Go to most recent revision | Details | Last modification | View Log | RSS feed
Rev | Author | Line No. | Line |
---|---|---|---|
3769 | Serge | 1 | /* |
2 | * All Video Processing kernels |
||
3 | * Copyright © <2010>, Intel Corporation. |
||
4 | * |
||
5 | * This program is licensed under the terms and conditions of the |
||
6 | * Eclipse Public License (EPL), version 1.0. The full text of the EPL is at |
||
7 | * http://www.opensource.org/licenses/eclipse-1.0.php. |
||
8 | * |
||
9 | */ |
||
10 | |||
11 | #define DI_ENABLE |
||
12 | |||
13 | #include "DNDI.inc" |
||
14 | |||
15 | #ifdef DI_ONLY |
||
16 | #undef nSMPL_RESP_LEN |
||
17 | #define nSMPL_RESP_LEN nSMPL_RESP_LEN_DI // set the number of GRF |
||
18 | #else |
||
19 | #undef nSMPL_RESP_LEN |
||
20 | #define nSMPL_RESP_LEN nSMPL_RESP_LEN_DNDI // set the number of GRF |
||
21 | #endif |
||
22 | |||
23 | #undef nDPW_BLOCK_SIZE_HIST |
||
24 | #define nDPW_BLOCK_SIZE_HIST nBLOCK_WIDTH_4+nBLOCK_HEIGHT_1 // HIST Block Size for Write is 4x2 |
||
25 | #undef nDPW_BLOCK_SIZE_DN |
||
26 | #define nDPW_BLOCK_SIZE_DN nBLOCK_WIDTH_32+nBLOCK_HEIGHT_4 // DN Block Size for Write is 32x4 |
||
27 | |||
28 | ////////////////////////////////////// Run the DN Algorithm /////////////////////////////////////// |
||
29 | #include "DNDI_Command.asm" |
||
30 | |||
31 | ////////////////////////////////////// Rearrange for Internal Planar ////////////////////////////// |
||
32 | //// move the previous frame Y component to internal planar format |
||
33 | //$for (0; |
||
34 | // mov (16) uwDEST_Y(%1,0)<1> ubRESP(nDI_PREV_FRAME_LUMA_OFFSET,%1*16) |
||
35 | //} |
||
36 | //// move the previous frame U,V components to internal planar format |
||
37 | //$for (0; |
||
38 | // mov (8) uwDEST_U(0,%1*8)<1> ubRESP(nDI_PREV_FRAME_CHROMA_OFFSET,%1*16+1)<16;8,2> //U pixels |
||
39 | // mov (8) uwDEST_V(0,%1*8)<1> ubRESP(nDI_PREV_FRAME_CHROMA_OFFSET,%1*16)<16;8,2> //V pixels |
||
40 | //} |
||
41 | //// move the current frame Y component to internal planar format |
||
42 | //$for (0; |
||
43 | // mov (16) uwDEST_Y(%1+4,0)<1> ubRESP(nDI_CURR_FRAME_LUMA_OFFSET,%1*16) |
||
44 | //} |
||
45 | //// move the current frame U,V components to internal planar format |
||
46 | //$for (0; |
||
47 | // mov (8) uwDEST_U(2,%1*8)<1> ubRESP(nDI_CURR_FRAME_CHROMA_OFFSET,%1*16+1)<16;8,2> //U pixels |
||
48 | // mov (8) uwDEST_V(2,%1*8)<1> ubRESP(nDI_CURR_FRAME_CHROMA_OFFSET,%1*16)<16;8,2> //V pixels |
||
49 | //} |
||
50 | |||
51 | ////////////////////////////////////// Save the STMM Data for Next Run ///////////////////////// |
||
52 | // Write STMM to memory |
||
53 | shr (1) rMSGSRC.0<1>:ud wORIX<0;1,0>:w 1:w NODDCLR // X origin / 2 |
||
54 | mov (1) rMSGSRC.1<1>:ud wORIY<0;1,0>:w NODDCLR_NODDCHK // Y origin |
||
55 | mov (1) rMSGSRC.2<1>:ud nDPW_BLOCK_SIZE_STMM:ud NODDCHK // block width and height (8x4) |
||
