Debian Codegraph @

src/shaders/post_processing/gen7/Save_AVS_RGBX.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 * Authors:
38 * Zhao Yakui <[email protected]>
39 */
40
41
42
43 // Module name: common.inc
44 //
45 // Common header file for all Video-Processing kernels
46 //
47
48 .default_execution_size (16)
49 .default_register_type :ub
50
51 .reg_count_total 128
52 .reg_count_payload 7
53
54 //========== Common constants ==========
55
56
57 //========== Macros ==========
58
59
60 //Fast Jump, For more details see "Set_Layer_N.asm"
61
62
63 //========== Defines ====================
64
65 //========== Static Parameters (Common To All) ==========
66 //r1
67
68
69 //r2
70
71 // e.g. byte0 byte1 byte2
72 // YUYV 0 1 3
73 // YVYU 0 3 1
74
75 //Color Pipe (IECP) parameters
76
77
78 //ByteCopy
79
80
81 //r4
82
83 // e.g. byte0 byte1 byte2
84 // YUYV 0 1 3
85 // YVYU 0 3 1
86
87
88 //========== Inline parameters (Common To All) ===========
89
90
91 //============== Binding Index Table===========
92 //Common between DNDI and DNUV
93
94
95 //================= Common Message Descriptor =====
96 // Message descriptor for thread spawning
97 // Message Descriptors
98 // = 000 0001 (min message len 1 ) 0,0000 (resp len 0 -add later)
99 // 0000,0000,0000
100 // 0001(Spawn a root thread),0001 (Root thread spawn thread)
101 // = 0x02000011
102 // Thread Spawner Message Descriptor
103
104
105 // Message descriptor for atomic operation add
106 // Message Descriptors
107 // = 000 0110 (min message len 6 ) 0,0000 (resp len 0 -add later)
108 // 1(header present)001,10(typed atomic operation)0(return enabled)0(slot group, low 8 bits),0111 (AOP_Add)
109 // 0000,0000 (Binding table index, added later)
110 // = 0x02000011
111
112 // Atomic Operation Add Message Descriptor
113
114
115 // Message descriptor for dataport media write
116 // Message Descriptors
117 // = 000 0001 (min message len 1 - add later) 00000 (resp len 0)
118 // 1 (header present 1) 0 1010 (media block write) 000000
119 // 00000000 (binding table index - set later)
120 // = 0x020A8000
121
122
123 // Message Length defines
124
125
126 // Response Length defines
127
128
129 // Block Width and Height Size defines
130
131
132 // Extended Message Descriptors
133
134
135 // Common message descriptors:
136
137
138 //===================== Math Function Control ===================================
139
140
141 //============ Message Registers ===============
142 // buf4 starts from r28
143
144
145 //#define mMSGHDR_EOT r43 // Dummy Message Register for EOT
146
147
148 .declare mubMSGPAYLOAD Base=r30 ElementSize=1 SrcRegion=<16;16,1> Type=ub
149 .declare muwMSGPAYLOAD Base=r30 ElementSize=2 SrcRegion=<16;16,1> Type=uw
150 .declare mudMSGPAYLOAD Base=r30 ElementSize=4 SrcRegion=<8;8,1> Type=ud
151 .declare mfMSGPAYLOAD Base=r30 ElementSize=4 SrcRegion=<8;8,1> Type=f
152
153 //=================== End of thread instruction ===========================
154
155
156 //=====================Pointers Used=====================================
157
158
159 //=======================================================================
160
161
162 //r11-r17
163 // Define temp space for any usages
164
165
166 // Common Buffers
167
168
169 // temp space for rotation
170
171 .declare fROBUF Base=r11.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=f
172
173 .declare udROBUF Base=r11.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=ud
174
175 .declare uwROBUF Base=r11.0 ElementSize=2 SrcRegion=<16;16,1> DstRegion=<1> Type=uw
176
177 .declare ubROBUF Base=r11.0 ElementSize=1 SrcRegion=<16;16,1> DstRegion=<1> Type=ub
178
179 .declare ub4ROBUF Base=r11.0 ElementSize=1 SrcRegion=<32;8,4> DstRegion=<4> Type=ub
180
181
182 // End of common.inc
183
184
185 // Module name: Save_AVS_RGBX.asm
186 //
187 // Save packed ARGB 444 frame data block of size 16x16
188 //
189 // To save 16x16 block (64x16 byte layout for ARGB8888) we need 4 send instructions with 32x8 in each
190 // --------
191 // | 0 | 1 |
192 // | 2 | 3 |
193 // ---------
194 // the 4 32x8 block send is used
195
196
197
198 // Module name: Save.inc
199
200
201
202
203 // Description: Includes all definitions explicit to Fast Composite.
