1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
|
/*
* Copyright © <2010>, Intel Corporation.
*
* This program is licensed under the terms and conditions of the
* Eclipse Public License (EPL), version 1.0. The full text of the EPL is at
* http://www.opensource.org/licenses/eclipse-1.0.php.
*
*/
//////////////////////////////////////////////////////////////////////////////////////////
// Module name: TransposeNV12_4x16.asm
//
// Transpose a 4x16 internal planar to 16x4 internal planar block
//
//----------------------------------------------------------------------------------------
// Symbols need to be defined before including this module
//
// Source region is :ub
// SRC_YB: SRC_YB Base=rxx ElementSize=1 SrcRegion=REGION(16,1) Type=ub // 8 GRFs
// SRC_UW: SRC_UB Base=rxx ElementSize=2 SrcRegion=REGION(8,1) Type=uw // 4 GRFs
//
// Temp buffer:
// BUF_B: BUF_B Base=rxx ElementSize=1 SrcRegion=REGION(16,1) Type=ub // 8 GRFs
// BUF_W: BUF_W Base=rxx ElementSize=2 SrcRegion=REGION(8,1) Type=uw // 4 GRFs
//
//////////////////////////////////////////////////////////////////////////////////////////
#if defined(_DEBUG)
mov (1) EntrySignatureC:w 0xDDDB:w
#endif
// Transpose Y (4x16) right most 4 columns
// The first step
mov (16) BUF_B(0,0)<1> SRC_YB(0,0)<16;4,1> // Read 2 rows, write 1 row
mov (16) BUF_B(0,16)<1> SRC_YB(2,0)<16;4,1>
mov (16) BUF_B(1,0)<1> SRC_YB(4,0)<16;4,1>
mov (16) BUF_B(1,16)<1> SRC_YB(6,0)<16;4,1>
// The second step
mov (16) BUF_B(2,0)<1> BUF_B(0,0)<32;8,4> // Read 2 rows, write 1 row
mov (16) BUF_B(2,16)<1> BUF_B(0,1)<32;8,4>
mov (16) BUF_B(3,0)<1> BUF_B(0,2)<32;8,4>
mov (16) BUF_B(3,16)<1> BUF_B(0,3)<32;8,4>
// Y is now transposed. the result is in BUF_B(2) and BUF_B(3).
// Transpose UV (4x8), right most 2 columns in word
// Use BUF_W(0) as temp buf
// Src U 8x8 and V 8x8 are mixed. (each pix is specified as yx)
// +-----------------------+-----------------------+-----------------------+-----------------------+
// |17 17 16 16 15 15 14 14 13 13 12 12 11 11 10 10 07 07 06 06 05 05 04 04 03 03 02 02 01 01 00 00|
// +-----------------------+-----------------------+-----------------------+-----------------------+
// |37 37 36 36 35 35 34 34 33 33 32 32 31 31 30 30 27 27 26 26 25 25 24 24 23 23 22 22 21 21 20 20|
// +-----------------------+-----------------------+-----------------------+-----------------------+
// |57 57 56 56 55 55 54 54 53 53 52 52 51 51 50 50 47 47 46 46 45 45 44 44 43 43 42 42 41 41 40 40|
// +-----------------------+-----------------------+-----------------------+-----------------------+
// |77 77 76 76 75 75 74 74 73 73 72 72 71 71 70 70 67 67 66 66 65 65 64 64 63 63 62 62 61 61 60 60|
// +-----------------------+-----------------------+-----------------------+-----------------------+
// First step (8) <1>:w <==== <8;2,1>:w
// +-----------------------+-----------------------+-----------------------+-----------------------+
// |71 71 70 70 61 61 60 60 51 51 50 50 41 41 40 40 31 31 30 30 21 21 20 20 11 11 10 10 01 01 00 00|
// +-----------------------+-----------------------+-----------------------+-----------------------+
mov (8) BUF_W(0,0)<1> SRC_UW(0,0)<8;2,1>
mov (8) BUF_W(0,8)<1> SRC_UW(2,0)<8;2,1>
// Second step (16) <1>:w <==== <1;8,2>:w
// +-----------------------+-----------------------+-----------------------+-----------------------+
// |71 71 61 61 51 51 41 41 31 31 21 21 11 11 01 01 70 70 60 60 50 50 40 40 30 30 20 20 10 10 00 00|
// +-----------------------+-----------------------+-----------------------+-----------------------+
mov (16) BUF_W(1,0)<1> BUF_W(0,0)<1;8,2>
// UV are now transposed. the result is in BUF_W(1).
//The first step
//mov (16) BUF_B(0,0)<1> SRC_UW(0,0)<8;2,1> // Read 2 rows, write 1 row
// The second step
//mov (8) SRC_UB(4,0)<1> BUF_B(0,0)<16;8,2> // Read 1 row, write 1 row
//mov (8) SRC_UB(4,8)<1> BUF_B(0,1)<16;8,2> // Read 1 row, write 1 row
// Transpose V (8x8), right most 2 columns
// The first step
//mov (16) BUF_B(0,0)<1> SRC_VB(0,1)<8;2,1> // Read 2 rows, write 1 row
// The second step
//mov (8) SRC_UB(4,16)<1> BUF_B(0,0)<16;8,2> // Read 1 row, write 1 row
//mov (8) SRC_UB(4,24)<1> BUF_B(0,1)<16;8,2> // Read 1 row, write 1 row
// U and V are now transposed. the result is in BUF_B(4).
|