2
* Copyright 2000-2011 Intel Corporation All Rights Reserved
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:
12
* The above copyright notice and this permission notice (including the
13
* next paragraph) shall be included in all copies or substantial portions
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.
24
* This file was originally licensed under the following license
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
30
* http://www.apache.org/licenses/LICENSE-2.0
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.
38
// 57 // Total instruction count
39
// 1 // Total kernel count
47
// Author: Vivek Kumar
48
// Description: Tasks for DI only case (16x4 block)
53
// Module name: common.inc
55
// Common header file for all Video-Processing kernels
58
.default_execution_size (16)
59
.default_register_type :ub
64
//========== Common constants ==========
67
//========== Macros ==========
70
//Fast Jump, For more details see "Set_Layer_N.asm"
73
//========== Defines ====================
75
//========== Static Parameters (Common To All) ==========
81
// e.g. byte0 byte1 byte2
85
//Color Pipe (IECP) parameters
93
// e.g. byte0 byte1 byte2
98
//========== Inline parameters (Common To All) ===========
101
//============== Binding Index Table===========
102
//Common between DNDI and DNUV
105
//================= Common Message Descriptor =====
106
// Message descriptor for thread spawning
107
// Message Descriptors
108
// = 000 0001 (min message len 1 ) 0,0000 (resp len 0 -add later)
110
// 0001(Spawn a root thread),0001 (Root thread spawn thread)
112
// Thread Spawner Message Descriptor
115
// Message descriptor for atomic operation add
116
// Message Descriptors
117
// = 000 0110 (min message len 6 ) 0,0000 (resp len 0 -add later)
118
// 1(header present)001,10(typed atomic operation)0(return enabled)0(slot group, low 8 bits),0111 (AOP_Add)
119
// 0000,0000 (Binding table index, added later)
122
// Atomic Operation Add Message Descriptor
125
// Message descriptor for dataport media write
126
// Message Descriptors
127
// = 000 0001 (min message len 1 - add later) 00000 (resp len 0)
128
// 1 (header present 1) 0 1010 (media block write) 000000
129
// 00000000 (binding table index - set later)
133
// Message Length defines
136
// Response Length defines
139
// Block Width and Height Size defines
142
// Extended Message Descriptors
145
// Common message descriptors:
148
//===================== Math Function Control ===================================
151
//============ Message Registers ===============
152
// buf4 starts from r28
155
//#define mMSGHDR_EOT r43 // Dummy Message Register for EOT
158
.declare mubMSGPAYLOAD Base=r30 ElementSize=1 SrcRegion=<16;16,1> Type=ub
159
.declare muwMSGPAYLOAD Base=r30 ElementSize=2 SrcRegion=<16;16,1> Type=uw
160
.declare mudMSGPAYLOAD Base=r30 ElementSize=4 SrcRegion=<8;8,1> Type=ud
161
.declare mfMSGPAYLOAD Base=r30 ElementSize=4 SrcRegion=<8;8,1> Type=f
163
//=================== End of thread instruction ===========================
166
//=====================Pointers Used=====================================
169
//=======================================================================
173
// Define temp space for any usages
179
// temp space for rotation
181
.declare fROBUF Base=r11.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=f
183
.declare udROBUF Base=r11.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=ud
185
.declare uwROBUF Base=r11.0 ElementSize=2 SrcRegion=<16;16,1> DstRegion=<1> Type=uw
187
.declare ubROBUF Base=r11.0 ElementSize=1 SrcRegion=<16;16,1> DstRegion=<1> Type=ub
189
.declare ub4ROBUF Base=r11.0 ElementSize=1 SrcRegion=<32;8,4> DstRegion=<4> Type=ub
195
// FileName: DNDI.inc
196
// Author: Vivek Kumar
197
// Description: Include file for DN, DI and DNDI
198
// Inputs: DI_ENABLE, DN_ENABLE, DN_PLANAR, DN_PACKED
211
//====================== Binding table (Explicit To DNDI)=========================================
214
.declare mudMSGHDR_DNDI Base=r18 ElementSize=4 Type=ud
215
.declare mdMSGHDR_DNDI Base=r18 ElementSize=4 Type=d
216
.declare mwMSGHDR_DNDI Base=r18 ElementSize=2 Type=w
219
.declare mudMSGHDR_STMM Base=r20 ElementSize=4 Type=ud
222
.declare mudMSGHDR_HIST Base=r22 ElementSize=4 Type=ud
225
.declare mudMSGHDR_ENC_STATS Base=r24 ElementSize=4 Type=ud
226
.declare muwMSGHDR_ENC_STATS Base=r24 ElementSize=2 Type=uw
227
.declare mubMSGHDR_ENC_STATS Base=r24 ElementSize=1 Type=ub
230
.declare mudMSGHDR_DN_OUT Base=r31.0 ElementSize=4 Type=ud
231
.declare mdMSGHDR_DN_OUT Base=r31.0 ElementSize=4 Type=d
232
.declare mubMSGHDR_DN_OUT Base=r31.0 ElementSize=1 Type=ub
235
.declare mudMSGHDR_UVCOPY Base=r36 ElementSize=4 Type=ud
236
.declare mdMSGHDR_UVCOPY Base=r36 ElementSize=4 Type=d
237
.declare mudMSGHDR_UCOPY Base=r36 ElementSize=4 Type=ud
238
.declare mudMSGHDR_VCOPY Base=r38 ElementSize=4 Type=ud
241
.declare mudMSGHDR_DI_OUT1 Base=r18.0 ElementSize=4 Type=ud
242
.declare mubMSGHDR_DI_OUT1 Base=r18.0 ElementSize=1 Type=ub
245
.declare mudMSGHDR_DI_OUT2 Base=r23.0 ElementSize=4 Type=ud
246
.declare mubMSGHDR_DI_OUT2 Base=r23.0 ElementSize=1 Type=ub
249
//Use r45 as message header, so no need to "mov" the data.