56 | mov (8) mudMSGHDR_STMM(0)<1> rMSGSRC.0<8;8,1>:ud // message header |
||
57 | mov (8) mudMSGHDR_STMM(1)<1> udRESP(nDI_STMM_OFFSET,0) // Move STMM to MRF |
||
58 | send (8) dNULLREG mMSGHDR_STMM udDUMMY_NULL nDATAPORT_WRITE nDPMW_MSGDSC+nDPMW_MSG_LEN_STMM+nBI_STMM_HISTORY_OUTPUT:ud |
||
59 | |||
60 | ////////////////////////////////////// Save the History Data for Next Run ///////////////////////// |
||
61 | #ifdef DI_ONLY |
||
62 | #else |
||
63 | |||
64 | #include "DI_Hist_Save.asm" |
||
65 | |||
66 | ////////////////////////////////////// Pack and Save the DN Curr Frame for Next Run /////////////// |
||
67 | // check top/bottom field first |
||
68 | cmp.e.f0.0 (1) null<1>:w ubTFLD_FIRST<0;1,0>:ub 1:w |
||
69 | |||
70 | add (4) pCF_Y_OFFSET<1>:uw ubSRC_CF_OFFSET<4;4,1>:ub npDN_YUV:uw |
||
71 | //set the save DN position |
||
72 | shl (1) rMSGSRC.0<1>:ud wORIX<0;1,0>:w 1:w NODDCLR // X origin * 2 |
||
73 | mov (1) rMSGSRC.1<1>:ud wORIY<0;1,0>:w NODDCLR_NODDCHK // Y origin |
||
74 | mov (1) rMSGSRC.2<1>:ud nDPW_BLOCK_SIZE_DN:ud NODDCHK // block width and height (8x4) |
||
75 | mov (8) mudMSGHDR_DN(0)<1> rMSGSRC.0<8;8,1>:ud |
||
76 | |||
77 | |||
78 | (f0.0) jmpi (1) TOP_FIELD_FIRST |
||
79 | |||
80 | BOTTOM_FIELD_FIRST: |
||
81 | //$for (0,0; |
||
82 | // mov (16) r[pCF_Y_OFFSET, %1*32]<2>:ub ubRESP(nDI_CURR_2ND_FIELD_LUMA_OFFSET,%2*16) // 2nd field luma from current frame (line 0,2) |
||
83 | // mov (16) r[pCF_Y_OFFSET, %1+1*32]<2>:ub ubRESP(nDI_CURR_FRAME_LUMA_OFFSET+%2,16) // 1st field luma from current frame (line 1,3) |
||
84 | // mov (8) r[pCF_U_OFFSET, %1*32]<4>:ub ubRESP(nDI_CURR_2ND_FIELD_CHROMA_OFFSET,%2*16+1)<16;8,2> // 2nd field U from current frame (line 0,2) |
||
85 | // mov (8) r[pCF_V_OFFSET, %1*32]<4>:ub ubRESP(nDI_CURR_2ND_FIELD_CHROMA_OFFSET,%2*16)<16;8,2> // 2nd field V from current frame (line 0,2) |
||
86 | // mov (8) r[pCF_U_OFFSET, %1+1*32]<4>:ub ubRESP(nDI_CURR_FRAME_CHROMA_OFFSET+%2,16+1)<16;8,2> // 1st field U from current frame (line 1,3) |
||
87 | // mov (8) r[pCF_V_OFFSET, %1+1*32]<4>:ub ubRESP(nDI_CURR_FRAME_CHROMA_OFFSET+%2,16)<16;8,2> // 1st field U from current frame (line 1,3) |
||
88 | //} |
||
89 | $for (0,0; |
||
90 | mov (16) r[pCF_Y_OFFSET, %1*32]<2>:ub ubRESP(nDI_CURR_2ND_FIELD_LUMA_OFFSET,%2*16) // 2nd field luma from current frame (line 0,2) |
||
91 | mov (16) r[pCF_Y_OFFSET, %1+1*32]<2>:ub ubRESP(nDI_CURR_FRAME_LUMA_OFFSET+%2,16) // 1st field luma from current frame (line 1,3) |
||
92 | } |
||
93 | |||
94 | $for (0,0; |
||
95 | mov (8) r[pCF_U_OFFSET, %1*32]<4>:ub ubRESP(nDI_CURR_2ND_FIELD_CHROMA_OFFSET,%2*16+1)<16;8,2> // 2nd field U from current frame (line 0,2) |
||
96 | mov (8) r[pCF_U_OFFSET, %1+1*32]<4>:ub ubRESP(nDI_CURR_FRAME_CHROMA_OFFSET+%2,16+1)<16;8,2> // 1st field U from current frame (line 1,3) |
||
97 | } |
||
98 | |||
99 | $for (0,0; |
||
