Debian Codegraph @

src/shaders/post_processing/gen7/PA_DNDI_PA.g4a

raw
1 /*
2 * Copyright 2000-2011 Intel Corporation All Rights Reserved
3 *
4 * Permission is hereby granted, free of charge, to any person obtaining a
5 * copy of this software and associated documentation files (the
6 * "Software"), to deal in the Software without restriction, including
7 * without limitation the rights to use, copy, modify, merge, publish,
8 * distribute, sub license, and/or sell copies of the Software, and to
9 * permit persons to whom the Software is furnished to do so, subject to
10 * the following conditions:
11 *
12 * The above copyright notice and this permission notice (including the
13 * next paragraph) shall be included in all copies or substantial portions
14 * of the Software.
15 *
16 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
17 * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
18 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
19 * IN NO EVENT SHALL PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR
20 * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
21 * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
22 * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
23 *
24 * This file was originally licensed under the following license
25 *
26 * Licensed under the Apache License, Version 2.0 (the "License");
27 * you may not use this file except in compliance with the License.
28 * You may obtain a copy of the License at
29 *
30 * http://www.apache.org/licenses/LICENSE-2.0
31 *
32 * Unless required by applicable law or agreed to in writing, software
33 * distributed under the License is distributed on an "AS IS" BASIS,
34 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
35 * See the License for the specific language governing permissions and
36 * limitations under the License.
37 */
38 // 97 // Total instruction count
39 // 1 // Total kernel count
40
41 .kernel PA_DNDI_PA
42 .code
43
44
45
46 // FileName: DNDI_PL_Core.asm
47 // Author: Tatiya, Rupesh
48
49
50
51 // FileName: DNDI_Core.asm
52 // Author: Vivek Kumar
53 // Description: Tasks for DN+DI case (16x4 block)
54
55
56
57
58 // Module name: common.inc
59 //
60 // Common header file for all Video-Processing kernels
61 //
62
63 .default_execution_size (16)
64 .default_register_type :ub
65
66 .reg_count_total 128
67 .reg_count_payload 7
68
69 //========== Common constants ==========
70
71
72 //========== Macros ==========
73
74
75 //Fast Jump, For more details see "Set_Layer_N.asm"
76
77
78 //========== Defines ====================
79
80 //========== Static Parameters (Common To All) ==========
81 //r1
82
83
84 //r2
85
86 // e.g. byte0 byte1 byte2
87 // YUYV 0 1 3
88 // YVYU 0 3 1
89
90 //Color Pipe (IECP) parameters
91
92
93 //ByteCopy
94
95
96 //r4
97
98 // e.g. byte0 byte1 byte2
99 // YUYV 0 1 3
100 // YVYU 0 3 1
101
102
103 //========== Inline parameters (Common To All) ===========
104
105
106 //============== Binding Index Table===========
107 //Common between DNDI and DNUV
108
109
110 //================= Common Message Descriptor =====
111 // Message descriptor for thread spawning
112 // Message Descriptors
113 // = 000 0001 (min message len 1 ) 0,0000 (resp len 0 -add later)
114 // 0000,0000,0000
115 // 0001(Spawn a root thread),0001 (Root thread spawn thread)
116 // = 0x02000011
117 // Thread Spawner Message Descriptor
118
119
120 // Message descriptor for atomic operation add
121 // Message Descriptors
122 // = 000 0110 (min message len 6 ) 0,0000 (resp len 0 -add later)
123 // 1(header present)001,10(typed atomic operation)0(return enabled)0(slot group, low 8 bits),0111 (AOP_Add)
124 // 0000,0000 (Binding table index, added later)
