Debian Codegraph @

src/shaders/post_processing/gen7/PA_AVS_Buf_3.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 // 37 // Total instruction count
39 // 1 // Total kernel count
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 // FileName: PA_AVS_Buf_3.asm
186 // Author: Vivek Kumar
187 // Description: Loads 8x8 AVS/IEF Packed data into Buffer 3
188
189
190
191 // FileName : PA_AVS_Buf.asm
192 // Author : Tatiya, Rupesh
193 // Description : Loads 8x8 AVS/IEF Packed data into Buffer N
194
195 //On IVB, for AVS module - set buffer pointers offset according to AVS Layout.
196 //Change it to Sample Unorm layout in Shuffle modules.
197
198
199
200 // Module name: Scaling.inc
201
202
203
204
205 // Description: Includes all definitions explicit to Fast Composite.
206
207
208
209
210 // End of common.inc
211
212
213 //========== GRF partition ==========
214 // r0 header : r0 (1 GRF)
215 // Static parameters : r1 - r6 (6 GRFS)
216 // Inline parameters : r7 - r8 (2 GRFs)
217 // MSGSRC : r27 (1 GRF)
218 //===================================
219
220 //Interface:
221 //========== Static Parameters (Explicit To Fast Composite) ==========
222 //r1
223 //CSC Set 0
224
225
226 .declare udCSC_CURBE Base=r1.0 ElementSize=4 Type=ud
227
228 //Constant alpha
229
230
231 //r2
232
233
234 // Gen7 AVS WA
235
236
237 // WiDi Definitions
238
239
240 //Colorfill
241
242
243 // 0: 0-degree, 1: 90, 2: 180, 3: 270-degree, clockwise.
244
245 .declare ubCOLOR_PIXEL_VAL Base=r2.20 ElementSize=1 SrcRegion=<0;1,0> DstRegion=<1> Type=ub
246
247 //r3
248 //Normalised Ratio of Horizontal step size with main video for all layers
249
250
251 //Normalised Ratio of Horizontal step size with main video for all layers becomes
252 //Normalised Horizontal step size for all layers in VP_Setup.asm
253
254
255 //r4
256 //Normalised Vertical step size for all layers
257
258
259 //r5
260 //Normalised Vertical Frame Origin for all layers
261
262
263 //r6
264 //Normalised Horizontal Frame Origin for all layers
265
266
267 //========== Inline Parameters (Explicit To Fast Composite) ==========
268
269
270 //Main video Step X
271
272
273 //====================== Binding table (Explicit To Fast Composite)=========================================
274
275
276 //Used by Interlaced Scaling Kernels
277
278
279 //========== Sampler State Table Index (Explicit To Fast Composite)==========
280 //Sampler Index for AVS/IEF messages
281
282
283 //Sampler Index for SIMD16 sampler messages
284
285
286 //=============================================================================
287
288 .declare fBUFFER_0 Base=r64.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=f
289 .declare fBUFFER_1 Base=r80.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=f
290 .declare fBUFFER_2 Base=r96.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=f
291 .declare fBUFFER_3 Base=r112.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=f
292 .declare fBUFFER_4 Base=r28.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=f
293 .declare fBUFFER_5 Base=r46.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=f
294
295 .declare udBUFFER_0 Base=r64.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=ud
