hpeldsp_alpha_asm.S

   1 /*
   2  * Alpha optimized DSP utils
   3  * Copyright (c) 2002 Falk Hueffner <falk@debian.org>
   4  *
   5  * This file is part of FFmpeg.
   6  *
   7  * FFmpeg is free software; you can redistribute it and/or
   8  * modify it under the terms of the GNU Lesser General Public
   9  * License as published by the Free Software Foundation; either
  10  * version 2.1 of the License, or (at your option) any later version.
  11  *
  12  * FFmpeg is distributed in the hope that it will be useful,
  13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  15  * Lesser General Public License for more details.
  16  *
  17  * You should have received a copy of the GNU Lesser General Public
  18  * License along with FFmpeg; if not, write to the Free Software
  19  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  20  */
  21
  22 /*
  23  * These functions are scheduled for pca56. They should work
  24  * reasonably on ev6, though.
  25  */
  26
  27 #include "regdef.h"
  28
  29
  30         .set noat
  31         .set noreorder
  32         .arch pca56
  33         .text
  34
  35 /************************************************************************
  36  * void put_pixels_axp_asm(uint8_t *block, const uint8_t *pixels,
  37  *                         int line_size, int h)
  38  */
  39         .align 6
  40         .globl put_pixels_axp_asm
  41         .ent put_pixels_axp_asm
  42 put_pixels_axp_asm:
  43         .frame sp, 0, ra
  44         .prologue 0
  45
  46         and     a1, 7, t0
  47         beq     t0, $aligned
  48
  49         .align 4
  50 $unaligned:
  51         ldq_u   t0, 0(a1)
  52         ldq_u   t1, 8(a1)
  53         addq    a1, a2, a1
  54         nop
  55
  56         ldq_u   t2, 0(a1)
  57         ldq_u   t3, 8(a1)
  58         addq    a1, a2, a1
  59         nop
  60
  61         ldq_u   t4, 0(a1)
  62         ldq_u   t5, 8(a1)
  63         addq    a1, a2, a1
  64         nop
  65
  66         ldq_u   t6, 0(a1)
  67         ldq_u   t7, 8(a1)
  68         extql   t0, a1, t0
  69         addq    a1, a2, a1
  70
  71         extqh   t1, a1, t1
  72         addq    a0, a2, t8
  73         extql   t2, a1, t2
  74         addq    t8, a2, t9
  75
  76         extqh   t3, a1, t3
  77         addq    t9, a2, ta
  78         extql   t4, a1, t4
  79         or      t0, t1, t0
  80
  81         extqh   t5, a1, t5
  82         or      t2, t3, t2
  83         extql   t6, a1, t6
  84         or      t4, t5, t4
  85
  86         extqh   t7, a1, t7
  87         or      t6, t7, t6
  88         stq     t0, 0(a0)
  89         stq     t2, 0(t8)
  90
  91         stq     t4, 0(t9)
  92         subq    a3, 4, a3
  93         stq     t6, 0(ta)
  94         addq    ta, a2, a0
  95
  96         bne     a3, $unaligned
  97         ret
  98
  99         .align 4
 100 $aligned:
 101         ldq     t0, 0(a1)
 102         addq    a1, a2, a1
 103         ldq     t1, 0(a1)
 104         addq    a1, a2, a1
 105
 106         ldq     t2, 0(a1)
 107         addq    a1, a2, a1
 108         ldq     t3, 0(a1)
 109
 110         addq    a0, a2, t4
 111         addq    a1, a2, a1
 112         addq    t4, a2, t5
 113         subq    a3, 4, a3
 114
 115         stq     t0, 0(a0)
 116         addq    t5, a2, t6
 117         stq     t1, 0(t4)
 118         addq    t6, a2, a0
 119
 120         stq     t2, 0(t5)
 121         stq     t3, 0(t6)
 122
 123         bne     a3, $aligned
 124         ret
 125         .end put_pixels_axp_asm