Debian Codegraph @

src/shaders/post_processing/gen7/PL2_AVS_Buf_0.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 // 44 // 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: PL2_AVS_Buf_0.asm
186 // Author: Tatiya, Rupesh
187 // Description: Loads 8x8 AVS/IEF PL2 data into Buffer 0
188
189
190
191 // FileName : PL2_AVS_Buf.asm
192 // Author : Tatiya, Rupesh
193 // Description : Loads 8x8 AVS/IEF PL2 data into Buffer N
194
195
196
197 // Module name: Scaling.inc
198
199
200
201
202 // Description: Includes all definitions explicit to Fast Composite.
203
204
205
206
207 // End of common.inc
208
209
210 //========== GRF partition ==========
211 // r0 header : r0 (1 GRF)
212 // Static parameters : r1 - r6 (6 GRFS)
213 // Inline parameters : r7 - r8 (2 GRFs)
214 // MSGSRC : r27 (1 GRF)
215 //===================================
216
217 //Interface:
218 //========== Static Parameters (Explicit To Fast Composite) ==========
219 //r1
220 //CSC Set 0
221
222
223 .declare udCSC_CURBE Base=r1.0 ElementSize=4 Type=ud
224
225 //Constant alpha
226
227
228 //r2
229
230
231 // Gen7 AVS WA
232
233
234 // WiDi Definitions
235
236
237 //Colorfill
238
239
240 // 0: 0-degree, 1: 90, 2: 180, 3: 270-degree, clockwise.
241
242 .declare ubCOLOR_PIXEL_VAL Base=r2.20 ElementSize=1 SrcRegion=<0;1,0> DstRegion=<1> Type=ub
243
244 //r3
245 //Normalised Ratio of Horizontal step size with main video for all layers
246
247
248 //Normalised Ratio of Horizontal step size with main video for all layers becomes
249 //Normalised Horizontal step size for all layers in VP_Setup.asm
250
251
252 //r4
253 //Normalised Vertical step size for all layers
254
255
256 //r5
257 //Normalised Vertical Frame Origin for all layers
258
259
260 //r6
261 //Normalised Horizontal Frame Origin for all layers
262
263
264 //========== Inline Parameters (Explicit To Fast Composite) ==========
265
266
267 //Main video Step X
268
269
270 //====================== Binding table (Explicit To Fast Composite)=========================================
271
272
273 //Used by Interlaced Scaling Kernels
274
275
276 //========== Sampler State Table Index (Explicit To Fast Composite)==========
277 //Sampler Index for AVS/IEF messages
278
279
280 //Sampler Index for SIMD16 sampler messages
281
282
283 //=============================================================================
284
285 .declare fBUFFER_0 Base=r64.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=f
286 .declare fBUFFER_1 Base=r80.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=f
287 .declare fBUFFER_2 Base=r96.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=f
288 .declare fBUFFER_3 Base=r112.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=f
289 .declare fBUFFER_4 Base=r28.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=f
290 .declare fBUFFER_5 Base=r46.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=f
291
292 .declare udBUFFER_0 Base=r64.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=ud
293 .declare udBUFFER_1 Base=r80.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=ud
294 .declare udBUFFER_2 Base=r96.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=ud
295 .declare udBUFFER_3 Base=r112.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=ud
296 .declare udBUFFER_4 Base=r28.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=ud
297 .declare udBUFFER_5 Base=r46.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=ud
298
299 .declare uwBUFFER_0 Base=r64.0 ElementSize=2 SrcRegion=<16;16,1> DstRegion=<1> Type=uw
300 .declare uwBUFFER_1 Base=r80.0 ElementSize=2 SrcRegion=<16;16,1> DstRegion=<1> Type=uw
301 .declare uwBUFFER_2 Base=r96.0 ElementSize=2 SrcRegion=<16;16,1> DstRegion=<1> Type=uw
302 .declare uwBUFFER_3 Base=r112.0 ElementSize=2 SrcRegion=<16;16,1> DstRegion=<1> Type=uw
