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    EXPORT  |vp8_short_inv_walsh4x4_neon|
11
12    ARM
13    REQUIRE8
14    PRESERVE8
15
16    AREA    |.text|, CODE, READONLY  ; name this block of code
17
18;short vp8_short_inv_walsh4x4_neon(short *input, short *mb_dqcoeff)
19|vp8_short_inv_walsh4x4_neon| PROC
20
21    ; read in all four lines of values: d0->d3
22    vld1.i16 {q0-q1}, [r0@128]
23
24    ; first for loop
25    vadd.s16 d4, d0, d3 ;a = [0] + [12]
26    vadd.s16 d6, d1, d2 ;b = [4] + [8]
27    vsub.s16 d5, d0, d3 ;d = [0] - [12]
28    vsub.s16 d7, d1, d2 ;c = [4] - [8]
29
30    vadd.s16 q0, q2, q3 ; a+b d+c
31    vsub.s16 q1, q2, q3 ; a-b d-c
32
33    vtrn.32 d0, d2 ;d0:  0  1  8  9
34                   ;d2:  2  3 10 11
35    vtrn.32 d1, d3 ;d1:  4  5 12 13
36                   ;d3:  6  7 14 15
37
38    vtrn.16 d0, d1 ;d0:  0  4  8 12
39                   ;d1:  1  5  9 13
40    vtrn.16 d2, d3 ;d2:  2  6 10 14
41                   ;d3:  3  7 11 15
42
43    ; second for loop
44
45    vadd.s16 d4, d0, d3 ;a = [0] + [3]
46    vadd.s16 d6, d1, d2 ;b = [1] + [2]
47    vsub.s16 d5, d0, d3 ;d = [0] - [3]
48    vsub.s16 d7, d1, d2 ;c = [1] - [2]
49
50    vmov.i16 q8, #3
51
52    vadd.s16 q0, q2, q3 ; a+b d+c
53    vsub.s16 q1, q2, q3 ; a-b d-c
54
55    vadd.i16 q0, q0, q8 ;e/f += 3
56    vadd.i16 q1, q1, q8 ;g/h += 3
57
58    vshr.s16 q0, q0, #3 ;e/f >> 3
59    vshr.s16 q1, q1, #3 ;g/h >> 3
60
61    mov      r2, #64
62    add      r3, r1, #32
63
64    vst1.i16 d0[0], [r1],r2
65    vst1.i16 d1[0], [r3],r2
66    vst1.i16 d2[0], [r1],r2
67    vst1.i16 d3[0], [r3],r2
68
69    vst1.i16 d0[1], [r1],r2
70    vst1.i16 d1[1], [r3],r2
71    vst1.i16 d2[1], [r1],r2
72    vst1.i16 d3[1], [r3],r2
73
74    vst1.i16 d0[2], [r1],r2
75    vst1.i16 d1[2], [r3],r2
76    vst1.i16 d2[2], [r1],r2
77    vst1.i16 d3[2], [r3],r2
78
79    vst1.i16 d0[3], [r1],r2
80    vst1.i16 d1[3], [r3],r2
81    vst1.i16 d2[3], [r1]
82    vst1.i16 d3[3], [r3]
83
84    bx lr
85    ENDP    ; |vp8_short_inv_walsh4x4_neon|
86
87    END
88