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
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
|
/*
* All inter-prediction macroblock 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.
*
*/
// Kernel name: AVCMCInter.asm
#ifdef INTERLABEL
#undef INTERLABEL
#endif
#if defined(MBAFF)
// < MBaff_Motion >
#define INTERLABEL(x) x##_##MBF
#elif defined(FIELD)
// < FieldMB_Motion >
#define INTERLABEL(x) x##_##FLD
#else // FRAME
// < FrameMB_Motion >
#define INTERLABEL(x) x##_##FRM
#endif
//
// Decoding an inter-prediction macroblock (conditional compile)
// -DMBAFF : MBAff picture MB
// -DFRAME : Frame picture MB
// -DFIELD : Field picture MB
// -DMBAFF -DMONO : MBAff mono picture MB
// -DFRAME -DMONO : Frame mono picture MB
// -DFIELD -DMONO : Field mono picture MB
//#if !defined(__AVCMCInter__) // Make sure this is only included once
//#define __AVCMCInter__
// TODO: header files need to be in sync with intra prediction
#include "header.inc"
#include "inter_Header.inc"
// TODO: Kernel names for mono cases
#if defined(MBAFF)
.kernel MBAff_Motion
MBAFF_MB:
#elif defined(FIELD)
.kernel FieldMB_Motion
FIELD_MB:
#else // Frame
.kernel FrameMB_Motion
FRAME_MB:
#endif
#ifdef _DEBUG
// WA for FULSIM so we'll know which kernel is being debugged
#if defined(MBAFF)
mov (1) acc0:ud 0x0aaa55a5:ud
#elif defined(FIELD)
mov (1) acc0:ud 0x0baa55a5:ud
#else // Frame
mov (1) acc0:ud 0x0caa55a5:ud
#endif
#endif
#ifdef SW_SCOREBOARD
CALL(scoreboard_start_inter,1)
#endif
mov (8) gMSGSRC<1>:ud r0.0<8;8,1>:ud // Initialize message header payload with R0
and (1) gwMBTYPE<1> gMBTYPE:ub nMBTYPE_MASK:w // MB type
shl (2) gX<1>:w gORIX<2;2,1>:ub 4:w // Convert MB origin to pixel unit
// #include "process_inter16x16.asm" // Handle B_L0_16x16 case with zero MVs and weighted pred off.
// In the case of B_L0_16x16 with zero MVs and weighted pred off, the kernel jumps to INTERLABEL(EXIT_LOOP).
INTERLABEL(INIT_MBPARA):
#include "initialize_MBPara.asm"
//========================= BEGIN - LOOP_SUBMB ===========================
mov (1) gLOOP_SUBMB:uw 0:uw // 0, 2, 4, 6
INTERLABEL(LOOP_SUBMB):
//========================== BEGIN - LOOP_DIR ============================
// Prediction flag (gPREDFLAG - 0:Pred_L0, 1:Pred_L1, 2:BiPred)
asr (1) gPREDFLAG:w gSUBMB_MODE:ub gLOOP_SUBMB:uw
mov (1) gLOOP_DIR:uw 1:uw // 1, 0
and (1) gPREDFLAG:w gPREDFLAG:w 0x3:w
INTERLABEL(LOOP_DIR):
cmp.e.f0.0 (1) null:w gLOOP_DIR:w gPREDFLAG:w
(f0.0) jmpi INTERLABEL(LOOP_DIR_CONTINUE)
// Get binding table index
// & reference picture parity (gREFPARITY - 0:top, 0x100:bottom, x:frame)
// & address of interpolation result
cmp.e.f0.1 (1) null:w gLOOP_DIR:w 1:w
(f0.1) mov (1) gpINTP:ud nOFFSET_INTP0:ud {NoDDClr} //
(f0.1) and (1) gBIDX:w r[pBIDX]:ub 0x7f:w {NoDDChk} //
(-f0.1) mov (1) gpINTP:ud nOFFSET_INTP1:ud {NoDDClr} //
(-f0.1) and (1) gBIDX:w r[pBIDX,4]:ub 0x7f:w {NoDDChk} //
#if defined(MBAFF) || defined(FIELD)
(f0.1) and (1) gREFPARITY:w r[pBIDX]:ub 0x80:w
(-f0.1) and (1) gREFPARITY:w r[pBIDX,4]:ub 0x80:w
shl (1) gREFPARITY:w gREFPARITY<0;1,0>:w 1:w
#endif
// Sub MB shape
asr (1) gSHAPETEMP:w gSUBMB_SHAPE:ub gLOOP_SUBMB:w
// Chroma MV adjustment & Set message descriptor for frame/field read
#if defined(MBAFF)
#include "chromaMVAdjust.asm"
and.nz.f0.0 (1) null:uw gFIELDMBFLAG:ub nFIELDMB_MASK:uw
(f0.0) add (1) gD0:ud gBIDX:uw nDWBRMSGDSC_SC_TF:ud
(-f0.0) add (1) gMSGDSC_R:ud gBIDX:uw nDWBRMSGDSC_SC:ud