100 | mov (8) r[pCF_V_OFFSET, %1*32]<4>:ub ubRESP(nDI_CURR_2ND_FIELD_CHROMA_OFFSET,%2*16)<16;8,2> // 2nd field V from current frame (line 0,2) |
||
101 | mov (8) r[pCF_V_OFFSET, %1+1*32]<4>:ub ubRESP(nDI_CURR_FRAME_CHROMA_OFFSET+%2,16)<16;8,2> // 1st field U from current frame (line 1,3) |
||
102 | } |
||
103 | |||
104 | jmpi (1) SAVE_DN_CURR |
||
105 | |||
106 | TOP_FIELD_FIRST: |
||
107 | //$for (0,0; |
||
108 | // mov (16) r[pCF_Y_OFFSET, %1*32]<2>:ub ubRESP(nDI_CURR_FRAME_LUMA_OFFSET+%2,0) // 1st field luma from current frame (line 0,2) |
||
109 | // mov (16) r[pCF_Y_OFFSET, %1+1*32]<2>:ub ubRESP(nDI_CURR_2ND_FIELD_LUMA_OFFSET,%2*16) // 2nd field luma from current frame (line 1,3) |
||
110 | // mov (8) r[pCF_U_OFFSET, %1*32]<4>:ub ubRESP(nDI_CURR_FRAME_CHROMA_OFFSET+%2,1)<16;8,2> // 1st field U from current frame (line 0,2) |
||
111 | // mov (8) r[pCF_V_OFFSET, %1*32]<4>:ub ubRESP(nDI_CURR_FRAME_CHROMA_OFFSET+%2,0)<16;8,2> // 1st field V from current frame (line 0,2) |
||
112 | // mov (8) r[pCF_U_OFFSET, %1+1*32]<4>:ub ubRESP(nDI_CURR_2ND_FIELD_CHROMA_OFFSET,%2*16+1)<16;8,2> // 2nd field U from current frame (line 1,3) |
||
113 | // mov (8) r[pCF_V_OFFSET, %1+1*32]<4>:ub ubRESP(nDI_CURR_2ND_FIELD_CHROMA_OFFSET,%2*16)<16;8,2> // 2nd field V from current frame (line 1,3) |
||
114 | //} |
||
115 | $for (0,0; |
||
116 | mov (16) r[pCF_Y_OFFSET, %1*32]<2>:ub ubRESP(nDI_CURR_FRAME_LUMA_OFFSET+%2,0) // 1st field luma from current frame (line 0,2) |
||
117 | mov (16) r[pCF_Y_OFFSET, %1+1*32]<2>:ub ubRESP(nDI_CURR_2ND_FIELD_LUMA_OFFSET,%2*16) // 2nd field luma from current frame (line 1,3) |
||
118 | } |
||
119 | $for (0,0; |
||
120 | mov (8) r[pCF_U_OFFSET, %1*32]<4>:ub ubRESP(nDI_CURR_FRAME_CHROMA_OFFSET+%2,1)<16;8,2> // 1st field U from current frame (line 0,2) |
||
121 | mov (8) r[pCF_U_OFFSET, %1+1*32]<4>:ub ubRESP(nDI_CURR_2ND_FIELD_CHROMA_OFFSET,%2*16+1)<16;8,2> // 2nd field U from current frame (line 1,3) |
||
122 | } |
||
123 | $for (0,0; |
||
124 | mov (8) r[pCF_V_OFFSET, %1*32]<4>:ub ubRESP(nDI_CURR_FRAME_CHROMA_OFFSET+%2,0)<16;8,2> // 1st field V from current frame (line 0,2) |
||
125 | mov (8) r[pCF_V_OFFSET, %1+1*32]<4>:ub ubRESP(nDI_CURR_2ND_FIELD_CHROMA_OFFSET,%2*16)<16;8,2> // 2nd field V from current frame (line 1,3) |
||
126 | } |
||
127 | |||
128 | SAVE_DN_CURR: |
||
129 | $for(0; |
||
130 | mov (8) mudMSGHDR_DN(%1+1)<1> udDN_YUV(%1)REGION(8,1) |
||
131 | } |
||
132 | send (8) dNULLREG mMSGHDR_DN udDUMMY_NULL nDATAPORT_WRITE nDPMW_MSGDSC+nDPMW_MSG_LEN_PA_DN_DI+nBI_DESTINATION_YUV:ud |
||
133 | #endif |
||
134 | |||
135 | // Save Processed frames |
||
136 | #include "DI_Save_PA.asm"1>16;8,2>4>16;8,2>4>16;8,2>4>16;8,2>4>2>2>16;8,2>4>16;8,2>4>16;8,2>4>16;8,2>4>2>2>16;8,2>4>16;8,2>4>16;8,2>4>16;8,2>4>2>2>16;8,2>4>16;8,2>4>16;8,2>4>16;8,2>4>2>2>8;8,1>1>1>0;1,0>1>0;1,0>1>4;4,1>1>0;1,0>1>1>8;8,1>1>1>0;1,0>1>0;1,0>1>16;8,2>1>16;8,2>1>1>16;8,2>1>16;8,2>1>1>2010> |
||
137 |