125 // = 0x02000011
126
127 // Atomic Operation Add Message Descriptor
128
129
130 // Message descriptor for dataport media write
131 // Message Descriptors
132 // = 000 0001 (min message len 1 - add later) 00000 (resp len 0)
133 // 1 (header present 1) 0 1010 (media block write) 000000
134 // 00000000 (binding table index - set later)
135 // = 0x020A8000
136
137
138 // Message Length defines
139
140
141 // Response Length defines
142
143
144 // Block Width and Height Size defines
145
146
147 // Extended Message Descriptors
148
149
150 // Common message descriptors:
151
152
153 //===================== Math Function Control ===================================
154
155
156 //============ Message Registers ===============
157 // buf4 starts from r28
158
159
160 //#define mMSGHDR_EOT r43 // Dummy Message Register for EOT
161
162
163 .declare mubMSGPAYLOAD Base=r30 ElementSize=1 SrcRegion=<16;16,1> Type=ub
164 .declare muwMSGPAYLOAD Base=r30 ElementSize=2 SrcRegion=<16;16,1> Type=uw
165 .declare mudMSGPAYLOAD Base=r30 ElementSize=4 SrcRegion=<8;8,1> Type=ud
166 .declare mfMSGPAYLOAD Base=r30 ElementSize=4 SrcRegion=<8;8,1> Type=f
167
168 //=================== End of thread instruction ===========================
169
170
171 //=====================Pointers Used=====================================
172
173
174 //=======================================================================
175
176
177 //r11-r17
178 // Define temp space for any usages
179
180
181 // Common Buffers
182
183
184 // temp space for rotation
185
186 .declare fROBUF Base=r11.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=f
187
188 .declare udROBUF Base=r11.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=ud
189
190 .declare uwROBUF Base=r11.0 ElementSize=2 SrcRegion=<16;16,1> DstRegion=<1> Type=uw
191
192 .declare ubROBUF Base=r11.0 ElementSize=1 SrcRegion=<16;16,1> DstRegion=<1> Type=ub
193
194 .declare ub4ROBUF Base=r11.0 ElementSize=1 SrcRegion=<32;8,4> DstRegion=<4> Type=ub
195
196
197 // End of common.inc
198
199
200 // FileName: DNDI.inc
201 // Author: Vivek Kumar
202 // Description: Include file for DN, DI and DNDI
203 // Inputs: DI_ENABLE, DN_ENABLE, DN_PLANAR, DN_PACKED
204
205
206
207
208 // End of common.inc
209
210
211 //Interface:
212 //Static Parameters:
213 //r1
214
215
216 //====================== Binding table (Explicit To DNDI)=========================================
217
218
219 .declare mudMSGHDR_DNDI Base=r18 ElementSize=4 Type=ud
220 .declare mdMSGHDR_DNDI Base=r18 ElementSize=4 Type=d
221 .declare mwMSGHDR_DNDI Base=r18 ElementSize=2 Type=w
222
223
224 .declare mudMSGHDR_STMM Base=r20 ElementSize=4 Type=ud
225
226
227 .declare mudMSGHDR_HIST Base=r22 ElementSize=4 Type=ud
228
229
230 .declare mudMSGHDR_ENC_STATS Base=r24 ElementSize=4 Type=ud
231 .declare muwMSGHDR_ENC_STATS Base=r24 ElementSize=2 Type=uw
232 .declare mubMSGHDR_ENC_STATS Base=r24 ElementSize=1 Type=ub
233
234
235 .declare mudMSGHDR_DN_OUT Base=r31.0 ElementSize=4 Type=ud
236 .declare mdMSGHDR_DN_OUT Base=r31.0 ElementSize=4 Type=d
237 .declare mubMSGHDR_DN_OUT Base=r31.0 ElementSize=1 Type=ub
238
239
240 .declare mudMSGHDR_UVCOPY Base=r36 ElementSize=4 Type=ud
241 .declare mdMSGHDR_UVCOPY Base=r36 ElementSize=4 Type=d
242 .declare mudMSGHDR_UCOPY Base=r36 ElementSize=4 Type=ud
243 .declare mudMSGHDR_VCOPY Base=r38 ElementSize=4 Type=ud
244
245
246 .declare mudMSGHDR_DI_OUT1 Base=r18.0 ElementSize=4 Type=ud
247 .declare mubMSGHDR_DI_OUT1 Base=r18.0 ElementSize=1 Type=ub
248
249
250 .declare mudMSGHDR_DI_OUT2 Base=r23.0 ElementSize=4 Type=ud
251 .declare mubMSGHDR_DI_OUT2 Base=r23.0 ElementSize=1 Type=ub
252
253 //r45
254 //Use r45 as message header, so no need to "mov" the data.