204
205
206
207
208 // End of common.inc
209
210
211 //========== GRF partition ==========
212 // r0 header : r0 (1 GRF)
213 // Static parameters : r1 - r6 (6 GRFS)
214 // Inline parameters : r7 - r8 (2 GRFs)
215 // MSGSRC : r27 (1 GRF)
216 //===================================
217
218 //Interface:
219 //========== Static Parameters (Explicit To Fast Composite) ==========
220 //r1
221 //CSC Set 0
222
223
224 .declare udCSC_CURBE Base=r1.0 ElementSize=4 Type=ud
225
226 //Constant alpha
227
228
229 //r2
230
231
232 // Gen7 AVS WA
233
234
235 // WiDi Definitions
236
237
238 //Colorfill
239
240
241 // 0: 0-degree, 1: 90, 2: 180, 3: 270-degree, clockwise.
242
243 .declare ubCOLOR_PIXEL_VAL Base=r2.20 ElementSize=1 SrcRegion=<0;1,0> DstRegion=<1> Type=ub
244
245 //r3
246 //Normalised Ratio of Horizontal step size with main video for all layers
247
248
249 //Normalised Ratio of Horizontal step size with main video for all layers becomes
250 //Normalised Horizontal step size for all layers in VP_Setup.asm
251
252
253 //r4
254 //Normalised Vertical step size for all layers
255
256
257 //r5
258 //Normalised Vertical Frame Origin for all layers
259
260
261 //r6
262 //Normalised Horizontal Frame Origin for all layers
263
264
265 //========== Inline Parameters (Explicit To Fast Composite) ==========
266
267
268 //Main video Step X
269
270
271 //====================== Binding table (Explicit To Fast Composite)=========================================
272
273
274 //Used by Interlaced Scaling Kernels
275
276
277 //========== Sampler State Table Index (Explicit To Fast Composite)==========
278 //Sampler Index for AVS/IEF messages
279
280
281 //Sampler Index for SIMD16 sampler messages
282
283
284 //=============================================================================
285
286 .declare fBUFFER_0 Base=r64.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=f
287 .declare fBUFFER_1 Base=r80.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=f
288 .declare fBUFFER_2 Base=r96.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=f
289 .declare fBUFFER_3 Base=r112.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=f
290 .declare fBUFFER_4 Base=r28.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=f
291 .declare fBUFFER_5 Base=r46.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=f
292
293 .declare udBUFFER_0 Base=r64.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=ud
294 .declare udBUFFER_1 Base=r80.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=ud
295 .declare udBUFFER_2 Base=r96.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=ud
296 .declare udBUFFER_3 Base=r112.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=ud
297 .declare udBUFFER_4 Base=r28.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=ud
298 .declare udBUFFER_5 Base=r46.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=ud
299
300 .declare uwBUFFER_0 Base=r64.0 ElementSize=2 SrcRegion=<16;16,1> DstRegion=<1> Type=uw