296 .declare udBUFFER_1 Base=r80.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=ud
297 .declare udBUFFER_2 Base=r96.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=ud
298 .declare udBUFFER_3 Base=r112.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=ud
299 .declare udBUFFER_4 Base=r28.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=ud
300 .declare udBUFFER_5 Base=r46.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=ud
301
302 .declare uwBUFFER_0 Base=r64.0 ElementSize=2 SrcRegion=<16;16,1> DstRegion=<1> Type=uw
303 .declare uwBUFFER_1 Base=r80.0 ElementSize=2 SrcRegion=<16;16,1> DstRegion=<1> Type=uw
304 .declare uwBUFFER_2 Base=r96.0 ElementSize=2 SrcRegion=<16;16,1> DstRegion=<1> Type=uw
305 .declare uwBUFFER_3 Base=r112.0 ElementSize=2 SrcRegion=<16;16,1> DstRegion=<1> Type=uw
306 .declare uwBUFFER_4 Base=r28.0 ElementSize=2 SrcRegion=<16;16,1> DstRegion=<1> Type=uw
307 .declare uwBUFFER_5 Base=r46.0 ElementSize=2 SrcRegion=<16;16,1> DstRegion=<1> Type=uw
308
309 .declare ubBUFFER_0 Base=r64.0 ElementSize=1 SrcRegion=<16;16,1> DstRegion=<1> Type=ub
310 .declare ubBUFFER_1 Base=r80.0 ElementSize=1 SrcRegion=<16;16,1> DstRegion=<1> Type=ub
311 .declare ubBUFFER_2 Base=r96.0 ElementSize=1 SrcRegion=<16;16,1> DstRegion=<1> Type=ub
312 .declare ubBUFFER_3 Base=r112.0 ElementSize=1 SrcRegion=<16;16,1> DstRegion=<1> Type=ub
313 .declare ubBUFFER_4 Base=r28.0 ElementSize=1 SrcRegion=<16;16,1> DstRegion=<1> Type=ub
314 .declare ubBUFFER_5 Base=r46.0 ElementSize=1 SrcRegion=<16;16,1> DstRegion=<1> Type=ub
315
316 .declare ub4BUFFER_0 Base=r64.0 ElementSize=1 SrcRegion=<32;8,4> DstRegion=<4> Type=ub
317 .declare ub4BUFFER_1 Base=r80.0 ElementSize=1 SrcRegion=<32;8,4> DstRegion=<4> Type=ub
318 .declare ub4BUFFER_2 Base=r96.0 ElementSize=1 SrcRegion=<32;8,4> DstRegion=<4> Type=ub
319 .declare ub4BUFFER_3 Base=r112.0 ElementSize=1 SrcRegion=<32;8,4> DstRegion=<4> Type=ub
320 .declare ub4BUFFER_4 Base=r28.0 ElementSize=1 SrcRegion=<32;8,4> DstRegion=<4> Type=ub
321 .declare ub4BUFFER_5 Base=r46.0 ElementSize=1 SrcRegion=<32;8,4> DstRegion=<4> Type=ub
322
323 //Pointer to mask reg
324
325
326 //r18
327
328
329 //Always keep Cannel Pointers and Offsets in same GRF, so that we can use
330 // NODDCLR, NODDCHK flags. -rT
331
332
333 .declare udCSC_COEFF_0 Base=r18.0 ElementSize=4 Type=ud // 1 GRF
334
335 //r19
336
337
338 .declare udCSC_COEFF_1 Base=r19.0 ElementSize=4 Type=ud // 1 GRF
339
340
341 //r20
342
343 .declare uwALPHA_MASK_REG_TEMP Base=r20.0 ElementSize=2 SrcRegion=<16;16,1> Type=uw // 1 GRF
344
345 //r21
346
347 .declare uwALPHA_MASK_REG Base=r21.0 ElementSize=2 SrcRegion=<16;16,1> Type=uw // 1 GRF
348
349 //r22
350
351
352 //Always keep Cannel Pointers and Offsets in same GRF, so that we can use
353 // NODDCLR, NODDCHK flags. -rT
354
355
356 //Keep fORIGIN_X_NLAS, fY_OFFSET_2ND_BLOCK, fSTEP_X_NLAS, pMSGDSC_COPY, ubCONST_ALPHA_COPY as
357 //sub registers of same GRF to enable using NODDCLR NODDCHK. -rT
358
359 //r23
360
361
362 //Lumakey
363
364
365 //r24
366
367
368 //r25
369
370
371 //r26
372
373
374 //defines to generate LABELS during compile time.