303 .declare uwBUFFER_4 Base=r28.0 ElementSize=2 SrcRegion=<16;16,1> DstRegion=<1> Type=uw
304 .declare uwBUFFER_5 Base=r46.0 ElementSize=2 SrcRegion=<16;16,1> DstRegion=<1> Type=uw
305
306 .declare ubBUFFER_0 Base=r64.0 ElementSize=1 SrcRegion=<16;16,1> DstRegion=<1> Type=ub
307 .declare ubBUFFER_1 Base=r80.0 ElementSize=1 SrcRegion=<16;16,1> DstRegion=<1> Type=ub
308 .declare ubBUFFER_2 Base=r96.0 ElementSize=1 SrcRegion=<16;16,1> DstRegion=<1> Type=ub
309 .declare ubBUFFER_3 Base=r112.0 ElementSize=1 SrcRegion=<16;16,1> DstRegion=<1> Type=ub
310 .declare ubBUFFER_4 Base=r28.0 ElementSize=1 SrcRegion=<16;16,1> DstRegion=<1> Type=ub
311 .declare ubBUFFER_5 Base=r46.0 ElementSize=1 SrcRegion=<16;16,1> DstRegion=<1> Type=ub
312
313 .declare ub4BUFFER_0 Base=r64.0 ElementSize=1 SrcRegion=<32;8,4> DstRegion=<4> Type=ub
314 .declare ub4BUFFER_1 Base=r80.0 ElementSize=1 SrcRegion=<32;8,4> DstRegion=<4> Type=ub
315 .declare ub4BUFFER_2 Base=r96.0 ElementSize=1 SrcRegion=<32;8,4> DstRegion=<4> Type=ub
316 .declare ub4BUFFER_3 Base=r112.0 ElementSize=1 SrcRegion=<32;8,4> DstRegion=<4> Type=ub
317 .declare ub4BUFFER_4 Base=r28.0 ElementSize=1 SrcRegion=<32;8,4> DstRegion=<4> Type=ub
318 .declare ub4BUFFER_5 Base=r46.0 ElementSize=1 SrcRegion=<32;8,4> DstRegion=<4> Type=ub
319
320 //Pointer to mask reg
321
322
323 //r18
324
325
326 //Always keep Cannel Pointers and Offsets in same GRF, so that we can use
327 // NODDCLR, NODDCHK flags. -rT
328
329
330 .declare udCSC_COEFF_0 Base=r18.0 ElementSize=4 Type=ud // 1 GRF
331
332 //r19
333
334
335 .declare udCSC_COEFF_1 Base=r19.0 ElementSize=4 Type=ud // 1 GRF
336
337
338 //r20
339
340 .declare uwALPHA_MASK_REG_TEMP Base=r20.0 ElementSize=2 SrcRegion=<16;16,1> Type=uw // 1 GRF
341
342 //r21
343
344 .declare uwALPHA_MASK_REG Base=r21.0 ElementSize=2 SrcRegion=<16;16,1> Type=uw // 1 GRF
345
346 //r22
347
348
349 //Always keep Cannel Pointers and Offsets in same GRF, so that we can use
350 // NODDCLR, NODDCHK flags. -rT
351
352
353 //Keep fORIGIN_X_NLAS, fY_OFFSET_2ND_BLOCK, fSTEP_X_NLAS, pMSGDSC_COPY, ubCONST_ALPHA_COPY as
354 //sub registers of same GRF to enable using NODDCLR NODDCHK. -rT
355
356 //r23
357
358
359 //Lumakey
360
361
362 //r24
363
364
365 //r25
366
367
368 //r26
369
370
371 //defines to generate LABELS during compile time.
372
373
374 // Message Header
375 // m0.7 31:0 Debug
376 // m0.6 31:0 Debug
377 // m0.5 31:0 Ignored
378 // m0.4 31:0 Ignored
379 // m0.3 31:0 Ignored
380 // m0.2 31:16 Ignored
381 // 15 Alpha Write Channel Mask enable=0, disable=1
382 // 14 Blue Write Channel Mask (U)
383 // 13 Green Write Channel Mask (Y)
384 // 12 Red Write Channel Mask (V)
385 // 11:0 Ignored
386 // m0.1 Ignored
387 // m0.0 Ignored
388
389
390 // AVS payload
391 // m1.7 Group ID Number
392 // m1.6 U 2nd Derivative ---> NLAS dx
393 // m1.5 Delta V ---> Step Y
394 // m1.4 Delta U ---> Step X
395 // m1.3 Pixel 0 V Address ---> ORIY (Y0)
396 // m1.2 Pixel 0 U Address ---> ORIX (X0)
397 // m1.1 Vertical Block Number
398 // m1.0 Reserved
399
400 // Sampler Message Descriptor
401 // 31:29 Reserved 000
402 // 28:25 Message length 0010
403 // 24:20 Response length xxxxx ---> 4GRFs for each enabled channel (AVS), 2GRFs for each enabled channel (sample unorm)
404 // 19 Header Present 1
405 // 18:17 SIMD Mode 11 ---> SIMD32/64
406 // 16:12 Message Type xxxxx ---> 01011 sample_8x8, 01100 (sample_unorm), 01010 (sample_unorm+killpix)
407 // 11:8 Sampler Index xxxx
408 // 7:0 Binding Table Index xxxxxxxx
409
410
411 // Msg Header M0.2
412 // 15:15 Alpha Write Channel Mask, 0: written back, 1: not written back
413 // 14:14 Blue Write Channel Mask
414 // 13:13 Green Write Channel Mask
415 // 12:12 Red Write Channel Mask
416
417
418 //By design, Buffer 0,1,2,3 always have Layer 0 and Buffer 4,5 always have L1-L7
419
420
421 //used to generate LABELS at compile time.