301 .declare uwBUFFER_1 Base=r80.0 ElementSize=2 SrcRegion=<16;16,1> DstRegion=<1> Type=uw
302 .declare uwBUFFER_2 Base=r96.0 ElementSize=2 SrcRegion=<16;16,1> DstRegion=<1> Type=uw
303 .declare uwBUFFER_3 Base=r112.0 ElementSize=2 SrcRegion=<16;16,1> DstRegion=<1> Type=uw
304 .declare uwBUFFER_4 Base=r28.0 ElementSize=2 SrcRegion=<16;16,1> DstRegion=<1> Type=uw
305 .declare uwBUFFER_5 Base=r46.0 ElementSize=2 SrcRegion=<16;16,1> DstRegion=<1> Type=uw
306
307 .declare ubBUFFER_0 Base=r64.0 ElementSize=1 SrcRegion=<16;16,1> DstRegion=<1> Type=ub
308 .declare ubBUFFER_1 Base=r80.0 ElementSize=1 SrcRegion=<16;16,1> DstRegion=<1> Type=ub
309 .declare ubBUFFER_2 Base=r96.0 ElementSize=1 SrcRegion=<16;16,1> DstRegion=<1> Type=ub
310 .declare ubBUFFER_3 Base=r112.0 ElementSize=1 SrcRegion=<16;16,1> DstRegion=<1> Type=ub
311 .declare ubBUFFER_4 Base=r28.0 ElementSize=1 SrcRegion=<16;16,1> DstRegion=<1> Type=ub
312 .declare ubBUFFER_5 Base=r46.0 ElementSize=1 SrcRegion=<16;16,1> DstRegion=<1> Type=ub
313
314 .declare ub4BUFFER_0 Base=r64.0 ElementSize=1 SrcRegion=<32;8,4> DstRegion=<4> Type=ub
315 .declare ub4BUFFER_1 Base=r80.0 ElementSize=1 SrcRegion=<32;8,4> DstRegion=<4> Type=ub
316 .declare ub4BUFFER_2 Base=r96.0 ElementSize=1 SrcRegion=<32;8,4> DstRegion=<4> Type=ub
317 .declare ub4BUFFER_3 Base=r112.0 ElementSize=1 SrcRegion=<32;8,4> DstRegion=<4> Type=ub
318 .declare ub4BUFFER_4 Base=r28.0 ElementSize=1 SrcRegion=<32;8,4> DstRegion=<4> Type=ub
319 .declare ub4BUFFER_5 Base=r46.0 ElementSize=1 SrcRegion=<32;8,4> DstRegion=<4> Type=ub
320
321 //Pointer to mask reg
322
323
324 //r18
325
326
327 //Always keep Cannel Pointers and Offsets in same GRF, so that we can use
328 // NODDCLR, NODDCHK flags. -rT
329
330
331 .declare udCSC_COEFF_0 Base=r18.0 ElementSize=4 Type=ud // 1 GRF
332
333 //r19
334
335
336 .declare udCSC_COEFF_1 Base=r19.0 ElementSize=4 Type=ud // 1 GRF
337
338
339 //r20
340
341 .declare uwALPHA_MASK_REG_TEMP Base=r20.0 ElementSize=2 SrcRegion=<16;16,1> Type=uw // 1 GRF
342
343 //r21
344
345 .declare uwALPHA_MASK_REG Base=r21.0 ElementSize=2 SrcRegion=<16;16,1> Type=uw // 1 GRF
346
347 //r22
348
349
350 //Always keep Cannel Pointers and Offsets in same GRF, so that we can use
351 // NODDCLR, NODDCHK flags. -rT
352
353
354 //Keep fORIGIN_X_NLAS, fY_OFFSET_2ND_BLOCK, fSTEP_X_NLAS, pMSGDSC_COPY, ubCONST_ALPHA_COPY as
355 //sub registers of same GRF to enable using NODDCLR NODDCHK. -rT
356
357 //r23
358
359
360 //Lumakey
361
362
363 //r24
364
365
366 //r25
367
368
369 //r26
370
371
372 //defines to generate LABELS during compile time.
373
374
375 //Msg payload buffers; upto 4 full-size messages can be written
376
377
378 .declare mudMSGPAYLOAD0 Base=r29.0 ElementSize=4 SrcRegion=<8;8,1> Type=ud
379 .declare mudMSGPAYLOAD1 Base=r38.0 ElementSize=4 SrcRegion=<8;8,1> Type=ud
380 .declare mudMSGPAYLOAD2 Base=r47.0 ElementSize=4 SrcRegion=<8;8,1> Type=ud
381 .declare mudMSGPAYLOAD3 Base=r56.0 ElementSize=4 SrcRegion=<8;8,1> Type=ud
382
