OLD | NEW |
| (Empty) |
1 ; | |
2 ; Copyright (c) 2015 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 %include "third_party/x86inc/x86inc.asm" | |
11 | |
12 SECTION .text | |
13 | |
14 %macro REORDER_INPUTS 0 | |
15 ; a c d b to a b c d | |
16 SWAP 1, 3, 2 | |
17 %endmacro | |
18 | |
19 %macro TRANSFORM_COLS 0 | |
20 ; input: | |
21 ; m0 a | |
22 ; m1 b | |
23 ; m2 c | |
24 ; m3 d | |
25 paddw m0, m2 | |
26 psubw m3, m1 | |
27 | |
28 ; wide subtract | |
29 punpcklwd m4, m0 | |
30 punpcklwd m5, m3 | |
31 psrad m4, 16 | |
32 psrad m5, 16 | |
33 psubd m4, m5 | |
34 psrad m4, 1 | |
35 packssdw m4, m4 ; e | |
36 | |
37 psubw m5, m4, m1 ; b | |
38 psubw m4, m2 ; c | |
39 psubw m0, m5 | |
40 paddw m3, m4 | |
41 ; m0 a | |
42 SWAP 1, 5 ; m1 b | |
43 SWAP 2, 4 ; m2 c | |
44 ; m3 d | |
45 %endmacro | |
46 | |
47 %macro TRANSPOSE_4X4 0 | |
48 punpcklwd m0, m2 | |
49 punpcklwd m1, m3 | |
50 mova m2, m0 | |
51 punpcklwd m0, m1 | |
52 punpckhwd m2, m1 | |
53 pshufd m1, m0, 0x0e | |
54 pshufd m3, m2, 0x0e | |
55 %endmacro | |
56 | |
57 ; transpose a 4x4 int16 matrix in xmm0 and xmm1 to the bottom half of xmm0-xmm3 | |
58 %macro TRANSPOSE_4X4_WIDE 0 | |
59 mova m3, m0 | |
60 punpcklwd m0, m1 | |
61 punpckhwd m3, m1 | |
62 mova m2, m0 | |
63 punpcklwd m0, m3 | |
64 punpckhwd m2, m3 | |
65 pshufd m1, m0, 0x0e | |
66 pshufd m3, m2, 0x0e | |
67 %endmacro | |
68 | |
69 %macro ADD_STORE_4P_2X 5 ; src1, src2, tmp1, tmp2, zero | |
70 movq m%3, [outputq] | |
71 movq m%4, [outputq + strideq] | |
72 punpcklbw m%3, m%5 | |
73 punpcklbw m%4, m%5 | |
74 paddw m%1, m%3 | |
75 paddw m%2, m%4 | |
76 packuswb m%1, m%5 | |
77 packuswb m%2, m%5 | |
78 movd [outputq], m%1 | |
79 movd [outputq + strideq], m%2 | |
80 %endmacro | |
81 | |
82 INIT_XMM sse2 | |
83 cglobal iwht4x4_16_add, 3, 3, 7, input, output, stride | |
84 mova m0, [inputq + 0] | |
85 mova m1, [inputq + 16] | |
86 | |
87 psraw m0, 2 | |
88 psraw m1, 2 | |
89 | |
90 TRANSPOSE_4X4_WIDE | |
91 REORDER_INPUTS | |
92 TRANSFORM_COLS | |
93 TRANSPOSE_4X4 | |
94 REORDER_INPUTS | |
95 TRANSFORM_COLS | |
96 | |
97 pxor m4, m4 | |
98 ADD_STORE_4P_2X 0, 1, 5, 6, 4 | |
99 lea outputq, [outputq + 2 * strideq] | |
100 ADD_STORE_4P_2X 2, 3, 5, 6, 4 | |
101 | |
102 RET | |
OLD | NEW |