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
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
|
/*
* 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: saveNV12_16x4T.asm
//
// Transpose 16x4 to 4x16 YNV12 data and write to memory
//
//----------------------------------------------------------------
// Symbols need to be defined before including this module
//
// Left MB region:
// PREV_MB_YB: Base=rxx ElementSize=1 SrcRegion=REGION(16,1) Type=ub
// PREV_MB_UW: Base=ryy ElementSize=2 SrcRegion=REGION(8,1) Type=uw
// Binding table index:
// BI_SRC_Y: Binding table index of Y surface
// BI_SRC_UV: Binding table index of UV surface (NV12)
//
// Temp buffer:
// BUF_B: BUF_B Base=rxx ElementSize=1 SrcRegion=REGION(16,1) Type=ub
// BUF_W: BUF_W Base=rxx ElementSize=1 SrcRegion=REGION(8,1) Type=uw
//
//
#if defined(_DEBUG)
mov (1) EntrySignatureC:w 0xDDD6:w
#endif
mov (2) MSGSRC.0<1>:ud ORIX_LEFT<2;2,1>:w // Block origin
mov (1) MSGSRC.2<1>:ud 0x000F0003:ud // 4x16
// Transpose Y, save them to MRFs
// 16x4 Y src in GRF (each pix is specified as yx)
// +-----------------------+-----------------------+-----------------------+-----------------------+
// |f1 e1 d1 c1 b1 a1 91 81 71 61 51 41 31 21 11 01 f0 e0 d0 c0 b0 a0 90 80 70 60 50 40 30 20 10 00|
// +-----------------------+-----------------------+-----------------------+-----------------------+
// |f3 e3 d3 c3 b3 a3 93 83 73 63 53 43 33 23 13 03 f2 e2 d2 c2 b2 a2 92 82 72 62 52 42 32 22 12 02|
// +-----------------------+-----------------------+-----------------------+-----------------------+
//
// First step (16) <1> <=== <16;4,1>
// +-----------------------+-----------------------+-----------------------+-----------------------+
// |73 63 53 43 72 62 52 42 71 61 51 41 70 60 50 40 33 23 13 03 32 22 12 02 31 21 11 01 30 20 10 00|
// +-----------------------+-----------------------+-----------------------+-----------------------+
// |f3 e3 d3 c3 f2 e2 d2 c2 f1 e1 d1 c1 f0 e0 d0 c0 b3 a3 93 83 b2 a2 92 82 b1 a1 91 81 b0 a0 90 80|
// +-----------------------+-----------------------+-----------------------+-----------------------+
// The first step
mov (16) BUF_B(0,0)<1> PREV_MB_YB(0,0)<16;4,1>
mov (16) BUF_B(0,16)<1> PREV_MB_YB(0,4)<16;4,1>
mov (16) BUF_B(1,0)<1> PREV_MB_YB(0,8)<16;4,1>
mov (16) BUF_B(1,16)<1> PREV_MB_YB(0,12)<16;4,1>
//
// Second step (16) <1> <=== <1;4,4>
// +-----------------------+-----------------------+-----------------------+-----------------------+
// |73 72 71 70 63 62 61 60 53 52 51 50 43 42 41 40 33 32 31 30 23 22 21 20 13 12 11 10 03 02 01 00|
// +-----------------------+-----------------------+-----------------------+-----------------------+
// |f3 f2 f1 f0 e3 e2 e1 e0 d3 d2 d1 d0 c3 c2 c1 c0 b3 b2 b1 b0 a3 a2 a1 a0 93 92 91 90 83 82 81 80|
// +-----------------------+-----------------------+-----------------------+-----------------------+
//
// The second step
// mov (16) MSGPAYLOADB(0,0)<1> BUF_B(0,0)<32;8,4> // Read 2 rows, write 1 row
// mov (16) MSGPAYLOADB(0,16)<1> BUF_B(0,1)<32;8,4>
// mov (16) MSGPAYLOADB(1,0)<1> BUF_B(0,2)<32;8,4>
// mov (16) MSGPAYLOADB(1,16)<1> BUF_B(0,3)<32;8,4>
mov (16) MSGPAYLOADB(0,0)<1> BUF_B(0,0)<1;4,4>
mov (16) MSGPAYLOADB(0,16)<1> BUF_B(0,16)<1;4,4>
mov (16) MSGPAYLOADB(1,0)<1> BUF_B(1,0)<1;4,4>
mov (16) MSGPAYLOADB(1,16)<1> BUF_B(1,16)<1;4,4>
// Transposed Y in 4x16 is ready for writting to dataport.
//
send (8) NULLREG MSGHDR MSGSRC<8;8,1>:ud DAPWRITE MSG_LEN(2)+DWBWMSGDSC+BI_DEST_Y // Write 2 GRFs
/////////////////////////////////////////////////////////////////////////////////////////////////////
// Transpose U/V, save them to MRFs in NV12 format
asr (1) MSGSRC.1:ud MSGSRC.1:ud 1:w // NV12 U+V block origin y = half of Y comp
mov (1) MSGSRC.2<1>:ud 0x00070003:ud // NV12 U+V block width and height (4x8)
// 16x2 UV src in GRF (each pix is specified as yx)
// +-----------------------+-----------------------+-----------------------+-----------------------+
// |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|
// +-----------------------+-----------------------+-----------------------+-----------------------+
//
// First step (8) <1> <=== <8;4,1>:w
// +-----------------------+-----------------------+-----------------------+-----------------------+
// |71 71 61 61 51 51 41 41 70 70 60 60 50 50 40 40 31 31 21 21 11 11 01 01 30 30 20 20 10 10 00 00|
// +-----------------------+-----------------------+-----------------------+-----------------------+
mov (8) BUF_W(0,0)<1> PREV_MB_UW(0,0)<8;4,1>
mov (8) BUF_W(0,8)<1> PREV_MB_UW(0,4)<8;4,1>
// Second step (8) <1> <=== <1;2,4>
// +-----------------------+-----------------------+-----------------------+-----------------------+
// |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) MSGPAYLOADW(0,0)<1> BUF_W(0,0)<1;2,4>
mov (8) MSGPAYLOADW(0,8)<1> BUF_W(0,8)<1;2,4>
// Transposed U+V in NV12 in 4x8 is ready for writting to dataport.
send (8) NULLREG MSGHDR MSGSRC<8;8,1>:ud DAPWRITE MSG_LEN(1)+DWBWMSGDSC+BI_DEST_UV // Write 1 GRF
|