97           0 :             c->idct      = ff_simple_idct10_sse2;
      98           0 :             c->perm_type = FF_IDCT_PERM_TRANSPOSE;
      99             : 
     100             :         }
     101          23 :         if (EXTERNAL_AVX(cpu_flags)) {
     102           0 :             c->idct_put  = ff_simple_idct10_put_avx;
     103           0 :             c->idct_add  = NULL;
     104           0 :             c->idct      = ff_simple_idct10_avx;
     105           0 :             c->perm_type = FF_IDCT_PERM_TRANSPOSE;
     106             :         }
     107             :         }
     108             : 
     109        1440 :         if (avctx->bits_per_raw_sample == 12 &&
     110           8 :             (avctx->idct_algo == FF_IDCT_AUTO ||
     111           4 :              avctx->idct_algo == FF_IDCT_SIMPLEMMX)) {
     112           0 :             if (EXTERNAL_SSE2(cpu_flags)) {
     113           0 :                 c->idct_put  = ff_simple_idct12_put_sse2;
     114           0 :                 c->idct_add  = NULL;
     115           0 :                 c->idct      = ff_simple_idct12_sse2;
     116           0 :                 c->perm_type = FF_IDCT_PERM_TRANSPOSE;
     117             :             }
     118           0 :             if (EXTERNAL_AVX(cpu_flags)) {
     119           0 :                 c->idct_put  = ff_simple_idct12_put_avx;
     120           0 :                 c->idct_add  = NULL;
     121           0 :                 c->idct      = ff_simple_idct12_avx;
     122           0 :                 c->perm_type = FF_IDCT_PERM_TRANSPOSE;
     123             :             }
     124             :         }
     125             :     }
     126        1438 : }

Generated by: LCOV version 1.12