255
256 .declare mudDN_Y_OUT Base=r45.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=ud
257
258 // Message response (Denoised & DI-ed pixels & statistics); Use buffer 5
259 .declare udDNDI_RESP Base=r46.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=ud
260 .declare uwDNDI_RESP Base=r46.0 ElementSize=2 SrcRegion=<16;16,1> DstRegion=<1> Type=uw
261 .declare ubDNDI_RESP Base=r46.0 ElementSize=1 SrcRegion=<16;16,1> DstRegion=<1> Type=ub
262
263 // Message response (UV Copy); Use buffer 5
264 .declare udDNDI_UV_RESP Base=r58.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=ud
265 .declare ubDNDI_UV_RESP Base=r58.0 ElementSize=1 SrcRegion=<16;16,1> DstRegion=<1> Type=ub
266
267 //Temp GRFs: For 42X to 422 Conversion
268 .declare uwDNDI_UVCOPY_TEMP Base=r10.0 ElementSize=2 SrcRegion=<16;16,1> DstRegion=<1> Type=uw //8 GRFs
269 .declare ubDNDI_UVCOPY_TEMP Base=r10.0 ElementSize=1 SrcRegion=<16;16,1> DstRegion=<1> Type=ub //8 GRFs
270 //---------------------------------------------------------------------------
271 // Message descriptors
272 //---------------------------------------------------------------------------
273 // Extended message descriptor
274 // Message descriptor for sampler read
275 // = 000 0010 (message len 2) 00000 (resp len - set later, 12 or 5 or 11)
276 // 1 (header present 1) 0 11 (SIMD32/64 mode)
277 // 1000 (message type) 0000 (DI state index)
278 // 00000000 (binding table index - set later)
279 // = 0x040b8000
280
281
282 // Attention: The Message Length is The Number of GRFs with Data Only, without the Header
283
284
285 //---------------------------------------------------------------------------
286 // VDI Return Data format
287 //---------------------------------------------------------------------------
288 // Defines for DI enabled
289
290
291 // Defines for DI disabled
292
293
294
295 // FileName: DNDI_Command.asm
296 // Author: Vivek Kumar
297 // Description: Sends a message to the VDI to process one DN (16x8) or DNDI (16x4) block
298
299 // Prepare the DNDI send command
300 mov (8) mudMSGHDR_DNDI(0)<1> r0.0<8;8,1>:ud // message header
301 mov (1) mwMSGHDR_DNDI(1,4)<1> r9.0<0;1,0>:w { NoDDClr } // horizontal origin // Do we need to add offset here? -vK
302 mov (1) mwMSGHDR_DNDI(1,12)<1> r9.1<0;1,0>:w { NoDDChk } // vertical origin // Can these 2 be combined? - vK
303
304 send (8) udDNDI_RESP(0)<1> r18 0x2 0x4CE8003:ud
305
306 // On Gen6, with VDI walker, use the XY pair returned rather than programmed above
307 // VDI_RETURNED_XY is ordered XY in case of walker enables and the same as programmed in case of walker disabled
308 mov (2) r9.0<1>:w uwDNDI_RESP(9,14)<2;2,1> // horizontal/Vertial origin in W.14 and W.15
309
310
311
312 // FileName: DI_STMM_Save.asm
313 // Author: Vivek Kumar
314 // Description: Saves DI STMM Data to statistics surface in case of DI enabled (for 16x4 block)
315
316 // Write STMM to memory
317 mov (8) mudMSGHDR_STMM(0)<1> r0.0<8;8,1>:ud // message header
318 mov (8) mudMSGHDR_STMM(1)<1> udDNDI_RESP(8,0) // Move STMM to MRF
319
320 shr (1) mudMSGHDR_STMM(0,0)<1> r9.0<0;1,0>:w 1:w { NoDDClr } // X origin / 2
