Debian Codegraph @

src/shaders/post_processing/gen8/Save_AVS_RGBX.g8a

raw
1 /*
2 * Copyright 2000-2013 Intel Corporation All Rights Reserved
3 *
4 * Licensed under the Apache License, Version 2.0 (the "License");
5 * you may not use this file except in compliance with the License.
6 * You may obtain a copy of the License at
7 *
8 * http://www.apache.org/licenses/LICENSE-2.0
9 *
10 * Unless required by applicable law or agreed to in writing, software
11 * distributed under the License is distributed on an "AS IS" BASIS,
12 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13 * See the License for the specific language governing permissions and
14 * limitations under the License.
15 * Authors:
16 * Zhao Yakui <[email protected]>
17 */
18
19
20
21 // Module name: common.inc
22 //
23 // Common header file for all Video-Processing kernels
24 //
25
26 .default_execution_size (16)
27 .default_register_type :ub
28
29 .reg_count_total 128
30 .reg_count_payload 7
31
32 //========== Common constants ==========
33
34
35 //========== Macros ==========
36
37
38 //Fast Jump, For more details see "Set_Layer_N.asm"
39
40
41 //========== Defines ====================
42
43 //========== Static Parameters (Common To All) ==========
44 //r1
45
46
47 //r2
48
49 // e.g. byte0 byte1 byte2
50 // YUYV 0 1 3
51 // YVYU 0 3 1
52
53 //Color Pipe (IECP) parameters
54
55
56 //ByteCopy
57
58
59 //r4
60
61 // e.g. byte0 byte1 byte2
62 // YUYV 0 1 3
63 // YVYU 0 3 1
64
65
66 //========== Inline parameters (Common To All) ===========
67
68
69 //============== Binding Index Table===========
70 //Common between DNDI and DNUV
71
72
73 //================= Common Message Descriptor =====
74 // Message descriptor for thread spawning
75 // Message Descriptors
76 // = 000 0001 (min message len 1 ) 0,0000 (resp len 0 -add later)
77 // 0000,0000,0000
78 // 0001(Spawn a root thread),0001 (Root thread spawn thread)
79 // = 0x02000011
80 // Thread Spawner Message Descriptor
81
82
83 // Message descriptor for atomic operation add
84 // Message Descriptors
85 // = 000 0110 (min message len 6 ) 0,0000 (resp len 0 -add later)
86 // 1(header present)001,10(typed atomic operation)0(return enabled)0(slot group, low 8 bits),0111 (AOP_Add)
87 // 0000,0000 (Binding table index, added later)
88 // = 0x02000011
89
90 // Atomic Operation Add Message Descriptor
91
92
93 // Message descriptor for dataport media write
94 // Message Descriptors
95 // = 000 0001 (min message len 1 - add later) 00000 (resp len 0)
96 // 1 (header present 1) 0 1010 (media block write) 000000
97 // 00000000 (binding table index - set later)
98 // = 0x020A8000
99
100
101 // Message Length defines
102
103
104 // Response Length defines
105
106
107 // Block Width and Height Size defines
108
109
110 // Extended Message Descriptors
111
112
113 // Common message descriptors:
114
115
116 //===================== Math Function Control ===================================
117
118
119 //============ Message Registers ===============
120 // buf4 starts from r28
121
122
123 //#define mMSGHDR_EOT r43 // Dummy Message Register for EOT
124
125
126 .declare mubMSGPAYLOAD Base=r30 ElementSize=1 SrcRegion=<16;16,1> Type=ub
127 .declare muwMSGPAYLOAD Base=r30 ElementSize=2 SrcRegion=<16;16,1> Type=uw
128 .declare mudMSGPAYLOAD Base=r30 ElementSize=4 SrcRegion=<8;8,1> Type=ud
129 .declare mfMSGPAYLOAD Base=r30 ElementSize=4 SrcRegion=<8;8,1> Type=f
130
131 //=================== End of thread instruction ===========================
132
133
134 //=====================Pointers Used=====================================
135
136
137 //=======================================================================
138
139
140 //r11-r17
141 // Define temp space for any usages
142
143
144 // Common Buffers
145
146
147 // temp space for rotation
148
149 .declare fROBUF Base=r11.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=f
150
151 .declare udROBUF Base=r11.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=ud
152
153 .declare uwROBUF Base=r11.0 ElementSize=2 SrcRegion=<16;16,1> DstRegion=<1> Type=uw
154
155 .declare ubROBUF Base=r11.0 ElementSize=1 SrcRegion=<16;16,1> DstRegion=<1> Type=ub
156
157 .declare ub4ROBUF Base=r11.0 ElementSize=1 SrcRegion=<32;8,4> DstRegion=<4> Type=ub
158
159
160 // End of common.inc
161
162
163 // Module name: Save_AVS_RGBX.asm
164 //
165 // Save packed ARGB 444 frame data block of size 16x16
166 //
167 // To save 16x16 block (64x16 byte layout for ARGB8888) we need 8 send instructions with 32x4 in each
168 // --------
169 // | 0 | 1 |
170 // | 2 | 3 |
171 // | 4 | 5 |
172 // | 6 | 7 |
173 // ---------
174 // the 8 32x4 block send is used
175
176
177
178 // Module name: Save.inc
179
180
181
182
183 // Description: Includes all definitions explicit to Fast Composite.