422
423
424 // 18:17 SIMD Mode 10 ---> SIMD16
425 // 16:12 Message Type xxxxx ---> 00000 (SIMD16)
426
427
428 //r10-17 - 8 GRFs to load SIMD16 data (upto 4 channels)
429 //r18-19 - 2 GRFs to store sampler ramp.
430
431 .declare mfSCALING_0X_34X_PAYLOAD Base=r14.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=f
432 .declare muwSCALING_0X_34X_PAYLOAD Base=r14.0 ElementSize=2 SrcRegion=<16;16,1> DstRegion=<1> Type=uw
433 .declare mudCALING_0X_34X_PAYLOAD Base=r14.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=ud
434 .declare mubCALING_0X_34X_PAYLOAD Base=r14.0 ElementSize=1 SrcRegion=<32;32,1> DstRegion=<1> Type=ub
435
436
437 .declare fSCALING_0X_34X_TEMP Base=r11.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=f
438 .declare udSCALING_0X_34X_TEMP Base=r11.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=ud
439 .declare ub4SCALING_0X_34X_TEMP Base=r11.0 ElementSize=1 SrcRegion=<32;8,4> DstRegion=<1> Type=ub
440 .declare uwSCALING_0X_34X_TEMP Base=r11.0 ElementSize=2 SrcRegion=<16;16,1> DstRegion=<1> Type=uw
441
442 // Sampler ramp is used for Scaling 0X_0.34X
443 .declare fSAMPLER_RAMP Base=r11.0 ElementSize=4 SrcRegion=<8;8,1> Type=f // 1 GRFs, 8 elements
444
445
446 //#define rMSGDSC_UV r23.0
447
448
449 //End of _SCALING_
450
451
452 //NOTE: We need offsets for second halfof LAYER 0 - even if we do not load it.
453 //Update the channel offset in the buffers for the lower 8x4 data for BUFFER_0.
454 mov (1) r22.4<1>:ud 0x400040:ud
455
456
457 //Check if layer is to be skipped
458
459
460 // f0.1 pre-computed in Set_Layer_0
461 (-f0.1) jmpi (1) SKIP_AVS_LOAD_L0_0_
462
463
464 //AVS_PAYLOAD already has all the data loaded at this point
465 add (1) a0.0:ud r23.5<0;1,0>:ud 0x44EB400:ud //msg desc
466
467 mov (1) r16.2:ud 0x0000D000:ud // Enable Red channel
468
469
470 mov (1) r25.7<1>:ud r9.7:ud { NoDDClr }
471 mov (1) r25.1<1>:ud r9.12:uw { NoDDChk }
472
473
474 // set the vertical block number
475
476
477 mov (8) r17.0:ud r25.0<8;8,1>:ud // Copy msg payload mirrors to MRFs
478
479 // Gen7 AVS WA Only for YUV packed surfaces, NV12 and Y-channel only for Planar surfaces
480 // if (((int)(u_left*width + 5.0/256) > (int)(u_left*width))
481 // {
482 // modified_u_coord = u_coord – 5.0/(256*width); //floating point
483 // }
484 // else if(((int)(u_left*width + 255.0/256) == (int)(u_left*width))
485 // {
486 // modified_u_coord = u_coord + 1.0/(256*width); //floating point
487 // }
488 // else{
489 // modified_u_coord = u_coord;
490 // }
491 // Where u_left = u – 2*du + 3*ddu for IEF On
492 // And u_left = u for IEF Off case
493 //
494
495 // check whether Gen7 AVS WA is enabled,
