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
|
/*
* All Video Processing kernels
* 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: PL8x4_Save_NV12.asm
//
// Save entire current planar frame data block of size 16x8
//---------------------------------------------------------------
// Symbols needed to be defined before including this module
//
// DWORD_ALIGNED_DEST: only if DEST_Y, DEST_U, DEST_V data are DWord aligned
// ORIX:
//---------------------------------------------------------------
#include "PL8x4_Save_NV12.inc"
mov (8) mMSGHDR<1>:ud rMSGSRC<8;8,1>:ud
#if !defined(SAVE_UV_ONLY)
// Save current planar frame Y block data (16x8) -------------------------------
mov (2) mMSGHDR.0<1>:d wORIX<2;2,1>:w // Block origin
mov (1) mMSGHDR.2<1>:ud nDPW_BLOCK_SIZE_Y:ud // Block width and height (16x8)
#endif
//Use the mask to determine which pixels shouldn't be over-written
and (1) acc0<1>:ud udBLOCK_MASK<0;1,0>:ud 0x00FFFFFF:ud
cmp.ge.f0.0 (1) dNULLREG acc0<0;1,0>:ud 0x00FFFFFF:ud //Check if all pixels in the block need to be modified
(f0.0) jmpi WritePlanarToDataPort
//If mask is not all 1's, then load the entire 16x8 block
//so that only those bytes may be modified that need to be (using the mask)
send (8) udSRC_Y(0)<1> mMSGHDR udDUMMY_NULL nDATAPORT_READ nDPMR_MSGDSC+nDPR_MSG_SIZE_Y+nBI_DESTINATION_Y:ud //16x8
asr (1) rMSGSRC.1<1>:ud wORIY<0;1,0>:w 1:w { NoDDClr } // U/V block origin should be half of Y's
mov (1) rMSGSRC.2<1>:ud nDPW_BLOCK_SIZE_UV:ud { NoDDChk } // Block width and height (16x4)
mov (8) mMSGHDR<1>:ud rMSGSRC<8;8,1>:ud //move message desrcptor to the message header
send (8) udSRC_U(0)<1> mMSGHDR udDUMMY_NULL nDATAPORT_READ nDPMR_MSGDSC+nDPR_MSG_SIZE_UV+nBI_DESTINATION_UV:ud
//Restore the origin information
mov (2) rMSGSRC.0<1>:ud wORIX<2;2,1>:w // Block origin
mov (1) rMSGSRC.2<1>:ud nDPW_BLOCK_SIZE_Y:ud // Block width and height (16x8)
mov (8) mMSGHDR<1>:ud rMSGSRC<8;8,1>:ud //move message desrcptor to the message header
//Merge the data
mov (1) f0.1:uw ubBLOCK_MASK_V:ub //Load the mask on flag reg
(f0.1) mov (8) rMASK_TEMP<1>:uw uwBLOCK_MASK_H:uw
(-f0.1) mov (8) rMASK_TEMP<1>:uw 0:uw
//convert the mask from 16bits to 8bits by selecting every other bit
mov (1) udMASK_TEMP1(0,0)<1> 0x00040001:ud
mov (1) udMASK_TEMP1(0,1)<1> 0x00400010:ud
mov (1) udMASK_TEMP1(0,2)<1> 0x04000100:ud
mov (1) udMASK_TEMP1(0,3)<1> 0x40001000:ud
//merge the loaded block with the current block
$for(0,0; <nY_NUM_OF_ROWS; 2,1) {
mov (1) f0.1:uw uwMASK_TEMP(0, %1)<0;1,0>
(-f0.1) mov (16) ubDEST_Y(0,%1*32)<2> ubSRC_Y(0,%1*16)
and.nz.f0.1 (8) wNULLREG uwMASK_TEMP(0,%1)<0;1,0> uwMASK_TEMP1(0,0) //change the mask by selecting every other bit
(-f0.1) mov (8) ubDEST_U(0, %2*16)<2> ub2SRC_U(0, %1*8)<16;8,2>
(-f0.1) mov (8) ubDEST_V(0, %2*16)<2> ub2SRC_U(0, %1*8+1)<16;8,2>
mov (1) f0.1:uw uwMASK_TEMP(0,1+%1)<0;1,0>
(-f0.1) mov (16) ubDEST_Y(0, (1+%1)*32)<2> ubSRC_Y(0, (1+%1)*16)
}
WritePlanarToDataPort:
#if !defined(SAVE_UV_ONLY)
$for(0,0; <nY_NUM_OF_ROWS; 2,1) {
mov (16) mubMSGPAYLOAD(%2,0)<1> ub2DEST_Y(%1)REGION(16,2)
mov (16) mubMSGPAYLOAD(%2,16)<1> ub2DEST_Y(%1+1)REGION(16,2)
}
send (8) dNULLREG mMSGHDR udDUMMY_NULL nDATAPORT_WRITE nDPMW_MSGDSC+nDPW_MSG_SIZE_Y+nBI_DESTINATION_Y:ud
#endif
//** Save 8x4 packed U and V -----------------------------------------------------
// we could write directly wORIX to mMSGHDR and then execute asr on it, that way we could
// avoid using rMSGSRC as a buffer and have one command less in code, but it is unknown whether
//it is possible to do asr on mMSGHDR so we use rMSGSRC.
mov (2) rMSGSRC.0<1>:d wORIX<2;2,1>:w // Block origin
asr (1) rMSGSRC.1<1>:d rMSGSRC.1<0;1,0>:d 1:w // U/V block origin should be half of Y's
mov (1) rMSGSRC.2<1>:ud nDPW_BLOCK_SIZE_UV:ud // U/V block width and height (16x4)
mov (8) mMSGHDR<1>:ud rMSGSRC<8;8,1>:ud
$for(0,0; <nY_NUM_OF_ROWS;4,1) {
mov (16) mubMSGPAYLOAD(%2,0)<2> ub2DEST_U(%2)REGION(16,2)
mov (16) mubMSGPAYLOAD(%2,1)<2> ub2DEST_V(%2)REGION(16,2)
}
send (8) dNULLREG mMSGHDR udDUMMY_NULL nDATAPORT_WRITE nDPMW_MSGDSC+nDPW_MSG_SIZE_UV+nBI_DESTINATION_UV:ud
// End of PL8x4_Save_NV12
|