184
185
186
187
188 // End of common.inc
189
190
191 //========== GRF partition ==========
192 // r0 header : r0 (1 GRF)
193 // Static parameters : r1 - r6 (6 GRFS)
194 // Inline parameters : r7 - r8 (2 GRFs)
195 // MSGSRC : r27 (1 GRF)
196 //===================================
197
198 //Interface:
199 //========== Static Parameters (Explicit To Fast Composite) ==========
200 //r1
201 //CSC Set 0
202
203
204 .declare udCSC_CURBE Base=r1.0 ElementSize=4 Type=ud
205
206 //Constant alpha
207
208
209 //r2
210
211
212 // Gen7 AVS WA
213
214
215 // WiDi Definitions
216
217
218 //Colorfill
219
220
221 // 0: 0-degree, 1: 90, 2: 180, 3: 270-degree, clockwise.
222
223 .declare ubCOLOR_PIXEL_VAL Base=r2.20 ElementSize=1 SrcRegion=<0;1,0> DstRegion=<1> Type=ub
224
225 //r3
226 //Normalised Ratio of Horizontal step size with main video for all layers
227
228
229 //Normalised Ratio of Horizontal step size with main video for all layers becomes
230 //Normalised Horizontal step size for all layers in VP_Setup.asm
231
232
233 //r4
234 //Normalised Vertical step size for all layers
235
236
237 //r5
238 //Normalised Vertical Frame Origin for all layers
239
240
241 //r6
242 //Normalised Horizontal Frame Origin for all layers
243
244
245 //========== Inline Parameters (Explicit To Fast Composite) ==========
246
247
248 //Main video Step X
249
250
251 //====================== Binding table (Explicit To Fast Composite)=========================================
252
253
254 //Used by Interlaced Scaling Kernels
255
256
257 //========== Sampler State Table Index (Explicit To Fast Composite)==========
258 //Sampler Index for AVS/IEF messages
259
260
261 //Sampler Index for SIMD16 sampler messages
262
263
264 //=============================================================================
265
266 .declare fBUFFER_0 Base=r64.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=f
267 .declare fBUFFER_1 Base=r80.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=f
268 .declare fBUFFER_2 Base=r96.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=f
269 .declare fBUFFER_3 Base=r112.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=f
270 .declare fBUFFER_4 Base=r28.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=f
271 .declare fBUFFER_5 Base=r46.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=f
272
273 .declare udBUFFER_0 Base=r64.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=ud
274 .declare udBUFFER_1 Base=r80.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=ud
275 .declare udBUFFER_2 Base=r96.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=ud
276 .declare udBUFFER_3 Base=r112.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=ud
277 .declare udBUFFER_4 Base=r28.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=ud