251
.declare mudDN_Y_OUT Base=r45.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=ud
253
// Message response (Denoised & DI-ed pixels & statistics); Use buffer 5
254
.declare udDNDI_RESP Base=r46.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=ud
255
.declare uwDNDI_RESP Base=r46.0 ElementSize=2 SrcRegion=<16;16,1> DstRegion=<1> Type=uw
256
.declare ubDNDI_RESP Base=r46.0 ElementSize=1 SrcRegion=<16;16,1> DstRegion=<1> Type=ub
258
// Message response (UV Copy); Use buffer 5
259
.declare udDNDI_UV_RESP Base=r58.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=ud
260
.declare ubDNDI_UV_RESP Base=r58.0 ElementSize=1 SrcRegion=<16;16,1> DstRegion=<1> Type=ub
262
//Temp GRFs: For 42X to 422 Conversion
263
.declare uwDNDI_UVCOPY_TEMP Base=r10.0 ElementSize=2 SrcRegion=<16;16,1> DstRegion=<1> Type=uw //8 GRFs
264
.declare ubDNDI_UVCOPY_TEMP Base=r10.0 ElementSize=1 SrcRegion=<16;16,1> DstRegion=<1> Type=ub //8 GRFs
265
//---------------------------------------------------------------------------
266
// Message descriptors
267
//---------------------------------------------------------------------------
268
// Extended message descriptor
269
// Message descriptor for sampler read
270
// = 000 0010 (message len 2) 00000 (resp len - set later, 12 or 5 or 11)
271
// 1 (header present 1) 0 11 (SIMD32/64 mode)
272
// 1000 (message type) 0000 (DI state index)
273
// 00000000 (binding table index - set later)
277
// Attention: The Message Length is The Number of GRFs with Data Only, without the Header
280
//---------------------------------------------------------------------------
281
// VDI Return Data format
282
//---------------------------------------------------------------------------
283
// Defines for DI enabled
286
// Defines for DI disabled
290
// FileName: DNDI_Command.asm
291
// Author: Vivek Kumar
292
// Description: Sends a message to the VDI to process one DN (16x8) or DNDI (16x4) block
294
// Prepare the DNDI send command
295
mov (8) mudMSGHDR_DNDI(0)<1> r0.0<8;8,1>:ud // message header
296
mov (1) mwMSGHDR_DNDI(1,4)<1> r9.0<0;1,0>:w { NoDDClr } // horizontal origin // Do we need to add offset here? -vK
297
mov (1) mwMSGHDR_DNDI(1,12)<1> r9.1<0;1,0>:w { NoDDChk } // vertical origin // Can these 2 be combined? - vK
299
send (8) udDNDI_RESP(0)<1> r18 0x2 0x4AE8003:ud
301
// On Gen6, with VDI walker, use the XY pair returned rather than programmed above
302
// VDI_RETURNED_XY is ordered XY in case of walker enables and the same as programmed in case of walker disabled
303
mov (2) r9.0<1>:w uwDNDI_RESP(9,14)<2;2,1> // horizontal/Vertial origin in W.14 and W.15
307
// FileName: DI_STMM_Save.asm
308
// Author: Vivek Kumar
309
// Description: Saves DI STMM Data to statistics surface in case of DI enabled (for 16x4 block)
311
// Write STMM to memory
312
mov (8) mudMSGHDR_STMM(0)<1> r0.0<8;8,1>:ud // message header
313
mov (8) mudMSGHDR_STMM(1)<1> udDNDI_RESP(8,0) // Move STMM to MRF
315
shr (1) mudMSGHDR_STMM(0,0)<1> r9.0<0;1,0>:w 1:w { NoDDClr } // X origin / 2
316
mov (1) mudMSGHDR_STMM(0,1)<1> r9.1<0;1,0>:w { NoDDClr, NoDDChk } // Y origin
317
mov (1) mudMSGHDR_STMM(0,2)<1> 0x30007:ud { NoDDChk } // block width and height (8x4)
319
send (8) null<1>:d r20 0x5 0x40A8021:ud
323
// FileName: DNDI_Enc_Stats_Save.asm
324
// Author: Vivek Kumar
325
// Description: Saves Encoder Statistics data to statistics surface in case of DI enabled (for 16x4 block)
327
