vp8/common/arm/neon/recon4b_neon.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     EXPORT  |vp8_recon4b_neon|
  13     ARM
  14     REQUIRE8
  15     PRESERVE8
  16
  17     AREA ||.text||, CODE, READONLY, ALIGN=2
  18
  19 ; r0    unsigned char  *pred_ptr,
  20 ; r1    short *diff_ptr,
  21 ; r2    unsigned char *dst_ptr,
  22 ; r3    int stride
  23
  24 |vp8_recon4b_neon| PROC
  25     vld1.u8         {q12, q13}, [r0]!   ;load data from pred_ptr
  26     vld1.16         {q8, q9}, [r1]!     ;load data from diff_ptr
  27     vld1.u8         {q14, q15}, [r0]
  28     vld1.16         {q10, q11}, [r1]!
  29
  30     vmovl.u8        q0, d24             ;modify Pred data from 8 bits to 16 bits
  31     vmovl.u8        q1, d25
  32     vmovl.u8        q2, d26
  33     vmovl.u8        q3, d27
  34     vmovl.u8        q4, d28
  35     vmovl.u8        q5, d29
  36     vmovl.u8        q6, d30
  37     vld1.16         {q12, q13}, [r1]!
  38     vmovl.u8        q7, d31
  39     vld1.16         {q14, q15}, [r1]
  40
  41     vadd.s16        q0, q0, q8          ;add Diff data and Pred data together
  42     vadd.s16        q1, q1, q9
  43     vadd.s16        q2, q2, q10
  44     vadd.s16        q3, q3, q11
  45     vadd.s16        q4, q4, q12
  46     vadd.s16        q5, q5, q13
  47     vadd.s16        q6, q6, q14
  48     vadd.s16        q7, q7, q15
  49
  50     vqmovun.s16     d0, q0              ;CLAMP() saturation
  51     vqmovun.s16     d1, q1
  52     vqmovun.s16     d2, q2
  53     vqmovun.s16     d3, q3
  54     vqmovun.s16     d4, q4
  55     vqmovun.s16     d5, q5
  56     vqmovun.s16     d6, q6
  57     vqmovun.s16     d7, q7
  58     add             r0, r2, r3
  59
  60     vst1.u8         {q0}, [r2]          ;store result
  61     vst1.u8         {q1}, [r0], r3
  62     add             r2, r0, r3
  63     vst1.u8         {q2}, [r0]
  64     vst1.u8         {q3}, [r2], r3
  65
  66     bx             lr
  67
  68     ENDP
  69     END