278 .declare udBUFFER_5 Base=r46.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=ud
279
280 .declare uwBUFFER_0 Base=r64.0 ElementSize=2 SrcRegion=<16;16,1> DstRegion=<1> Type=uw
281 .declare uwBUFFER_1 Base=r80.0 ElementSize=2 SrcRegion=<16;16,1> DstRegion=<1> Type=uw
282 .declare uwBUFFER_2 Base=r96.0 ElementSize=2 SrcRegion=<16;16,1> DstRegion=<1> Type=uw
283 .declare uwBUFFER_3 Base=r112.0 ElementSize=2 SrcRegion=<16;16,1> DstRegion=<1> Type=uw
284 .declare uwBUFFER_4 Base=r28.0 ElementSize=2 SrcRegion=<16;16,1> DstRegion=<1> Type=uw
285 .declare uwBUFFER_5 Base=r46.0 ElementSize=2 SrcRegion=<16;16,1> DstRegion=<1> Type=uw
286
287 .declare ubBUFFER_0 Base=r64.0 ElementSize=1 SrcRegion=<16;16,1> DstRegion=<1> Type=ub
288 .declare ubBUFFER_1 Base=r80.0 ElementSize=1 SrcRegion=<16;16,1> DstRegion=<1> Type=ub
289 .declare ubBUFFER_2 Base=r96.0 ElementSize=1 SrcRegion=<16;16,1> DstRegion=<1> Type=ub
290 .declare ubBUFFER_3 Base=r112.0 ElementSize=1 SrcRegion=<16;16,1> DstRegion=<1> Type=ub
291 .declare ubBUFFER_4 Base=r28.0 ElementSize=1 SrcRegion=<16;16,1> DstRegion=<1> Type=ub
292 .declare ubBUFFER_5 Base=r46.0 ElementSize=1 SrcRegion=<16;16,1> DstRegion=<1> Type=ub
293
294 .declare ub4BUFFER_0 Base=r64.0 ElementSize=1 SrcRegion=<32;8,4> DstRegion=<4> Type=ub
295 .declare ub4BUFFER_1 Base=r80.0 ElementSize=1 SrcRegion=<32;8,4> DstRegion=<4> Type=ub
296 .declare ub4BUFFER_2 Base=r96.0 ElementSize=1 SrcRegion=<32;8,4> DstRegion=<4> Type=ub
297 .declare ub4BUFFER_3 Base=r112.0 ElementSize=1 SrcRegion=<32;8,4> DstRegion=<4> Type=ub
298 .declare ub4BUFFER_4 Base=r28.0 ElementSize=1 SrcRegion=<32;8,4> DstRegion=<4> Type=ub
299 .declare ub4BUFFER_5 Base=r46.0 ElementSize=1 SrcRegion=<32;8,4> DstRegion=<4> Type=ub
300
301 //Pointer to mask reg
302
303
304 //r18
305
306
307 //Always keep Cannel Pointers and Offsets in same GRF, so that we can use
308 // NODDCLR, NODDCHK flags. -rT
309
310
311 .declare udCSC_COEFF_0 Base=r18.0 ElementSize=4 Type=ud // 1 GRF
312
313 //r19
314
315
316 .declare udCSC_COEFF_1 Base=r19.0 ElementSize=4 Type=ud // 1 GRF
317
318
319 //r20
320
321 .declare uwALPHA_MASK_REG_TEMP Base=r20.0 ElementSize=2 SrcRegion=<16;16,1> Type=uw // 1 GRF
322
323 //r21
324
325 .declare uwALPHA_MASK_REG Base=r21.0 ElementSize=2 SrcRegion=<16;16,1> Type=uw // 1 GRF
326
327 //r22
328
329
330 //Always keep Cannel Pointers and Offsets in same GRF, so that we can use
331 // NODDCLR, NODDCHK flags. -rT
332
333
334 //Keep fORIGIN_X_NLAS, fY_OFFSET_2ND_BLOCK, fSTEP_X_NLAS, pMSGDSC_COPY, ubCONST_ALPHA_COPY as
335 //sub registers of same GRF to enable using NODDCLR NODDCHK. -rT
336
337 //r23
338
339
340 //Lumakey
341
342
343 //r24
344
345
346 //r25
347
348
349 //r26
350
351
352 //defines to generate LABELS during compile time.