321 mov (1) mudMSGHDR_STMM(0,1)<1> r9.1<0;1,0>:w { NoDDClr, NoDDChk } // Y origin
322 mov (1) mudMSGHDR_STMM(0,2)<1> 0x30007:ud { NoDDChk } // block width and height (8x4)
323
324 send (8) null<1>:d r20 0x5 0x40A8021:ud
325
326
327
328 // FileName: DN_Hist_Save.asm
329 // Author: Vivek Kumar
330 // Description: Saves DN history data to statistics surface
331
332 // Write denoise history to memory
333 mov (8) r27<1>:ud r0.0<8;8,1>:ud // message header
334
335 mov (1) mudMSGHDR_HIST(1)<1> udDNDI_RESP(9,0)<0;1,0> // Move denoise history to MRF (4x1)
336
337
338 shr (2) r27.0<1>:ud r9.0<2;2,1>:w 2:w // X,Y origin / 4
339 add (1) r27.0<1>:ud r27.0<0;1,0>:ud r1.12<0;1,0>:uw { NoDDClr } // Add pitch to X origin
340 mov (1) r27.2<1>:ud 0x3:ud { NoDDChk } // block width and height
341
342 mov (8) mudMSGHDR_HIST(0)<1> r27.0<8;8,1>:ud
343 send (8) null<1>:d r22 0x5 0x40A8021:ud
344
345
346
347 // FileName: DNDI_Enc_Stats_Save.asm
348 // Author: Vivek Kumar
349 // Description: Saves Encoder Statistics data to statistics surface in case of DI enabled (for 16x4 block)
350
351 // Write encoder statistics to memory
352 //Currently enable this only on Gen6 validation
353 mov (8) mudMSGHDR_ENC_STATS(1)<1> 0x0:ud // Init payload MRF
354 mov (8) mudMSGHDR_ENC_STATS(0)<1> r0.0<8;8,1>:ud // message header
355
356 shr (1) mudMSGHDR_ENC_STATS(0,0)<1> r9.0<0;1,0>:w 1:w { NoDDClr } //enable the flag after testing on si { NoDDClr } // X origin / 2
357 mul (1) acc0.1<1>:ud r9.1<0;1,0>:w 3:w // Y origin * 3
358 shr (1) mudMSGHDR_ENC_STATS(0,1)<1> acc0.1<0;1,0>:ud 2:w { NoDDClr, NoDDChk } //enable the flag after testing on si { NoDDClr, NoDDChk } // Y origin * 3/4
359 mov (1) mudMSGHDR_ENC_STATS(0,2)<1> 0x20007:ud { NoDDChk } //enable the flag after testing on si { NoDDChk } // block width and height (8x3)
360 add (2) mudMSGHDR_ENC_STATS(0,0)<1> mudMSGHDR_ENC_STATS(0,0)<2;2,1> r1.12<2;2,1>:uw // Add pitch to X,Y origin
361
362
363 //Data block for Encoder Statistics
364 //----------------------------------------------------
365 //| 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 | Bytes
366 //----------------------------------------------------
367 //| BNE | MCNT | FCNT | TCNT | X | X | X | X |
368 //----------------------------------------------------
369 //| DcTpT | SVCM | DcBpT | DcTpB |
370 //----------------------------------------------------
371 //| SHCM | STAD | DcTcB | DcBpB |
372 //----------------------------------------------------
373 mov (1) mudMSGHDR_ENC_STATS(1,0)<1> udDNDI_RESP(9,1)<0;1,0> { NoDDClr } // Move encoder statistics to MRF
374 mov (2) mudMSGHDR_ENC_STATS(1,3)<2> udDNDI_RESP(9,3)<2;2,1> { NoDDClr, NoDDChk } // Move encoder statistics to MRF
375 mov (2) mudMSGHDR_ENC_STATS(1,2)<2> udDNDI_RESP(9,5)<2;2,1> { NoDDChk } // Move encoder statistics to MRF
376
377
378 send (8) null<1>:d r24 0x5 0x40A8021:ud
379
380
381
382 // FileName: DN_Save_PA_16x4.asm
383 // Author: Vivek Kumar
384 // Description: Save one 16x4 blocks of DN output in Packed format for reference
385
386
387 // check top/bottom field first
388 cmp.e.f0.0 (1) null<1>:w r1.28<0;1,0>:ub 1:w
389