383 .declare muwMSGPAYLOAD0 Base=r29.0 ElementSize=2 SrcRegion=<16;16,1> Type=uw
384 .declare muwMSGPAYLOAD1 Base=r38.0 ElementSize=2 SrcRegion=<16;16,1> Type=uw
385 .declare muwMSGPAYLOAD2 Base=r47.0 ElementSize=2 SrcRegion=<16;16,1> Type=uw
386 .declare muwMSGPAYLOAD3 Base=r56.0 ElementSize=2 SrcRegion=<16;16,1> Type=uw
387
388 .declare mubMSGPAYLOAD0 Base=r29.0 ElementSize=1 SrcRegion=<16;16,1> Type=ub
389 .declare mubMSGPAYLOAD1 Base=r38.0 ElementSize=1 SrcRegion=<16;16,1> Type=ub
390 .declare mubMSGPAYLOAD2 Base=r47.0 ElementSize=1 SrcRegion=<16;16,1> Type=ub
391 .declare mubMSGPAYLOAD3 Base=r56.0 ElementSize=1 SrcRegion=<16;16,1> Type=ub
392 .declare mubMSGPAYLOAD4 Base=r32.0 ElementSize=1 SrcRegion=<16;16,1> Type=ub
393 .declare mubMSGPAYLOAD5 Base=r41.0 ElementSize=1 SrcRegion=<16;16,1> Type=ub
394 .declare mubMSGPAYLOAD6 Base=r50.0 ElementSize=1 SrcRegion=<16;16,1> Type=ub
395 .declare mubMSGPAYLOAD7 Base=r59.0 ElementSize=1 SrcRegion=<16;16,1> Type=ub
396
397
398 // the r17 register (nTEMP0) is originally defined from "Common.inc"
399 // instead of re-defining a nTEMP0 here, we use "SAVE_RGB" suffix for its naming
400
401 .declare uwTemp0 Base=r17.0 ElementSize=2 Type=uw
402
403
404 //_SAVE_INC_
405
406
407 // At the save module we have all 8 address sub-registers available.
408 // So we will use PING-PONG type of scheme to save the data using
409 // pointers pBUF_CHNL_TOP_8x4 and pBUF_CHNL_BOT_8x4. This will help
410 // reduce dependency. - rT
411
412 //Internal LAYOUT:(RRGGBBAA)
413 //Assign buffer channel order for Buffer 0123 in the order RGBA a0.3>A, a0.2>B, a0.1>G, a0.0>R
414 // R = 0, G= 4, B = 8, A = 12.
415 mov (4) acc0.0<1>:w 0x62EA:v
416 add (4) acc0.0<1>:w acc0<4;4,1>:w 70:uw
417 shl (4) r22.0<1>:w acc0<4;4,1>:w 5:uw
418
419 // if channel swap?
420 // This means that it should be BGRA(B is the LSB) or RGBA
421 // the internal format is always RGBA(MSB-A-B-G-R).
422 and.nz.f0.0 null<1>:w r2.3<0;1,0>:uw 0x01:w
423
424 //wBUFF_CHNL_PTR points to either buffer 0 or buffer 4.
425 //Add appropriate offsets to get pointers for all buffers (1,2,3 or 5).
426 //Offsets are zero for buffer 0 and buffer 4.
427 add (4) a0.0:uw r22.0<4;4,1>:w 0:uw
428
429 // pointer swap
430 (f0.0) mov (1) uwTemp0<1> a0.0<0;1,0>:uw
431 (f0.0) mov (1) a0.0<1>:uw a0.2<0;1,0>:uw
432 (f0.0) mov (1) a0.2<1>:uw uwTemp0<0;1,0>
433
434 shl (1) r27.0<1>:d r9.0<0;1,0>:w 2:w { NoDDClr } // H. block origin need to be quadrupled
435 mov (1) r27.1<1>:d r9.1<0;1,0>:w { NoDDClr, NoDDChk } // Block origin (1st quadrant)
436 mov (1) r27.2<1>:ud 0x3001F:ud { NoDDChk } // Block width and height (32x4)
437
438 mov (4) a0.4<1>:uw a0.0<4;4,1>:uw
439
440 mov (8) r28<1>:ud r27<8;8,1>:ud
441 mov (8) r37<1>:ud r27<8;8,1>:ud
442 mov (8) r46<1>:ud r27<8;8,1>:ud
443 mov (8) r55<1>:ud r27<8;8,1>:ud
444
445 mov (8) r31<1>:ud r27<8;8,1>:ud
446 mov (8) r40<1>:ud r27<8;8,1>:ud
447 mov (8) r49<1>:ud r27<8;8,1>:ud
448 mov (8) r58<1>:ud r27<8;8,1>:ud
449
450 //Buffer 0/1 are written by using 4 32x4.