496 mov (1) r14.8:uw f0.0:uw // save f0.0
497 mov (1) r14.5:f r17.2<0;1,0>:f // save pixel 0 U for chroma
498
499 and.nz.f0.0 (1) null<1>:uw r2.3:uw 0x2:uw
500 (-f0.0)jmpi (1) GEN7_PL2_AVS_WA_DONE_L0_0_
501
502 // Gen7 AVS WA, check if IEF is ON for choosing Gen7 AVS WA formula
503
504 and.nz.f0.0 (8) null<1>:uw r2.3<0;1,0>:uw 0x4:uw
505 (f0.0)mov (8) acc0.0:f r17.2<0;1,0>:f
506 (f0.0)mac (8) acc0.0:f r17.4<0;1,0>:f -2.0:f
507 (f0.0)mac (8) acc0.0:f r17.6<0;1,0>:f 3.0:f
508 (f0.0)mov (1) r14.2:f acc0:f // IEF ON, rTEMP3.2 = u_left
509 (-f0.0)mov (1) r14.2:f r17.2<0;1,0>:f // IEF OFF, rTEMP3.2 = u_left
510
511 and (1) r14.1:ud r2.3:uw 0xFFF8:uw
512 asr (1) r14.1:ud r14.1:ud 3:d
513 mov (1) r14.1:f r14.1:ud
514
515 // Gen7 AVS WA, if (int)(u_left*width + 5.0/256) > (int)(u_left*width)
516 mul (1) r14.0:f r14.2:f r14.1:f // rTEMP3.0 = u_left*width
517 add (1) r14.2:f r14.0:f 0.01953125:f // rTEMP3.2 = u_left*width + 5.0/256
518 add (1) r14.3:f r14.0:f 0.99609375:f // rTEMP3.3 = u_left*width + 255.0/256
519
520 //Check if the values are < 0 and account for (int) cast of negative numbers
521
522 //(int)(u_left*width)
523 cmp.l.f0.0 (1) null<1>:f r14.0:f 0.00000000:f
524 mov (1) r14.0:d r14.0:f
525 (f0.0)add (1) r14.0:d r14.0<0;1,0>:d -1:d
526
527 //(int)(u_left*width + 5.0/256)
528 cmp.l.f0.0 (1) null<1>:f r14.2:f 0.00000000:f
529 mov (1) r14.2:d r14.2:f
530 (f0.0)add (1) r14.2:d r14.2<0;1,0>:d -1:d
531
532 //(int)(u_left*width + 255.0/256)
533 cmp.l.f0.0 (1) null<1>:f r14.3:f 0.00000000:f
534 mov (1) r14.3:d r14.3:f
535 (f0.0)add (1) r14.3:d r14.3<0;1,0>:d -1:d
536
537 mov (1) f0.0:uw 0:uw // clear flag
538 //if (((int)(u_left*width + 5.0/256) > (int)(u_left*width))
539 cmp.g.f1.0 (1) null<1>:d r14.2:d r14.0:d
540 // modified_u_coord = u_coord – 5.0/(256*width); //floating point
541 (f1.0) add (1) r17.2:f r17.2<0;1,0>:f -r2.3:f
542 //else if(((int)(u_left*width + 255.0/256) == (int)(u_left*width))
543 (-f1.0) cmp.e.f0.0 (1) null<1>:d r14.3:d r14.0:d
544 // modified_u_coord = u_coord + 1.0/(256*width); //floating point
545 (f0.0) add (1) r17.2:f r17.2<0;1,0>:f r2.2:f
546
547 GEN7_PL2_AVS_WA_DONE_L0_0_:
548 mov (1) f0.0:uw r14.8:uw // restore f0.0
549
550 send (1) uwBUFFER_0(0)<1> r16 0x2 a0.0:ud
551 // Returns Y data in 4 GRFs in scrambled order
552
553 mov (1) r17.2:f r14.5:f // restore pixel 0 U for chroma, No AVS WA for chroma
554
555 add (1) a0.0:ud r23.5<0;1,0>:ud 0x48EB801:ud // msg desc; 1 is added to change BI to UV
556 mov (1) r16.2:ud 0x0000A000:ud // Enable Red+Blue channel
557
558 send (1) uwBUFFER_0(4)<1> r16 0x2 a0.0:ud
559 // Returns UV data in 8 GRFs in scrambled order
560
561 SKIP_AVS_LOAD_L0_0_:
562 nop
563
564