353
354
355 //Msg payload buffers; upto 4 full-size messages can be written
356
357
358 .declare mudMSGPAYLOAD0 Base=r29.0 ElementSize=4 SrcRegion=<8;8,1> Type=ud
359 .declare mudMSGPAYLOAD1 Base=r38.0 ElementSize=4 SrcRegion=<8;8,1> Type=ud
360 .declare mudMSGPAYLOAD2 Base=r47.0 ElementSize=4 SrcRegion=<8;8,1> Type=ud
361 .declare mudMSGPAYLOAD3 Base=r56.0 ElementSize=4 SrcRegion=<8;8,1> Type=ud
362
363 .declare muwMSGPAYLOAD0 Base=r29.0 ElementSize=2 SrcRegion=<16;16,1> Type=uw
364 .declare muwMSGPAYLOAD1 Base=r38.0 ElementSize=2 SrcRegion=<16;16,1> Type=uw
365 .declare muwMSGPAYLOAD2 Base=r47.0 ElementSize=2 SrcRegion=<16;16,1> Type=uw
366 .declare muwMSGPAYLOAD3 Base=r56.0 ElementSize=2 SrcRegion=<16;16,1> Type=uw
367
368 .declare mubMSGPAYLOAD0 Base=r29.0 ElementSize=1 SrcRegion=<16;16,1> Type=ub
369 .declare mubMSGPAYLOAD1 Base=r38.0 ElementSize=1 SrcRegion=<16;16,1> Type=ub
370 .declare mubMSGPAYLOAD2 Base=r47.0 ElementSize=1 SrcRegion=<16;16,1> Type=ub
371 .declare mubMSGPAYLOAD3 Base=r56.0 ElementSize=1 SrcRegion=<16;16,1> Type=ub
372 .declare mubMSGPAYLOAD4 Base=r32.0 ElementSize=1 SrcRegion=<16;16,1> Type=ub
373 .declare mubMSGPAYLOAD5 Base=r41.0 ElementSize=1 SrcRegion=<16;16,1> Type=ub
374 .declare mubMSGPAYLOAD6 Base=r50.0 ElementSize=1 SrcRegion=<16;16,1> Type=ub
375 .declare mubMSGPAYLOAD7 Base=r59.0 ElementSize=1 SrcRegion=<16;16,1> Type=ub
376
377
378 // the r17 register (nTEMP0) is originally defined from "Common.inc"
379 // instead of re-defining a nTEMP0 here, we use "SAVE_RGB" suffix for its naming
380
381 .declare uwTemp0 Base=r17.0 ElementSize=2 Type=uw
382
383
384 //_SAVE_INC_
385
386
387 // At the save module we have all 8 address sub-registers available.
388 // So we will use PING-PONG type of scheme to save the data using
389 // pointers pBUF_CHNL_TOP_8x4 and pBUF_CHNL_BOT_8x4. This will help
390 // reduce dependency. - rT
391
392 //Internal LAYOUT:(RRGGBBAA)
393 //Assign buffer channel order for Buffer 0123 in the order RGBA a0.3>A, a0.2>B, a0.1>G, a0.0>R
394 // R = 0, G= 4, B = 8, A = 12.
395 mov (4) r22.0<1>:w 0x62EA:v
396 add (4) r22.0<1>:w r22.0<4;4,1>:w 70:uw
397 shl (4) r22.0<1>:w r22.0<4;4,1>:w 5:uw
398
399 // if channel swap?
400 // This means that it should be BGRA(B is the LSB) or RGBA
401 // the internal format is always RGBA(MSB-A-B-G-R).
402 and.nz.f0.0 null<1>:w r2.3<0;1,0>:uw 0x01:w
403
404 //wBUFF_CHNL_PTR points to either buffer 0 or buffer 4.
405 //Add appropriate offsets to get pointers for all buffers (1,2,3 or 5).
406 //Offsets are zero for buffer 0 and buffer 4.
407 add (4) a0.0<1>:uw r22.0<4;4,1>:w 0:uw
408
409 // pointer swap
410 (f0.0) mov (1) uwTemp0<1> a0.0<0;1,0>:uw
411 (f0.0) mov (1) a0.0<1>:uw a0.2<0;1,0>:uw
412 (f0.0) mov (1) a0.2<1>:uw uwTemp0<0;1,0>
413
414 shl (1) r27.0<1>:d r9.0<0;1,0>:w 2:w { NoDDClr } // H. block origin need to be quadrupled
415 mov (1) r27.1<1>:d r9.1<0;1,0>:w { NoDDClr, NoDDChk } // Block origin (1st quadrant)
416 mov (1) r27.2<1>:ud 0x3001F:ud { NoDDChk } // Block width and height (32x4)
417
418 mov (4) a0.4<1>:uw a0.0<4;4,1>:uw
419
420 mov (8) r28<1>:ud r27<8;8,1>:ud
421 mov (8) r37<1>:ud r27<8;8,1>:ud
422 mov (8) r46<1>:ud r27<8;8,1>:ud
423 mov (8) r55<1>:ud r27<8;8,1>:ud
424
425 mov (8) r31<1>:ud r27<8;8,1>:ud
426 mov (8) r40<1>:ud r27<8;8,1>:ud
427 mov (8) r49<1>:ud r27<8;8,1>:ud
428 mov (8) r58<1>:ud r27<8;8,1>:ud
429
430 //Buffer 0/1 are written by using 4 32x4.