451
452 add (1) r37.0<1>:d r27.0<0;1,0>:d 32:d
453
454 add (1) r46.1<1>:d r27.1<0;1,0>:d 4:d
455
456 add (1) r55.1<1>:d r27.1<0;1,0>:d 4:d
457 add (1) r55.0<1>:d r27.0<0;1,0>:d 32:d
458
459 // write Buf_0 to 1st quarter of four horizontal output blocks
460
461 // Please note the scattered order of NODDCLR, NODDCHK flags. Since the sub-registers
462 // of destination reg are not updated at one place and hence even flags are scattered. -rT
463
464 /* for block 0 the left part of buffer 0 and 1 */
465 mov (8) mubMSGPAYLOAD0(0, 0)<4> r[a0.0, 1]<16;8,2>:ub
466 mov (8) mubMSGPAYLOAD0(0, 1)<4> r[a0.1, 1]<16;8,2>:ub
467 mov (8) mubMSGPAYLOAD0(0, 2)<4> r[a0.2, 1]<16;8,2>:ub
468 mov (8) mubMSGPAYLOAD0(0, 3)<4> r2.31:ub
469
470 mov (8) mubMSGPAYLOAD0(1, 0)<4> r[a0.0, 33]<16;8,2>:ub
471 mov (8) mubMSGPAYLOAD0(1, 1)<4> r[a0.1, 33]<16;8,2>:ub
472 mov (8) mubMSGPAYLOAD0(1, 2)<4> r[a0.2, 33]<16;8,2>:ub
473 mov (8) mubMSGPAYLOAD0(1, 3)<4> r2.31:ub
474
475 mov (8) mubMSGPAYLOAD1(0, 0)<4> r[a0.0, 17]<16;8,2>:ub
476 mov (8) mubMSGPAYLOAD1(0, 1)<4> r[a0.1, 17]<16;8,2>:ub
477 mov (8) mubMSGPAYLOAD1(0, 2)<4> r[a0.2, 17]<16;8,2>:ub
478 mov (8) mubMSGPAYLOAD1(0, 3)<4> r2.31:ub
479
480 mov (8) mubMSGPAYLOAD1(1, 0)<4> r[a0.0, 49]<16;8,2>:ub
481 mov (8) mubMSGPAYLOAD1(1, 1)<4> r[a0.1, 49]<16;8,2>:ub
482 mov (8) mubMSGPAYLOAD1(1, 2)<4> r[a0.2, 49]<16;8,2>:ub
483 mov (8) mubMSGPAYLOAD1(1, 3)<4> r2.31:ub
484
485 mov (8) mubMSGPAYLOAD0(2, 0)<4> r[a0.0, 65]<16;8,2>:ub
486 mov (8) mubMSGPAYLOAD0(2, 1)<4> r[a0.1, 65]<16;8,2>:ub
487 mov (8) mubMSGPAYLOAD0(2, 2)<4> r[a0.2, 65]<16;8,2>:ub
488 mov (8) mubMSGPAYLOAD0(2, 3)<4> r2.31:ub
489
490 mov (8) mubMSGPAYLOAD0(3, 0)<4> r[a0.0, 97]<16;8,2>:ub
491 mov (8) mubMSGPAYLOAD0(3, 1)<4> r[a0.1, 97]<16;8,2>:ub
492 mov (8) mubMSGPAYLOAD0(3, 2)<4> r[a0.2, 97]<16;8,2>:ub
493 mov (8) mubMSGPAYLOAD0(3, 3)<4> r2.31:ub
494
495 mov (8) mubMSGPAYLOAD1(2, 0)<4> r[a0.0, 81]<16;8,2>:ub
496 mov (8) mubMSGPAYLOAD1(2, 1)<4> r[a0.1, 81]<16;8,2>:ub
497 mov (8) mubMSGPAYLOAD1(2, 2)<4> r[a0.2, 81]<16;8,2>:ub
498 mov (8) mubMSGPAYLOAD1(2, 3)<4> r2.31:ub
499
500 mov (8) mubMSGPAYLOAD1(3, 0)<4> r[a0.0, 113]<16;8,2>:ub
501 mov (8) mubMSGPAYLOAD1(3, 1)<4> r[a0.1, 113]<16;8,2>:ub
