media/libvpx/vp8/common/x86/recon_sse2.asm

   1 ;
   2 ;  Copyright (c) 2010 The WebM project authors. All Rights Reserved.
   3 ;
   4 ;  Use of this source code is governed by a BSD-style license
   5 ;  that can be found in the LICENSE file in the root of the source
   6 ;  tree. An additional intellectual property rights grant can be found
   7 ;  in the file PATENTS.  All contributing project authors may
   8 ;  be found in the AUTHORS file in the root of the source tree.
   9 ;
  10
  11
  12 %include "vpx_ports/x86_abi_support.asm"
  13 ;void vp8_recon2b_sse2(unsigned char *s, short *q, unsigned char *d, int stride)
  14 global sym(vp8_recon2b_sse2)
  15 sym(vp8_recon2b_sse2):
  16     push        rbp
  17     mov         rbp, rsp
  18     SHADOW_ARGS_TO_STACK 4
  19     push        rsi
  20     push        rdi
  21     ; end prolog
  22
  23         mov         rsi,        arg(0) ;s
  24         mov         rdi,        arg(2) ;d
  25         mov         rdx,        arg(1) ;q
  26         movsxd      rax,        dword ptr arg(3) ;stride
  27         pxor        xmm0,       xmm0
  28
  29         movq        xmm1,       MMWORD PTR [rsi]
  30         punpcklbw   xmm1,       xmm0
  31         paddsw      xmm1,       XMMWORD PTR [rdx]
  32         packuswb    xmm1,       xmm0              ; pack and unpack to saturate
  33         movq        MMWORD PTR [rdi],   xmm1
  34
  35
  36         movq        xmm2,       MMWORD PTR [rsi+8]
  37         punpcklbw   xmm2,       xmm0
  38         paddsw      xmm2,       XMMWORD PTR [rdx+16]
  39         packuswb    xmm2,       xmm0              ; pack and unpack to saturate
  40         movq        MMWORD PTR [rdi+rax],   xmm2
  41
  42
  43         movq        xmm3,       MMWORD PTR [rsi+16]
  44         punpcklbw   xmm3,       xmm0
  45         paddsw      xmm3,       XMMWORD PTR [rdx+32]
  46         packuswb    xmm3,       xmm0              ; pack and unpack to saturate
  47         movq        MMWORD PTR [rdi+rax*2], xmm3
  48
  49         add         rdi, rax
  50         movq        xmm4,       MMWORD PTR [rsi+24]
  51         punpcklbw   xmm4,       xmm0
  52         paddsw      xmm4,       XMMWORD PTR [rdx+48]
  53         packuswb    xmm4,       xmm0              ; pack and unpack to saturate
  54         movq        MMWORD PTR [rdi+rax*2], xmm4
  55
  56     ; begin epilog
  57     pop rdi
  58     pop rsi
  59     UNSHADOW_ARGS
  60     pop         rbp
  61     ret
  62
  63
  64 ;void vp8_recon4b_sse2(unsigned char *s, short *q, unsigned char *d, int stride)
  65 global sym(vp8_recon4b_sse2)
  66 sym(vp8_recon4b_sse2):
  67     push        rbp
  68     mov         rbp, rsp
  69     SHADOW_ARGS_TO_STACK 4
  70     SAVE_XMM
  71     push        rsi
  72     push        rdi
  73     ; end prolog
  74
  75         mov         rsi,        arg(0) ;s
  76         mov         rdi,        arg(2) ;d
  77         mov         rdx,        arg(1) ;q
  78         movsxd      rax,        dword ptr arg(3) ;stride
  79         pxor        xmm0,       xmm0
  80
  81         movdqa      xmm1,       XMMWORD PTR [rsi]
  82         movdqa      xmm5,       xmm1
  83         punpcklbw   xmm1,       xmm0
  84         punpckhbw   xmm5,       xmm0
  85         paddsw      xmm1,       XMMWORD PTR [rdx]
  86         paddsw      xmm5,       XMMWORD PTR [rdx+16]
  87         packuswb    xmm1,       xmm5              ; pack and unpack to saturate
  88         movdqa      XMMWORD PTR [rdi],  xmm1
  89
  90
  91         movdqa      xmm2,       XMMWORD PTR [rsi+16]
  92         movdqa      xmm6,       xmm2
  93         punpcklbw   xmm2,       xmm0
  94         punpckhbw   xmm6,       xmm0
  95         paddsw      xmm2,       XMMWORD PTR [rdx+32]
  96         paddsw      xmm6,       XMMWORD PTR [rdx+48]
  97         packuswb    xmm2,       xmm6              ; pack and unpack to saturate
  98         movdqa      XMMWORD PTR [rdi+rax],  xmm2
  99
 100
 101         movdqa      xmm3,       XMMWORD PTR [rsi+32]
 102         movdqa      xmm7,       xmm3
 103         punpcklbw   xmm3,       xmm0