375
376
377 // Message Header
378 // m0.7 31:0 Debug
379 // m0.6 31:0 Debug
380 // m0.5 31:0 Ignored
381 // m0.4 31:0 Ignored
382 // m0.3 31:0 Ignored
383 // m0.2 31:16 Ignored
384 // 15 Alpha Write Channel Mask enable=0, disable=1
385 // 14 Blue Write Channel Mask (U)
386 // 13 Green Write Channel Mask (Y)
387 // 12 Red Write Channel Mask (V)
388 // 11:0 Ignored
389 // m0.1 Ignored
390 // m0.0 Ignored
391
392
393 // AVS payload
394 // m1.7 Group ID Number
395 // m1.6 U 2nd Derivative ---> NLAS dx
396 // m1.5 Delta V ---> Step Y
397 // m1.4 Delta U ---> Step X
398 // m1.3 Pixel 0 V Address ---> ORIY (Y0)
399 // m1.2 Pixel 0 U Address ---> ORIX (X0)
400 // m1.1 Vertical Block Number
401 // m1.0 Reserved
402
403 // Sampler Message Descriptor
404 // 31:29 Reserved 000
405 // 28:25 Message length 0010
406 // 24:20 Response length xxxxx ---> 4GRFs for each enabled channel (AVS), 2GRFs for each enabled channel (sample unorm)
407 // 19 Header Present 1
408 // 18:17 SIMD Mode 11 ---> SIMD32/64
409 // 16:12 Message Type xxxxx ---> 01011 sample_8x8, 01100 (sample_unorm), 01010 (sample_unorm+killpix)
410 // 11:8 Sampler Index xxxx
411 // 7:0 Binding Table Index xxxxxxxx
412
413
414 // Msg Header M0.2
415 // 15:15 Alpha Write Channel Mask, 0: written back, 1: not written back
416 // 14:14 Blue Write Channel Mask
417 // 13:13 Green Write Channel Mask
418 // 12:12 Red Write Channel Mask
419
420
421 //By design, Buffer 0,1,2,3 always have Layer 0 and Buffer 4,5 always have L1-L7
422
423
424 //used to generate LABELS at compile time.
425
426
427 // 18:17 SIMD Mode 10 ---> SIMD16
428 // 16:12 Message Type xxxxx ---> 00000 (SIMD16)
429
430
431 //r10-17 - 8 GRFs to load SIMD16 data (upto 4 channels)
432 //r18-19 - 2 GRFs to store sampler ramp.
433
434 .declare mfSCALING_0X_34X_PAYLOAD Base=r14.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=f
435 .declare muwSCALING_0X_34X_PAYLOAD Base=r14.0 ElementSize=2 SrcRegion=<16;16,1> DstRegion=<1> Type=uw
436 .declare mudCALING_0X_34X_PAYLOAD Base=r14.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=ud
437 .declare mubCALING_0X_34X_PAYLOAD Base=r14.0 ElementSize=1 SrcRegion=<32;32,1> DstRegion=<1> Type=ub
438
439
440 .declare fSCALING_0X_34X_TEMP Base=r11.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=f
441 .declare udSCALING_0X_34X_TEMP Base=r11.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=ud
442 .declare ub4SCALING_0X_34X_TEMP Base=r11.0 ElementSize=1 SrcRegion=<32;8,4> DstRegion=<1> Type=ub
443 .declare uwSCALING_0X_34X_TEMP Base=r11.0 ElementSize=2 SrcRegion=<16;16,1> DstRegion=<1> Type=uw
444
445 // Sampler ramp is used for Scaling 0X_0.34X
446 .declare fSAMPLER_RAMP Base=r11.0 ElementSize=4 SrcRegion=<8;8,1> Type=f // 1 GRFs, 8 elements
447
448
449 //#define rMSGDSC_UV r23.0
450
451
452 //End of _SCALING_
453
454
455 //Check if layer is to be skipped
456
457
458 // f0.1 pre-computed in Set_Layer_0
459 (-f0.1) jmpi (1) SKIP_AVS_LOAD_L0_3_
460
461
462 //AVS_PAYLOAD already has all the data loaded at this point
463 add (1) a0.0:ud r23.5<0;1,0>:ud 0x50EB400:ud //msg desc