502 mov (8) mubMSGPAYLOAD1(3, 2)<4> r[a0.2, 113]<16;8,2>:ub
503 mov (8) mubMSGPAYLOAD1(3, 3)<4> r2.31:ub
504
505 /* For Buffer 0 */
506 send (16) null<1>:d r28 0x5 0x0A0A8018:ud
507 send (16) null<1>:d r37 0x5 0x0A0A8018:ud
508
509 add (4) a0.0<1>:uw a0.4<4;4,1>:uw 512:uw
510 mov (8) mubMSGPAYLOAD2(0, 0)<4> r[a0.0, 1]<16;8,2>:ub
511 mov (8) mubMSGPAYLOAD2(0, 1)<4> r[a0.1, 1]<16;8,2>:ub
512 mov (8) mubMSGPAYLOAD2(0, 2)<4> r[a0.2, 1]<16;8,2>:ub
513 mov (8) mubMSGPAYLOAD2(0, 3)<4> r2.31:ub
514
515 mov (8) mubMSGPAYLOAD2(1, 0)<4> r[a0.0, 33]<16;8,2>:ub
516 mov (8) mubMSGPAYLOAD2(1, 1)<4> r[a0.1, 33]<16;8,2>:ub
517 mov (8) mubMSGPAYLOAD2(1, 2)<4> r[a0.2, 33]<16;8,2>:ub
518 mov (8) mubMSGPAYLOAD2(1, 3)<4> r2.31:ub
519
520 mov (8) mubMSGPAYLOAD3(0, 0)<4> r[a0.0, 17]<16;8,2>:ub
521 mov (8) mubMSGPAYLOAD3(0, 1)<4> r[a0.1, 17]<16;8,2>:ub
522 mov (8) mubMSGPAYLOAD3(0, 2)<4> r[a0.2, 17]<16;8,2>:ub
523 mov (8) mubMSGPAYLOAD3(0, 3)<4> r2.31:ub
524
525 mov (8) mubMSGPAYLOAD3(1, 0)<4> r[a0.0, 49]<16;8,2>:ub
526 mov (8) mubMSGPAYLOAD3(1, 1)<4> r[a0.1, 49]<16;8,2>:ub
527 mov (8) mubMSGPAYLOAD3(1, 2)<4> r[a0.2, 49]<16;8,2>:ub
528 mov (8) mubMSGPAYLOAD3(1, 3)<4> r2.31:ub
529
530 mov (8) mubMSGPAYLOAD2(2, 0)<4> r[a0.0, 65]<16;8,2>:ub
531 mov (8) mubMSGPAYLOAD2(2, 1)<4> r[a0.1, 65]<16;8,2>:ub
532 mov (8) mubMSGPAYLOAD2(2, 2)<4> r[a0.2, 65]<16;8,2>:ub
533 mov (8) mubMSGPAYLOAD2(2, 3)<4> r2.31:ub
534
535 mov (8) mubMSGPAYLOAD2(3, 0)<4> r[a0.0, 97]<16;8,2>:ub
536 mov (8) mubMSGPAYLOAD2(3, 1)<4> r[a0.1, 97]<16;8,2>:ub
537 mov (8) mubMSGPAYLOAD2(3, 2)<4> r[a0.2, 97]<16;8,2>:ub
538 mov (8) mubMSGPAYLOAD2(3, 3)<4> r2.31:ub
539
540 mov (8) mubMSGPAYLOAD3(2, 0)<4> r[a0.0, 81]<16;8,2>:ub
541 mov (8) mubMSGPAYLOAD3(2, 1)<4> r[a0.1, 81]<16;8,2>:ub
542 mov (8) mubMSGPAYLOAD3(2, 2)<4> r[a0.2, 81]<16;8,2>:ub
543 mov (8) mubMSGPAYLOAD3(2, 3)<4> r2.31:ub
544
545 mov (8) mubMSGPAYLOAD3(3, 0)<4> r[a0.0, 113]<16;8,2>:ub
546 mov (8) mubMSGPAYLOAD3(3, 1)<4> r[a0.1, 113]<16;8,2>:ub
547 mov (8) mubMSGPAYLOAD3(3, 2)<4> r[a0.2, 113]<16;8,2>:ub
548 mov (8) mubMSGPAYLOAD3(3, 3)<4> r2.31:ub
549 // send Buffer 1
550 send (16) null<1>:d r46 0x5 0x0A0A8018:ud
551 send (16) null<1>:d r55 0x5 0x0A0A8018:ud
552
553