 104         punpckhbw   xmm7,       xmm0
 105         paddsw      xmm3,       XMMWORD PTR [rdx+64]
 106         paddsw      xmm7,       XMMWORD PTR [rdx+80]
 107         packuswb    xmm3,       xmm7              ; pack and unpack to saturate
 108         movdqa      XMMWORD PTR [rdi+rax*2],    xmm3
 109
 110         add       rdi, rax
 111         movdqa      xmm4,       XMMWORD PTR [rsi+48]
 112         movdqa      xmm5,       xmm4
 113         punpcklbw   xmm4,       xmm0
 114         punpckhbw   xmm5,       xmm0
 115         paddsw      xmm4,       XMMWORD PTR [rdx+96]
 116         paddsw      xmm5,       XMMWORD PTR [rdx+112]
 117         packuswb    xmm4,       xmm5              ; pack and unpack to saturate
 118         movdqa      XMMWORD PTR [rdi+rax*2],    xmm4
 119
 120     ; begin epilog
 121     pop rdi
 122     pop rsi
 123     RESTORE_XMM
 124     UNSHADOW_ARGS
 125     pop         rbp
 126     ret
 127
 128
 129 ;void copy_mem16x16_sse2(
 130 ;    unsigned char *src,
 131 ;    int src_stride,
 132 ;    unsigned char *dst,
 133 ;    int dst_stride
 134 ;    )
 135 global sym(vp8_copy_mem16x16_sse2)
 136 sym(vp8_copy_mem16x16_sse2):
 137     push        rbp
 138     mov         rbp, rsp
 139     SHADOW_ARGS_TO_STACK 4
 140     push        rsi
 141     push        rdi
 142     ; end prolog
 143
 144         mov         rsi,        arg(0) ;src;
 145         movdqu      xmm0,       [rsi]
 146
 147         movsxd      rax,        dword ptr arg(1) ;src_stride;
 148         mov         rdi,        arg(2) ;dst;
 149
 150         movdqu      xmm1,       [rsi+rax]
 151         movdqu      xmm2,       [rsi+rax*2]
 152
 153         movsxd      rcx,        dword ptr arg(3) ;dst_stride
 154         lea         rsi,        [rsi+rax*2]
 155
 156         movdqa      [rdi],      xmm0
 157         add         rsi,        rax
 158
 159         movdqa      [rdi+rcx],  xmm1
 160         movdqa      [rdi+rcx*2],xmm2
 161
 162         lea         rdi,        [rdi+rcx*2]
 163         movdqu      xmm3,       [rsi]
 164
 165         add         rdi,        rcx
 166         movdqu      xmm4,       [rsi+rax]
 167
 168         movdqu      xmm5,       [rsi+rax*2]
 169         lea         rsi,        [rsi+rax*2]
 170
 171         movdqa      [rdi],  xmm3
 172         add         rsi,        rax
 173
 174         movdqa      [rdi+rcx],  xmm4
 175         movdqa      [rdi+rcx*2],xmm5
 176
 177         lea         rdi,        [rdi+rcx*2]
 178         movdqu      xmm0,       [rsi]
 179
 180         add         rdi,        rcx
 181         movdqu      xmm1,       [rsi+rax]
 182
 183         movdqu      xmm2,       [rsi+rax*2]
 184         lea         rsi,        [rsi+rax*2]
 185
 186         movdqa      [rdi],      xmm0
 187         add         rsi,        rax
 188
 189         movdqa      [rdi+rcx],  xmm1
 190
 191         movdqa      [rdi+rcx*2],    xmm2
 192         movdqu      xmm3,       [rsi]
 193
 194         movdqu      xmm4,       [rsi+rax]
 195         lea         rdi,        [rdi+rcx*2]
 196
 197         add         rdi,        rcx
 198         movdqu      xmm5,       [rsi+rax*2]
 199
 200         lea         rsi,        [rsi+rax*2]
 201         movdqa      [rdi],  xmm3
 202
 203         add         rsi,        rax
 204         movdqa      [rdi+rcx],  xmm4
 205
 206         movdqa      [rdi+rcx*2],xmm5
 207         movdqu      xmm0,       [rsi]
 208
 209         lea         rdi,        [rdi+rcx*2]
 210         movdqu      xmm1,       [rsi+rax]
 211
 212         add         rdi,        rcx
 213         movdqu      xmm2,       [rsi+rax*2]
 214
 215         lea         rsi,        [rsi+rax*2]
 216         movdqa      [rdi],      xmm0
 217
 218         movdqa      [rdi+rcx],  xmm1
 219         movdqa      [rdi+rcx*2],xmm2
 220
 221         movdqu      xmm3,       [rsi+rax]
 222         lea         rdi,        [rdi+rcx*2]
 223
 224         movdqa      [rdi+rcx],  xmm3
 225
 226     ; begin epilog
 227     pop rdi
 228     pop rsi
 229     UNSHADOW_ARGS
 230     pop         rbp
 231     ret