431
432 add (1) r37.0<1>:d r27.0<0;1,0>:d 32:d
433
434 add (1) r46.1<1>:d r27.1<0;1,0>:d 4:d
435
436 add (1) r55.1<1>:d r27.1<0;1,0>:d 4:d
437 add (1) r55.0<1>:d r27.0<0;1,0>:d 32:d
438
439 // write Buf_0 to 1st quarter of four horizontal output blocks
440
441 // Please note the scattered order of NODDCLR, NODDCHK flags. Since the sub-registers
442 // of destination reg are not updated at one place and hence even flags are scattered. -rT
443
444 /* for block 0 the left part of buffer 0 and 1 */
445 mov (8) mubMSGPAYLOAD0(0, 0)<4> r[a0.0, 1]<16;8,2>:ub
446 mov (8) mubMSGPAYLOAD0(0, 1)<4> r[a0.1, 1]<16;8,2>:ub
447 mov (8) mubMSGPAYLOAD0(0, 2)<4> r[a0.2, 1]<16;8,2>:ub
448 mov (8) mubMSGPAYLOAD0(0, 3)<4> r2.31:ub
449
450 mov (8) mubMSGPAYLOAD0(1, 0)<4> r[a0.0, 33]<16;8,2>:ub
451 mov (8) mubMSGPAYLOAD0(1, 1)<4> r[a0.1, 33]<16;8,2>:ub
452 mov (8) mubMSGPAYLOAD0(1, 2)<4> r[a0.2, 33]<16;8,2>:ub
453 mov (8) mubMSGPAYLOAD0(1, 3)<4> r2.31:ub
454
455 mov (8) mubMSGPAYLOAD1(0, 0)<4> r[a0.0, 17]<16;8,2>:ub
456 mov (8) mubMSGPAYLOAD1(0, 1)<4> r[a0.1, 17]<16;8,2>:ub
457 mov (8) mubMSGPAYLOAD1(0, 2)<4> r[a0.2, 17]<16;8,2>:ub
458 mov (8) mubMSGPAYLOAD1(0, 3)<4> r2.31:ub
459
460 mov (8) mubMSGPAYLOAD1(1, 0)<4> r[a0.0, 49]<16;8,2>:ub
461 mov (8) mubMSGPAYLOAD1(1, 1)<4> r[a0.1, 49]<16;8,2>:ub
462 mov (8) mubMSGPAYLOAD1(1, 2)<4> r[a0.2, 49]<16;8,2>:ub
463 mov (8) mubMSGPAYLOAD1(1, 3)<4> r2.31:ub
464
465 mov (8) mubMSGPAYLOAD0(2, 0)<4> r[a0.0, 65]<16;8,2>:ub
466 mov (8) mubMSGPAYLOAD0(2, 1)<4> r[a0.1, 65]<16;8,2>:ub
467 mov (8) mubMSGPAYLOAD0(2, 2)<4> r[a0.2, 65]<16;8,2>:ub
468 mov (8) mubMSGPAYLOAD0(2, 3)<4> r2.31:ub
469
470 mov (8) mubMSGPAYLOAD0(3, 0)<4> r[a0.0, 97]<16;8,2>:ub
471 mov (8) mubMSGPAYLOAD0(3, 1)<4> r[a0.1, 97]<16;8,2>:ub
472 mov (8) mubMSGPAYLOAD0(3, 2)<4> r[a0.2, 97]<16;8,2>:ub
473 mov (8) mubMSGPAYLOAD0(3, 3)<4> r2.31:ub
474
475 mov (8) mubMSGPAYLOAD1(2, 0)<4> r[a0.0, 81]<16;8,2>:ub
476 mov (8) mubMSGPAYLOAD1(2, 1)<4> r[a0.1, 81]<16;8,2>:ub
477 mov (8) mubMSGPAYLOAD1(2, 2)<4> r[a0.2, 81]<16;8,2>:ub
478 mov (8) mubMSGPAYLOAD1(2, 3)<4> r2.31:ub
479
480 mov (8) mubMSGPAYLOAD1(3, 0)<4> r[a0.0, 113]<16;8,2>:ub
481 mov (8) mubMSGPAYLOAD1(3, 1)<4> r[a0.1, 113]<16;8,2>:ub