554 /* for Buffer 2/3 */
555 mov (8) r28<1>:ud r27<8;8,1>:ud
556 mov (8) r37<1>:ud r27<8;8,1>:ud
557 mov (8) r46<1>:ud r27<8;8,1>:ud
558 mov (8) r55<1>:ud r27<8;8,1>:ud
559
560 add (1) r28.1<1>:d r27.1<0;1,0>:d 8:d
561
562 add (1) r37.0<1>:d r27.0<0;1,0>:d 32:d
563 add (1) r37.1<1>:d r27.1<0;1,0>:d 8:d
564
565 add (1) r46.1<1>:d r27.1<0;1,0>:d 12:d
566
567 add (1) r55.1<1>:d r27.1<0;1,0>:d 12:d
568 add (1) r55.0<1>:d r27.0<0;1,0>:d 32:d
569
570 add (4) a0.0<1>:uw a0.4<4;4,1>:uw 1024:uw
571
572 mov (8) mubMSGPAYLOAD0(0, 0)<4> r[a0.0, 1]<16;8,2>:ub
573 mov (8) mubMSGPAYLOAD0(0, 1)<4> r[a0.1, 1]<16;8,2>:ub
574 mov (8) mubMSGPAYLOAD0(0, 2)<4> r[a0.2, 1]<16;8,2>:ub
575 mov (8) mubMSGPAYLOAD0(0, 3)<4> r2.31:ub
576
577 mov (8) mubMSGPAYLOAD0(1, 0)<4> r[a0.0, 33]<16;8,2>:ub
578 mov (8) mubMSGPAYLOAD0(1, 1)<4> r[a0.1, 33]<16;8,2>:ub
579 mov (8) mubMSGPAYLOAD0(1, 2)<4> r[a0.2, 33]<16;8,2>:ub
580 mov (8) mubMSGPAYLOAD0(1, 3)<4> r2.31:ub
581
582 mov (8) mubMSGPAYLOAD1(0, 0)<4> r[a0.0, 17]<16;8,2>:ub
583 mov (8) mubMSGPAYLOAD1(0, 1)<4> r[a0.1, 17]<16;8,2>:ub
584 mov (8) mubMSGPAYLOAD1(0, 2)<4> r[a0.2, 17]<16;8,2>:ub
585 mov (8) mubMSGPAYLOAD1(0, 3)<4> r2.31:ub
586
587 mov (8) mubMSGPAYLOAD1(1, 0)<4> r[a0.0, 49]<16;8,2>:ub
588 mov (8) mubMSGPAYLOAD1(1, 1)<4> r[a0.1, 49]<16;8,2>:ub
589 mov (8) mubMSGPAYLOAD1(1, 2)<4> r[a0.2, 49]<16;8,2>:ub
590 mov (8) mubMSGPAYLOAD1(1, 3)<4> r2.31:ub
591
592 mov (8) mubMSGPAYLOAD0(2, 0)<4> r[a0.0, 65]<16;8,2>:ub
593 mov (8) mubMSGPAYLOAD0(2, 1)<4> r[a0.1, 65]<16;8,2>:ub
594 mov (8) mubMSGPAYLOAD0(2, 2)<4> r[a0.2, 65]<16;8,2>:ub
595 mov (8) mubMSGPAYLOAD0(2, 3)<4> r2.31:ub
596
597 mov (8) mubMSGPAYLOAD0(3, 0)<4> r[a0.0, 97]<16;8,2>:ub
598 mov (8) mubMSGPAYLOAD0(3, 1)<4> r[a0.1, 97]<16;8,2>:ub
599 mov (8) mubMSGPAYLOAD0(3, 2)<4> r[a0.2, 97]<16;8,2>:ub
600 mov (8) mubMSGPAYLOAD0(3, 3)<4> r2.31:ub
601
602 mov (8) mubMSGPAYLOAD1(2, 0)<4> r[a0.0, 81]<16;8,2>:ub
603 mov (8) mubMSGPAYLOAD1(2, 1)<4> r[a0.1, 81]<16;8,2>:ub
604 mov (8) mubMSGPAYLOAD1(2, 2)<4> r[a0.2, 81]<16;8,2>:ub
605 mov (8) mubMSGPAYLOAD1(2, 3)<4> r2.31:ub
606
607 mov (8) mubMSGPAYLOAD1(3, 0)<4> r[a0.0, 113]<16;8,2>:ub
608 mov (8) mubMSGPAYLOAD1(3, 1)<4> r[a0.1, 113]<16;8,2>:ub