390 add (4) a0.4<1>:uw r4.0<4;4,1>:ub 1024:w // Initial Y,U,V offset in YUV422 block; it starts at m14
391
392 mov (8) mudMSGHDR_DN_OUT(0)<1> r0<8;8,1>:ud // message header
393 shl (1) mdMSGHDR_DN_OUT(0,0)<1> r9.0<0;1,0>:w 1:w { NoDDClr } // X origin * 2 (422 output)
394 mov (1) mdMSGHDR_DN_OUT(0,1)<1> r9.1<0;1,0>:w { NoDDClr, NoDDChk } // Y origin
395 mov (1) mudMSGHDR_DN_OUT(0,2)<1> 0x3001F:ud { NoDDChk } // block width and height (32x8)
396
397 (f0.0) jmpi (1) TOP_FIELD_FIRST
398
399 BOTTOM_FIELD_FIRST:
400 mov (16) r[a0.4, 0]<2>:ub ubDNDI_RESP(10,0) { NoDDClr } // 2nd field luma from current frame (line 0,2)
401 mov (16) r[a0.4, 32]<2>:ub ubDNDI_RESP(4,16) { NoDDClr } // 1st field luma from current frame (line 1,3)
402 mov (16) r[a0.4, 64]<2>:ub ubDNDI_RESP(10,16) { NoDDClr } // 2nd field luma from current frame (line 0,2)
403 mov (16) r[a0.4, 96]<2>:ub ubDNDI_RESP(5,16) { NoDDClr } // 1st field luma from current frame (line 1,3)
404 mov (8) r[a0.5, 0]<4>:ub ubDNDI_RESP(11,1)<16;8,2> { NoDDClr, NoDDChk } // 2nd field U from current frame (line 0,2)
405 mov (8) r[a0.5, 32]<4>:ub ubDNDI_RESP(6,17)<16;8,2> { NoDDClr, NoDDChk } // 1st field U from current frame (line 1,3)
406 mov (8) r[a0.5, 64]<4>:ub ubDNDI_RESP(11,17)<16;8,2> { NoDDClr, NoDDChk } // 2nd field U from current frame (line 0,2)
407 mov (8) r[a0.5, 96]<4>:ub ubDNDI_RESP(7,17)<16;8,2> { NoDDClr, NoDDChk } // 1st field U from current frame (line 1,3)
408 mov (8) r[a0.6, 0]<4>:ub ubDNDI_RESP(11,0)<16;8,2> { NoDDChk } // 2nd field V from current frame (line 0,2)
409 mov (8) r[a0.6, 32]<4>:ub ubDNDI_RESP(6,16)<16;8,2> { NoDDChk } // 1st field U from current frame (line 1,3)
410 mov (8) r[a0.6, 64]<4>:ub ubDNDI_RESP(11,16)<16;8,2> { NoDDChk } // 2nd field V from current frame (line 0,2)
411 mov (8) r[a0.6, 96]<4>:ub ubDNDI_RESP(7,16)<16;8,2> { NoDDChk } // 1st field U from current frame (line 1,3)
412 jmpi (1) SAVE_DN_CURR
413
414 TOP_FIELD_FIRST:
415 mov (16) r[a0.4, 0]<2>:ub ubDNDI_RESP(4,0) { NoDDClr } // 1st field luma from current frame (line 0,2)
416 mov (16) r[a0.4, 32]<2>:ub ubDNDI_RESP(10,0) { NoDDClr } // 2nd field luma from current frame (line 1,3)
417 mov (16) r[a0.4, 64]<2>:ub ubDNDI_RESP(5,0) { NoDDClr } // 1st field luma from current frame (line 0,2)
418 mov (16) r[a0.4, 96]<2>:ub ubDNDI_RESP(10,16) { NoDDClr } // 2nd field luma from current frame (line 1,3)
419 mov (8) r[a0.5, 0]<4>:ub ubDNDI_RESP(6,1)<16;8,2> { NoDDClr, NoDDChk } // 1st field U from current frame (line 0,2)
420 mov (8) r[a0.5, 32]<4>:ub ubDNDI_RESP(11,1)<16;8,2> { NoDDClr, NoDDChk } // 2nd field U from current frame (line 1,3)
421 mov (8) r[a0.5, 64]<4>:ub ubDNDI_RESP(7,1)<16;8,2> { NoDDClr, NoDDChk } // 1st field U from current frame (line 0,2)
422 mov (8) r[a0.5, 96]<4>:ub ubDNDI_RESP(11,17)<16;8,2> { NoDDClr, NoDDChk } // 2nd field U from current frame (line 1,3)
423 mov (8) r[a0.6, 0]<4>:ub ubDNDI_RESP(6,0)<16;8,2> { NoDDChk } // 1st field V from current frame (line 0,2)
424 mov (8) r[a0.6, 32]<4>:ub ubDNDI_RESP(11,0)<16;8,2> { NoDDChk } // 2nd field V from current frame (line 1,3)
425 mov (8) r[a0.6, 64]<4>:ub ubDNDI_RESP(7,0)<16;8,2> { NoDDChk } // 1st field V from current frame (line 0,2)