482 mov (8) mubMSGPAYLOAD1(3, 2)<4> r[a0.2, 113]<16;8,2>:ub
483 mov (8) mubMSGPAYLOAD1(3, 3)<4> r2.31:ub
484
485 /* For Buffer 0 */
486 send (16) null<1>:d r28 0xc 0x0A0A8018:ud
487 send (16) null<1>:d r37 0xc 0x0A0A8018:ud
488
489 add (4) a0.0<1>:uw a0.4<4;4,1>:uw 512:uw
490 mov (8) mubMSGPAYLOAD2(0, 0)<4> r[a0.0, 1]<16;8,2>:ub
491 mov (8) mubMSGPAYLOAD2(0, 1)<4> r[a0.1, 1]<16;8,2>:ub
492 mov (8) mubMSGPAYLOAD2(0, 2)<4> r[a0.2, 1]<16;8,2>:ub
493 mov (8) mubMSGPAYLOAD2(0, 3)<4> r2.31:ub
494
495 mov (8) mubMSGPAYLOAD2(1, 0)<4> r[a0.0, 33]<16;8,2>:ub
496 mov (8) mubMSGPAYLOAD2(1, 1)<4> r[a0.1, 33]<16;8,2>:ub
497 mov (8) mubMSGPAYLOAD2(1, 2)<4> r[a0.2, 33]<16;8,2>:ub
498 mov (8) mubMSGPAYLOAD2(1, 3)<4> r2.31:ub
499
500 mov (8) mubMSGPAYLOAD3(0, 0)<4> r[a0.0, 17]<16;8,2>:ub
501 mov (8) mubMSGPAYLOAD3(0, 1)<4> r[a0.1, 17]<16;8,2>:ub
502 mov (8) mubMSGPAYLOAD3(0, 2)<4> r[a0.2, 17]<16;8,2>:ub
503 mov (8) mubMSGPAYLOAD3(0, 3)<4> r2.31:ub
504
505 mov (8) mubMSGPAYLOAD3(1, 0)<4> r[a0.0, 49]<16;8,2>:ub
506 mov (8) mubMSGPAYLOAD3(1, 1)<4> r[a0.1, 49]<16;8,2>:ub
507 mov (8) mubMSGPAYLOAD3(1, 2)<4> r[a0.2, 49]<16;8,2>:ub
508 mov (8) mubMSGPAYLOAD3(1, 3)<4> r2.31:ub
509
510 mov (8) mubMSGPAYLOAD2(2, 0)<4> r[a0.0, 65]<16;8,2>:ub
511 mov (8) mubMSGPAYLOAD2(2, 1)<4> r[a0.1, 65]<16;8,2>:ub
512 mov (8) mubMSGPAYLOAD2(2, 2)<4> r[a0.2, 65]<16;8,2>:ub
513 mov (8) mubMSGPAYLOAD2(2, 3)<4> r2.31:ub
514
515 mov (8) mubMSGPAYLOAD2(3, 0)<4> r[a0.0, 97]<16;8,2>:ub
516 mov (8) mubMSGPAYLOAD2(3, 1)<4> r[a0.1, 97]<16;8,2>:ub
517 mov (8) mubMSGPAYLOAD2(3, 2)<4> r[a0.2, 97]<16;8,2>:ub
518 mov (8) mubMSGPAYLOAD2(3, 3)<4> r2.31:ub
519
520 mov (8) mubMSGPAYLOAD3(2, 0)<4> r[a0.0, 81]<16;8,2>:ub
521 mov (8) mubMSGPAYLOAD3(2, 1)<4> r[a0.1, 81]<16;8,2>:ub
522 mov (8) mubMSGPAYLOAD3(2, 2)<4> r[a0.2, 81]<16;8,2>:ub
523 mov (8) mubMSGPAYLOAD3(2, 3)<4> r2.31:ub
524
525 mov (8) mubMSGPAYLOAD3(3, 0)<4> r[a0.0, 113]<16;8,2>:ub
526 mov (8) mubMSGPAYLOAD3(3, 1)<4> r[a0.1, 113]<16;8,2>:ub
527 mov (8) mubMSGPAYLOAD3(3, 2)<4> r[a0.2, 113]<16;8,2>:ub
528 mov (8) mubMSGPAYLOAD3(3, 3)<4> r2.31:ub
529 // send Buffer 1
530 send (16) null<1>:d r46 0xc 0x0A0A8018:ud
531 send (16) null<1>:d r55 0xc 0x0A0A8018:ud
532
533