609 mov (8) mubMSGPAYLOAD1(3, 2)<4> r[a0.2, 113]<16;8,2>:ub
610 mov (8) mubMSGPAYLOAD1(3, 3)<4> r2.31:ub
611
612 // Send Buffer 2
613 send (16) null<1>:d r28 0x5 0x0A0A8018:ud
614 send (16) null<1>:d r37 0x5 0x0A0A8018:ud
615
616 add (4) a0.0<1>:uw a0.4<4;4,1>:uw 1536:uw
617 mov (8) mubMSGPAYLOAD2(0, 0)<4> r[a0.0, 1]<16;8,2>:ub
618 mov (8) mubMSGPAYLOAD2(0, 1)<4> r[a0.1, 1]<16;8,2>:ub
619 mov (8) mubMSGPAYLOAD2(0, 2)<4> r[a0.2, 1]<16;8,2>:ub
620 mov (8) mubMSGPAYLOAD2(0, 3)<4> r2.31:ub
621
622 mov (8) mubMSGPAYLOAD2(1, 0)<4> r[a0.0, 33]<16;8,2>:ub
623 mov (8) mubMSGPAYLOAD2(1, 1)<4> r[a0.1, 33]<16;8,2>:ub
624 mov (8) mubMSGPAYLOAD2(1, 2)<4> r[a0.2, 33]<16;8,2>:ub
625 mov (8) mubMSGPAYLOAD2(1, 3)<4> r2.31:ub
626
627 mov (8) mubMSGPAYLOAD3(0, 0)<4> r[a0.0, 17]<16;8,2>:ub
628 mov (8) mubMSGPAYLOAD3(0, 1)<4> r[a0.1, 17]<16;8,2>:ub
629 mov (8) mubMSGPAYLOAD3(0, 2)<4> r[a0.2, 17]<16;8,2>:ub
630 mov (8) mubMSGPAYLOAD3(0, 3)<4> r2.31:ub
631
632 mov (8) mubMSGPAYLOAD3(1, 0)<4> r[a0.0, 49]<16;8,2>:ub
633 mov (8) mubMSGPAYLOAD3(1, 1)<4> r[a0.1, 49]<16;8,2>:ub
634 mov (8) mubMSGPAYLOAD3(1, 2)<4> r[a0.2, 49]<16;8,2>:ub
635 mov (8) mubMSGPAYLOAD3(1, 3)<4> r2.31:ub
636
637 mov (8) mubMSGPAYLOAD2(2, 0)<4> r[a0.0, 65]<16;8,2>:ub
638 mov (8) mubMSGPAYLOAD2(2, 1)<4> r[a0.1, 65]<16;8,2>:ub
639 mov (8) mubMSGPAYLOAD2(2, 2)<4> r[a0.2, 65]<16;8,2>:ub
640 mov (8) mubMSGPAYLOAD2(2, 3)<4> r2.31:ub
641
642 mov (8) mubMSGPAYLOAD2(3, 0)<4> r[a0.0, 97]<16;8,2>:ub
643 mov (8) mubMSGPAYLOAD2(3, 1)<4> r[a0.1, 97]<16;8,2>:ub
644 mov (8) mubMSGPAYLOAD2(3, 2)<4> r[a0.2, 97]<16;8,2>:ub
645 mov (8) mubMSGPAYLOAD2(3, 3)<4> r2.31:ub
646
647 mov (8) mubMSGPAYLOAD3(2, 0)<4> r[a0.0, 81]<16;8,2>:ub
648 mov (8) mubMSGPAYLOAD3(2, 1)<4> r[a0.1, 81]<16;8,2>:ub
649 mov (8) mubMSGPAYLOAD3(2, 2)<4> r[a0.2, 81]<16;8,2>:ub
650 mov (8) mubMSGPAYLOAD3(2, 3)<4> r2.31:ub
651
652 mov (8) mubMSGPAYLOAD3(3, 0)<4> r[a0.0, 113]<16;8,2>:ub
653 mov (8) mubMSGPAYLOAD3(3, 1)<4> r[a0.1, 113]<16;8,2>:ub
654 mov (8) mubMSGPAYLOAD3(3, 2)<4> r[a0.2, 113]<16;8,2>:ub
655 mov (8) mubMSGPAYLOAD3(3, 3)<4> r2.31:ub
656 // send buffer 3
657 send (16) null<1>:d r46 0x5 0x0A0A8018:ud
658 send (16) null<1>:d r55 0x5 0x0A0A8018:ud
659
660
661