426 mov (8) r[a0.6, 96]<4>:ub ubDNDI_RESP(11,16)<16;8,2> { NoDDChk } // 2nd field V from current frame (line 1,3)
427
428 SAVE_DN_CURR:
429 //send out data through data port
430 send (8) null<1>:d r31.0 0x5 0xA0A8018:ud
431
432
433
434 // FileName: DI_Save_PA_16x4.asm
435 // Author: Vivek Kumar
436 // Description: Save two 16x4 blocks of DI output in Packed format
437
438
439 add (4) a0.4<1>:uw r2.28<4;4,1>:ub 608:w // Initial Y,U,V offset in YUV422 block; it starts at m20
440
441 mov (8) r27.0<1>:ud r0.0<8;8,1>:ud
442 shl (1) r27.0<1>:d r9.0<0;1,0>:w 1:w { NoDDClr } // H. block origin need to be doubled
443 mov (1) r27.1<1>:d r9.1<0;1,0>:w { NoDDClr, NoDDChk } // Block origin
444 mov (1) r27.2<1>:ud 0x3001F:ud { NoDDChk } // Block width and height (32x8)
445
446 //prepare the message headers
447 mov (8) r18.0<1>:ud r27<8;8,1>:ud
448 mov (8) r23.0<1>:ud r27<8;8,1>:ud
449
450 // Pack 2nd field Y
451 mov (16) r[a0.4, 0]<2> ubDNDI_RESP(0,0) { NoDDClr }
452 mov (16) r[a0.4, 32]<2> ubDNDI_RESP(0,16) { NoDDClr }
453 mov (16) r[a0.4, 64]<2> ubDNDI_RESP(0,32) { NoDDClr }
454 mov (16) r[a0.4, 96]<2> ubDNDI_RESP(0,48) { NoDDClr }
455 // Pack 2nd field U
456 mov (8) r[a0.5, 0]<4> ubDNDI_RESP(2,1)<16;8,2> { NoDDClr, NoDDChk } //U pixels
457 mov (8) r[a0.5, 32]<4> ubDNDI_RESP(2,17)<16;8,2> { NoDDClr, NoDDChk } //U pixels
458 mov (8) r[a0.5, 64]<4> ubDNDI_RESP(2,33)<16;8,2> { NoDDClr, NoDDChk } //U pixels
459 mov (8) r[a0.5, 96]<4> ubDNDI_RESP(2,49)<16;8,2> { NoDDClr, NoDDChk } //U pixels
460 // Pack 2nd field V
461 mov (8) r[a0.6, 0]<4> ubDNDI_RESP(2,0)<16;8,2> { NoDDChk } //Vpixels
462 mov (8) r[a0.6, 32]<4> ubDNDI_RESP(2,16)<16;8,2> { NoDDChk } //Vpixels
463 mov (8) r[a0.6, 64]<4> ubDNDI_RESP(2,32)<16;8,2> { NoDDChk } //Vpixels
464 mov (8) r[a0.6, 96]<4> ubDNDI_RESP(2,48)<16;8,2> { NoDDChk } //Vpixels
465
466 // Pack 1st field Y
467 mov (16) r[a0.4, 160]<2> ubDNDI_RESP(4,0) { NoDDClr }
468 mov (16) r[a0.4, 192]<2> ubDNDI_RESP(4,16) { NoDDClr }
469 mov (16) r[a0.4, 224]<2> ubDNDI_RESP(4,32) { NoDDClr }
470 mov (16) r[a0.4, 256]<2> ubDNDI_RESP(4,48) { NoDDClr }
471 // Pack 1st field U
472 mov (8) r[a0.5, 160]<4> ubDNDI_RESP(6,1)<16;8,2> { NoDDClr, NoDDChk } //U pixels
473 mov (8) r[a0.5, 192]<4> ubDNDI_RESP(6,17)<16;8,2> { NoDDClr, NoDDChk } //U pixels
474 mov (8) r[a0.5, 224]<4> ubDNDI_RESP(6,33)<16;8,2> { NoDDClr, NoDDChk } //U pixels
475 mov (8) r[a0.5, 256]<4> ubDNDI_RESP(6,49)<16;8,2> { NoDDClr, NoDDChk } //U pixels
476 // Pack 1st field V
477 mov (8) r[a0.6, 160]<4> ubDNDI_RESP(6,0)<16;8,2> { NoDDChk } //Vpixels
478 mov (8) r[a0.6, 192]<4> ubDNDI_RESP(6,16)<16;8,2> { NoDDChk } //Vpixels
479 mov (8) r[a0.6, 224]<4> ubDNDI_RESP(6,32)<16;8,2> { NoDDChk } //Vpixels
480 mov (8) r[a0.6, 256]<4> ubDNDI_RESP(6,48)<16;8,2> { NoDDChk } //Vpixels
481
482 //save the previous frame
483 send (8) null<1>:d r18.0 0x5 0xA0A801B:ud
484
485 //save the current frame
486 send (8) null<1>:d r23.0 0x5 0xA0A801E:ud
487
488
489
490 //End of Thread message
491
492 mov (8) r127<1>:ud r0.0<8;8,1>:ud
493 send (1) null<1>:d r127 0x27 0x02000010
494
495
496 .end_code
497 .end_kernel