534 /* for Buffer 2/3 */
535 mov (8) r28<1>:ud r27<8;8,1>:ud
536 mov (8) r37<1>:ud r27<8;8,1>:ud
537 mov (8) r46<1>:ud r27<8;8,1>:ud
538 mov (8) r55<1>:ud r27<8;8,1>:ud
539
540 add (1) r28.1<1>:d r27.1<0;1,0>:d 8:d
541
542 add (1) r37.0<1>:d r27.0<0;1,0>:d 32:d
543 add (1) r37.1<1>:d r27.1<0;1,0>:d 8:d
544
545 add (1) r46.1<1>:d r27.1<0;1,0>:d 12:d
546
547 add (1) r55.1<1>:d r27.1<0;1,0>:d 12:d
548 add (1) r55.0<1>:d r27.0<0;1,0>:d 32:d
549
550 add (4) a0.0<1>:uw a0.4<4;4,1>:uw 1024:uw
551
552 mov (8) mubMSGPAYLOAD0(0, 0)<4> r[a0.0, 1]<16;8,2>:ub
553 mov (8) mubMSGPAYLOAD0(0, 1)<4> r[a0.1, 1]<16;8,2>:ub
554 mov (8) mubMSGPAYLOAD0(0, 2)<4> r[a0.2, 1]<16;8,2>:ub
555 mov (8) mubMSGPAYLOAD0(0, 3)<4> r2.31:ub
556
557 mov (8) mubMSGPAYLOAD0(1, 0)<4> r[a0.0, 33]<16;8,2>:ub
558 mov (8) mubMSGPAYLOAD0(1, 1)<4> r[a0.1, 33]<16;8,2>:ub
559 mov (8) mubMSGPAYLOAD0(1, 2)<4> r[a0.2, 33]<16;8,2>:ub
560 mov (8) mubMSGPAYLOAD0(1, 3)<4> r2.31:ub
561
562 mov (8) mubMSGPAYLOAD1(0, 0)<4> r[a0.0, 17]<16;8,2>:ub
563 mov (8) mubMSGPAYLOAD1(0, 1)<4> r[a0.1, 17]<16;8,2>:ub
564 mov (8) mubMSGPAYLOAD1(0, 2)<4> r[a0.2, 17]<16;8,2>:ub
565 mov (8) mubMSGPAYLOAD1(0, 3)<4> r2.31:ub
566
567 mov (8) mubMSGPAYLOAD1(1, 0)<4> r[a0.0, 49]<16;8,2>:ub
568 mov (8) mubMSGPAYLOAD1(1, 1)<4> r[a0.1, 49]<16;8,2>:ub
569 mov (8) mubMSGPAYLOAD1(1, 2)<4> r[a0.2, 49]<16;8,2>:ub
570 mov (8) mubMSGPAYLOAD1(1, 3)<4> r2.31:ub
571
572 mov (8) mubMSGPAYLOAD0(2, 0)<4> r[a0.0, 65]<16;8,2>:ub
573 mov (8) mubMSGPAYLOAD0(2, 1)<4> r[a0.1, 65]<16;8,2>:ub
574 mov (8) mubMSGPAYLOAD0(2, 2)<4> r[a0.2, 65]<16;8,2>:ub
575 mov (8) mubMSGPAYLOAD0(2, 3)<4> r2.31:ub
576
577 mov (8) mubMSGPAYLOAD0(3, 0)<4> r[a0.0, 97]<16;8,2>:ub
578 mov (8) mubMSGPAYLOAD0(3, 1)<4> r[a0.1, 97]<16;8,2>:ub
579 mov (8) mubMSGPAYLOAD0(3, 2)<4> r[a0.2, 97]<16;8,2>:ub
580 mov (8) mubMSGPAYLOAD0(3, 3)<4> r2.31:ub
581
582 mov (8) mubMSGPAYLOAD1(2, 0)<4> r[a0.0, 81]<16;8,2>:ub
583 mov (8) mubMSGPAYLOAD1(2, 1)<4> r[a0.1, 81]<16;8,2>:ub
584 mov (8) mubMSGPAYLOAD1(2, 2)<4> r[a0.2, 81]<16;8,2>:ub
585 mov (8) mubMSGPAYLOAD1(2, 3)<4> r2.31:ub
586
587 mov (8) mubMSGPAYLOAD1(3, 0)<4> r[a0.0, 113]<16;8,2>:ub
588 mov (8) mubMSGPAYLOAD1(3, 1)<4> r[a0.1, 113]<16;8,2>:ub
