libavcodec/ps2/dsputil_mmi.c

   1 /*
   2  * MMI optimized DSP utils
   3  * Copyright (c) 2000, 2001 Fabrice Bellard.
   4  *
   5  * MMI optimization by Leon van Stuivenberg
   6  * clear_blocks_mmi() by BroadQ
   7  *
   8  * This file is part of FFmpeg.
   9  *
  10  * FFmpeg is free software; you can redistribute it and/or
  11  * modify it under the terms of the GNU Lesser General Public
  12  * License as published by the Free Software Foundation; either
  13  * version 2.1 of the License, or (at your option) any later version.
  14  *
  15  * FFmpeg is distributed in the hope that it will be useful,
  16  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  17  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  18  * Lesser General Public License for more details.
  19  *
  20  * You should have received a copy of the GNU Lesser General Public
  21  * License along with FFmpeg; if not, write to the Free Software
  22  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  23  */
  24
  25 #include "libavcodec/dsputil.h"
  26 #include "mmi.h"
  27
  28 void ff_mmi_idct_put(uint8_t *dest, int line_size, DCTELEM *block);
  29 void ff_mmi_idct_add(uint8_t *dest, int line_size, DCTELEM *block);
  30 void ff_mmi_idct(DCTELEM *block);
  31
  32 static void clear_blocks_mmi(DCTELEM * blocks)
  33 {
  34         __asm__ volatile(
  35         ".set noreorder    \n"
  36         "addiu $9, %0, 768 \n"
  37         "nop               \n"
  38         "1:                \n"
  39         "sq $0, 0(%0)      \n"
  40         "move $8, %0       \n"
  41         "addi %0, %0, 64   \n"
  42         "sq $0, 16($8)     \n"
  43         "slt $10, %0, $9   \n"
  44         "sq $0, 32($8)     \n"
  45         "bnez $10, 1b      \n"
  46         "sq $0, 48($8)     \n"
  47         ".set reorder      \n"
  48         : "+r" (blocks) ::  "$8", "$9", "memory" );
  49 }
  50
  51
  52 static void get_pixels_mmi(DCTELEM *block, const uint8_t *pixels, int line_size)
  53 {
  54         __asm__ volatile(
  55         ".set   push            \n\t"
  56         ".set   mips3           \n\t"
  57         "ld     $8, 0(%0)       \n\t"
  58         "add    %0, %0, %2      \n\t"
  59         "ld     $9, 0(%0)       \n\t"
  60         "add    %0, %0, %2      \n\t"
  61         "ld     $10, 0(%0)      \n\t"
  62         "pextlb $8, $0, $8      \n\t"
  63         "sq     $8, 0(%1)       \n\t"
  64         "add    %0, %0, %2      \n\t"
  65         "ld     $8, 0(%0)       \n\t"
  66         "pextlb $9, $0, $9      \n\t"
  67         "sq     $9, 16(%1)      \n\t"
  68         "add    %0, %0, %2      \n\t"
  69         "ld     $9, 0(%0)       \n\t"
  70         "pextlb $10, $0, $10    \n\t"
  71         "sq     $10, 32(%1)     \n\t"
  72         "add    %0, %0, %2      \n\t"
  73         "ld     $10, 0(%0)      \n\t"
  74         "pextlb $8, $0, $8      \n\t"
  75         "sq     $8, 48(%1)      \n\t"
  76         "add    %0, %0, %2      \n\t"
  77         "ld     $8, 0(%0)       \n\t"
  78         "pextlb $9, $0, $9      \n\t"
  79         "sq     $9, 64(%1)      \n\t"
  80         "add    %0, %0, %2      \n\t"
  81         "ld     $9, 0(%0)       \n\t"
  82         "pextlb $10, $0, $10    \n\t"
  83         "sq     $10, 80(%1)     \n\t"
  84         "pextlb $8, $0, $8      \n\t"
  85         "sq     $8, 96(%1)      \n\t"
  86         "pextlb $9, $0, $9      \n\t"
  87         "sq     $9, 112(%1)     \n\t"
  88         ".set   pop             \n\t"
  89         : "+r" (pixels) : "r" (block), "r" (line_size) : "$8", "$9", "$10", "memory" );
  90 }
  91
  92
  93 static void put_pixels8_mmi(uint8_t *block, const uint8_t *pixels, int line_size, int h)
  94 {
  95         __asm__ volatile(
  96         ".set   push            \n\t"
  97         ".set   mips3           \n\t"
  98         "1:                     \n\t"
  99         "ldr    $8, 0(%1)       \n\t"
 100         "addiu  %2, %2, -1      \n\t"
 101         "ldl    $8, 7(%1)       \n\t"
 102         "add    %1, %1, %3      \n\t"
 103         "sd     $8, 0(%0)       \n\t"
 104         "add    %0, %0, %3      \n\t"
 105         "bgtz   %2, 1b          \n\t"
 106         ".set   pop             \n\t"
 107         : "+r" (block), "+r" (pixels), "+r" (h) : "r" (line_size)
 108         : "$8", "memory" );
 109 }
 110
 111
 112 static void put_pixels16_mmi(uint8_t *block, const uint8_t *pixels, int line_size, int h)
 113 {
 114         __asm__ volatile (
 115         ".set   push            \n\t"
 116         ".set   mips3           \n\t"
 117         "1:                     \n\t"
 118         "ldr    $8, 0(%1)       \n\t"
 119         "add    $11, %1, %3     \n\t"
 120         "ldl    $8, 7(%1)       \n\t"
 121         "add    $10, %0, %3     \n\t"
 122         "ldr    $9, 8(%1)       \n\t"
 123         "ldl    $9, 15(%1)      \n\t"
 124         "ldr    $12, 0($11)     \n\t"
 125         "add    %1, $11, %3     \n\t"
 126         "ldl    $12, 7($11)     \n\t"
 127         "pcpyld $8, $9, $8      \n\t"
 128         "sq     $8, 0(%0)       \n\t"
 129         "ldr    $13, 8($11)     \n\t"
 130         "addiu  %2, %2, -2      \n\t"
 131         "ldl    $13, 15($11)    \n\t"
 132         "add    %0, $10, %3     \n\t"
 133         "pcpyld $12, $13, $12   \n\t"
 134         "sq     $12, 0($10)     \n\t"
 135         "bgtz   %2, 1b          \n\t"
 136         ".set   pop             \n\t"
 137         : "+r" (block), "+r" (pixels), "+r" (h) : "r" (line_size)
 138         : "$8", "$9", "$10", "$11", "$12", "$13", "memory" );
 139 }
 140
 141
 142 void dsputil_init_mmi(DSPContext* c, AVCodecContext *avctx)
 143 {
 144     const int idct_algo= avctx->idct_algo;
 145
 146     c->clear_blocks = clear_blocks_mmi;
 147
 148     c->put_pixels_tab[1][0] = put_pixels8_mmi;
 149     c->put_no_rnd_pixels_tab[1][0] = put_pixels8_mmi;
 150
 151     c->put_pixels_tab[0][0] = put_pixels16_mmi;
 152     c->put_no_rnd_pixels_tab[0][0] = put_pixels16_mmi;
 153
 154     c->get_pixels = get_pixels_mmi;
 155
 156     if(idct_algo==FF_IDCT_AUTO || idct_algo==FF_IDCT_PS2){
 157         c->idct_put= ff_mmi_idct_put;
 158         c->idct_add= ff_mmi_idct_add;
 159         c->idct    = ff_mmi_idct;
 160         c->idct_permutation_type= FF_LIBMPEG2_IDCT_PERM;
 161     }
 162 }
 163