464
465 mov (1) r16.2:ud 0x00000000:ud // Enable ARGB channels
466
467
468 // set the vertical block number
469
470
471 add (1) r25.1<1>:ud r9.12:uw 3:ud
472
473
474 mov (8) r17.0:ud r25.0<8;8,1>:ud
475
476 // Gen7 AVS WA Only for YUV packed surfaces, NV12 and Y-channel only for Planar surfaces
477 // if (((int)(u_left*width + 5.0/256) > (int)(u_left*width))
478 // {
479 // modified_u_coord = u_coord – 5.0/(256*width); //floating point
480 // }
481 // else if(((int)(u_left*width + 255.0/256) == (int)(u_left*width))
482 // {
483 // modified_u_coord = u_coord + 1.0/(256*width); //floating point
484 // }
485 // else{
486 // modified_u_coord = u_coord;
487 // }
488 // Where u_left = u – 2*du + 3*ddu for IEF On
489 // And u_left = u for IEF Off case
490 //
491
492 // check whether Gen7 AVS WA is enabled,
493 mov (1) r14.8:uw f0.0:uw // save f0.0
494
495 and.nz.f0.0 (1) null<1>:uw r2.3:uw 0x2:uw
496 (-f0.0)jmpi (1) GEN7_AVS_WA_DONE_L0_3_
497
498
499 // Gen7 AVS WA, check if IEF is ON for choosing Gen7 AVS WA formula
500
501 and.nz.f0.0 (8) null<1>:uw r2.3<0;1,0>:uw 0x4:uw
502 (f0.0)mov (8) acc0.0:f r17.2<0;1,0>:f
503 (f0.0)mac (8) acc0.0:f r17.4<0;1,0>:f -2.0:f
504 (f0.0)mac (8) acc0.0:f r17.6<0;1,0>:f 3.0:f
505 (f0.0)mov (1) r14.2:f acc0:f // IEF ON, rTEMP3.2 = u_left
506 (-f0.0)mov (1) r14.2:f r17.2<0;1,0>:f // IEF OFF, rTEMP3.2 = u_left
507
508 and (1) r14.1:ud r2.3:uw 0xFFF8:uw
509 asr (1) r14.1:ud r14.1:ud 3:d
510 mov (1) r14.1:f r14.1:ud
511
512 // Gen7 AVS WA, if (int)(u_left*width + 5.0/256) > (int)(u_left*width)
513 mul (1) r14.0:f r14.2:f r14.1:f // rTEMP3.0 = u_left*width
514 add (1) r14.2:f r14.0:f 0.01953125:f // rTEMP3.2 = u_left*width + 5.0/256
515 add (1) r14.3:f r14.0:f 0.99609375:f // rTEMP3.3 = u_left*width + 255.0/256
516
517 //Check if the values are < 0 and account for (int) cast of negative numbers
518
519 //(int)(u_left*width)
520 cmp.l.f0.0 (1) null<1>:f r14.0:f 0.00000000:f
521 mov (1) r14.0:d r14.0:f
522 (f0.0)add (1) r14.0:d r14.0<0;1,0>:d -1:d
523
524 //(int)(u_left*width + 5.0/256)
525 cmp.l.f0.0 (1) null<1>:f r14.2:f 0.00000000:f
526 mov (1) r14.2:d r14.2:f
527 (f0.0)add (1) r14.2:d r14.2<0;1,0>:d -1:d
528
529 //(int)(u_left*width + 255.0/256)
530 cmp.l.f0.0 (1) null<1>:f r14.3:f 0.00000000:f
531 mov (1) r14.3:d r14.3:f
532 (f0.0)add (1) r14.3:d r14.3<0;1,0>:d -1:d
533
534 mov (1) f0.0:uw 0:uw // clear flag
535 //if (((int)(u_left*width + 5.0/256) > (int)(u_left*width))
536 cmp.g.f1.0 (1) null<1>:d r14.2:d r14.0:d
537 // modified_u_coord = u_coord – 5.0/(256*width); //floating point
538 (f1.0) add (1) r17.2:f r17.2<0;1,0>:f -r2.3:f
539 //else if(((int)(u_left*width + 255.0/256) == (int)(u_left*width))
540 (-f1.0) cmp.e.f0.0 (1) null<1>:d r14.3:d r14.0:d
541 // modified_u_coord = u_coord + 1.0/(256*width); //floating point
542 (f0.0) add (1) r17.2:f r17.2<0;1,0>:f r2.2:f
543
544
545 GEN7_AVS_WA_DONE_L0_3_:
546 mov (1) f0.0:uw r14.8:uw // restore f0.0
547
548 send (1) uwBUFFER_3(0)<1> r16 0x2 a0.0:ud
549 // Returns packed data in 16 GRFs in scrambled order
550
551 SKIP_AVS_LOAD_L0_3_:
552 nop
553
554