589 mov (8) mubMSGPAYLOAD1(3, 2)<4> r[a0.2, 113]<16;8,2>:ub
590 mov (8) mubMSGPAYLOAD1(3, 3)<4> r2.31:ub
591
592 // Send Buffer 2
593 send (16) null<1>:d r28 0xc 0x0A0A8018:ud
594 send (16) null<1>:d r37 0xc 0x0A0A8018:ud
595
596 add (4) a0.0<1>:uw a0.4<4;4,1>:uw 1536:uw
597 mov (8) mubMSGPAYLOAD2(0, 0)<4> r[a0.0, 1]<16;8,2>:ub
598 mov (8) mubMSGPAYLOAD2(0, 1)<4> r[a0.1, 1]<16;8,2>:ub
599 mov (8) mubMSGPAYLOAD2(0, 2)<4> r[a0.2, 1]<16;8,2>:ub
600 mov (8) mubMSGPAYLOAD2(0, 3)<4> r2.31:ub
601
602 mov (8) mubMSGPAYLOAD2(1, 0)<4> r[a0.0, 33]<16;8,2>:ub
603 mov (8) mubMSGPAYLOAD2(1, 1)<4> r[a0.1, 33]<16;8,2>:ub
604 mov (8) mubMSGPAYLOAD2(1, 2)<4> r[a0.2, 33]<16;8,2>:ub
605 mov (8) mubMSGPAYLOAD2(1, 3)<4> r2.31:ub
606
607 mov (8) mubMSGPAYLOAD3(0, 0)<4> r[a0.0, 17]<16;8,2>:ub
608 mov (8) mubMSGPAYLOAD3(0, 1)<4> r[a0.1, 17]<16;8,2>:ub
609 mov (8) mubMSGPAYLOAD3(0, 2)<4> r[a0.2, 17]<16;8,2>:ub
610 mov (8) mubMSGPAYLOAD3(0, 3)<4> r2.31:ub
611
612 mov (8) mubMSGPAYLOAD3(1, 0)<4> r[a0.0, 49]<16;8,2>:ub
613 mov (8) mubMSGPAYLOAD3(1, 1)<4> r[a0.1, 49]<16;8,2>:ub
614 mov (8) mubMSGPAYLOAD3(1, 2)<4> r[a0.2, 49]<16;8,2>:ub
615 mov (8) mubMSGPAYLOAD3(1, 3)<4> r2.31:ub
616
617 mov (8) mubMSGPAYLOAD2(2, 0)<4> r[a0.0, 65]<16;8,2>:ub
618 mov (8) mubMSGPAYLOAD2(2, 1)<4> r[a0.1, 65]<16;8,2>:ub
619 mov (8) mubMSGPAYLOAD2(2, 2)<4> r[a0.2, 65]<16;8,2>:ub
620 mov (8) mubMSGPAYLOAD2(2, 3)<4> r2.31:ub
621
622 mov (8) mubMSGPAYLOAD2(3, 0)<4> r[a0.0, 97]<16;8,2>:ub
623 mov (8) mubMSGPAYLOAD2(3, 1)<4> r[a0.1, 97]<16;8,2>:ub
624 mov (8) mubMSGPAYLOAD2(3, 2)<4> r[a0.2, 97]<16;8,2>:ub
625 mov (8) mubMSGPAYLOAD2(3, 3)<4> r2.31:ub
626
627 mov (8) mubMSGPAYLOAD3(2, 0)<4> r[a0.0, 81]<16;8,2>:ub
628 mov (8) mubMSGPAYLOAD3(2, 1)<4> r[a0.1, 81]<16;8,2>:ub
629 mov (8) mubMSGPAYLOAD3(2, 2)<4> r[a0.2, 81]<16;8,2>:ub
630 mov (8) mubMSGPAYLOAD3(2, 3)<4> r2.31:ub
631
632 mov (8) mubMSGPAYLOAD3(3, 0)<4> r[a0.0, 113]<16;8,2>:ub
633 mov (8) mubMSGPAYLOAD3(3, 1)<4> r[a0.1, 113]<16;8,2>:ub
634 mov (8) mubMSGPAYLOAD3(3, 2)<4> r[a0.2, 113]<16;8,2>:ub
635 mov (8) mubMSGPAYLOAD3(3, 3)<4> r2.31:ub
636 // send buffer 3
637 send (16) null<1>:d r46 0xc 0x0A0A8018:ud
638 send (16) null<1>:d r55 0xc 0x0A0A8018:ud
639
640
641