(f0.0) add (1) gMSGDSC_R:ud gD0:ud gREFPARITY:uw
#elif defined(FIELD)
#include "chromaMVAdjust.asm"
add (1) gMSGDSC_R:ud gBIDX:uw nDWBRMSGDSC_SC_TF:ud
add (1) gMSGDSC_R:ud gMSGDSC_R:ud gREFPARITY:uw
#else // FRAME
add (1) gMSGDSC_R:ud gBIDX:uw nDWBRMSGDSC_SC:ud
#endif
and.nz.f0.1 (1) null:w gSHAPETEMP:w 3:w
(f0.1) jmpi INTERLABEL(PROCESS4x4)
//======================== BEGIN - PROCESS 8x8 ===========================
// Reference block load
#include "loadRef_Y_16x13.asm"
#ifndef MONO
#if defined(MBAFF) || defined(FIELD)
add (1) r[pMV,2]:w r[pMV,2]:w gCHRMVADJ:w
#endif
#include "loadRef_C_10x5.asm"
#endif
// Interpolation
//CALL_INTER(INTERLABEL(Interpolate_Y_8x8_Func), 1)
#include "interpolate_Y_8x8.asm"
#ifndef MONO
//CALL_INTER(INTERLABEL(Interpolate_C_4x4_Func), 1)
#include "interpolate_C_4x4.asm"
#endif
jmpi INTERLABEL(ROUND_SHIFT_C)
//========================= END - PROCESS 8x8 ============================
//======================== BEGIN - LOOP_SUBMBPT ==========================
INTERLABEL(PROCESS4x4):
mov (1) gLOOP_SUBMBPT:uw 4:uw // 4, 3, 2, 1
INTERLABEL(LOOP_SUBMBPT):
// Reference block load
#include "loadRef_Y_16x9.asm"
#ifndef MONO
#if defined(MBAFF) || defined(FIELD)
add (1) r[pMV,2]:w r[pMV,2]:w gCHRMVADJ:w
#endif
#include "loadRef_C_6x3.asm"
#endif
// Interpolation
#include "interpolate_Y_4x4.asm"
#ifndef MONO
#include "interpolate_C_2x2.asm"
#endif
cmp.e.f0.0 (1) null:w gLOOP_SUBMBPT:uw 3:w
add.z.f0.1 (1) gLOOP_SUBMBPT:uw gLOOP_SUBMBPT:uw -1:w
add (1) pMV:w pMV:w 8:w
(-f0.0) add (1) gpINTP:ud gpINTP:ud 0x00080008:ud // 8 & 8
(f0.0) add (1) gpINTP:ud gpINTP:ud 0x00180038:ud // 24 & 56
(-f0.1) jmpi INTERLABEL(LOOP_SUBMBPT)
cmp.e.f0.1 null:w gLOOP_DIR:w 1:w
add (1) pMV:w pMV:w -32:w
(f0.1) mov (1) gpINTP:ud nOFFSET_INTP0:ud
(-f0.1) mov (1) gpINTP:ud nOFFSET_INTP1:ud
mov (1) pRESULT:uw gpINTPC:uw
//========================= END - LOOP_SUBMBPT ===========================
INTERLABEL(ROUND_SHIFT_C):
#ifndef MONO
#include "roundShift_C_4x4.asm"
#endif
INTERLABEL(LOOP_DIR_CONTINUE):
add.nz.f0.1 (1) gLOOP_DIR:uw gLOOP_DIR:uw -1:w
add (1) pMV:w pMV:w 4:w
(-f0.1) jmpi INTERLABEL(LOOP_DIR)
//=========================== END - LOOP_DIR =============================
INTERLABEL(Weighted_Prediction):
#include "weightedPred.asm"
and.z.f0.1 (16) null<1>:w gLOOP_SUBMB<0;1,0>:uw 2:w
#include "recon_Y_8x8.asm"
#ifndef MONO
#include "recon_C_4x4.asm"
(-f0.1) add (1) pERRORC:w pERRORC:w 48:w
#endif
cmp.e.f0.1 (1) null:w gLOOP_SUBMB:uw 6:w
add (1) gLOOP_SUBMB:uw gLOOP_SUBMB:uw 2:w
add (1) pWGT_BIDX:ud pWGT_BIDX:ud 0x00100001:ud // 12 & 1
add (1) pMV:w pMV:w gMVSTEP:w
(-f0.1) jmpi INTERLABEL(LOOP_SUBMB)
//========================== END - LOOP_SUBMB ============================
INTERLABEL(EXIT_LOOP):
#include "writeRecon_YC.asm"
#ifdef SW_SCOREBOARD
wait n0:ud // Now wait for scoreboard to response
#include "Soreboard_update.asm" // scorboard update function
#else
// Check for write commit first if SW scoreboard is disabled
mov (1) gREG_WRITE_COMMIT_Y<1>:ud gREG_WRITE_COMMIT_Y<0;1,0>:ud // Make sure Y write is committed
mov (1) gREG_WRITE_COMMIT_UV<1>:ud gREG_WRITE_COMMIT_UV<0;1,0>:ud // Make sure U/V write is committed
#endif
// Terminate the thread
//
END_THREAD
//#include "Interpolate_Y_8x8_Func.asm"
//#include "Interpolate_C_4x4_Func.asm"
//#include "WeightedPred_Y_Func.asm"
//#include "WeightedPred_C_Func.asm"
.end_code
.end_kernel
//#endif // !defined(__AVCMCInter__)
|