// Write encoder statistics to memory
328
//Currently enable this only on Gen6 validation
329
mov (8) mudMSGHDR_ENC_STATS(1)<1> 0x0:ud // Init payload MRF
330
mov (8) mudMSGHDR_ENC_STATS(0)<1> r0.0<8;8,1>:ud // message header
332
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
333
mul (1) acc0.1<1>:ud r9.1<0;1,0>:w 3:w // Y origin * 3
334
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
335
mov (1) mudMSGHDR_ENC_STATS(0,2)<1> 0x20007:ud { NoDDChk } //enable the flag after testing on si { NoDDChk } // block width and height (8x3)
336
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
339
//Data block for Encoder Statistics
340
//----------------------------------------------------
341
//| 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 | Bytes
342
//----------------------------------------------------
343
//| BNE | MCNT | FCNT | TCNT | X | X | X | X |
344
//----------------------------------------------------
345
//| DcTpT | SVCM | DcBpT | DcTpB |
346
//----------------------------------------------------
347
//| SHCM | STAD | DcTcB | DcBpB |
348
//----------------------------------------------------
349
mov (1) mudMSGHDR_ENC_STATS(1,0)<1> udDNDI_RESP(9,1)<0;1,0> { NoDDClr } // Move encoder statistics to MRF
350
mov (2) mudMSGHDR_ENC_STATS(1,3)<2> udDNDI_RESP(9,3)<2;2,1> { NoDDClr, NoDDChk } // Move encoder statistics to MRF
351
mov (2) mudMSGHDR_ENC_STATS(1,2)<2> udDNDI_RESP(9,5)<2;2,1> { NoDDChk } // Move encoder statistics to MRF
354
send (8) null<1>:d r24 0x5 0x40A8021:ud
358
// FileName: DI_Save_PA_16x4.asm
359
// Author: Vivek Kumar
360
// Description: Save two 16x4 blocks of DI output in Packed format
363
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
365
mov (8) r27.0<1>:ud r0.0<8;8,1>:ud
366
shl (1) r27.0<1>:d r9.0<0;1,0>:w 1:w { NoDDClr } // H. block origin need to be doubled
367
mov (1) r27.1<1>:d r9.1<0;1,0>:w { NoDDClr, NoDDChk } // Block origin
368
mov (1) r27.2<1>:ud 0x3001F:ud { NoDDChk } // Block width and height (32x8)
370
//prepare the message headers
371
mov (8) r18.0<1>:ud r27<8;8,1>:ud
372
mov (8) r23.0<1>:ud r27<8;8,1>:ud
375
mov (16) r[a0.4, 0]<2> ubDNDI_RESP(0,0) { NoDDClr }
376
mov (16) r[a0.4, 32]<2> ubDNDI_RESP(0,16) { NoDDClr }
377
mov (16) r[a0.4, 64]<2> ubDNDI_RESP(0,32) { NoDDClr }
378
mov (16) r[a0.4, 96]<2> ubDNDI_RESP(0,48) { NoDDClr }
380
mov (8) r[a0.5, 0]<4> ubDNDI_RESP(2,1)<16;8,2> { NoDDClr, NoDDChk } //U pixels
381
mov (8) r[a0.5, 32]<4> ubDNDI_RESP(2,17)<16;8,2> { NoDDClr, NoDDChk } //U pixels
382
mov (8) r[a0.5, 64]<4> ubDNDI_RESP(2,33)<16;8,2> { NoDDClr, NoDDChk } //U pixels
383
mov (8) r[a0.5, 96]<4> ubDNDI_RESP(2,49)<16;8,2> { NoDDClr, NoDDChk } //U pixels
385
mov (8) r[a0.6, 0]<4> ubDNDI_RESP(2,0)<16;8,2> { NoDDChk } //Vpixels
386
mov (8) r[a0.6, 32]<4> ubDNDI_RESP(2,16)<16;8,2> { NoDDChk } //Vpixels
387
mov (8) r[a0.6, 64]<4> ubDNDI_RESP(2,32)<16;8,2> { NoDDChk } //Vpixels
388
mov (8) r[a0.6, 96]<4> ubDNDI_RESP(2,48)<16;8,2> { NoDDChk } //Vpixels
391
mov (16) r[a0.4, 160]<2> ubDNDI_RESP(4,0) { NoDDClr }
392
mov (16) r[a0.4, 192]<2> ubDNDI_RESP(4,16) { NoDDClr }
393
mov (16) r[a0.4, 224]<2> ubDNDI_RESP(4,32) { NoDDClr }
394
mov (16) r[a0.4, 256]<2> ubDNDI_RESP(4,48) { NoDDClr }
396
mov (8) r[a0.5, 160]<4> ubDNDI_RESP(6,1)<16;8,2> { NoDDClr, NoDDChk } //U pixels
397
mov (8) r[a0.5, 192]<4> ubDNDI_RESP(6,17)<16;8,2> { NoDDClr, NoDDChk } //U pixels
398
mov (8) r[a0.5, 224]<4> ubDNDI_RESP(6,33)<16;8,2> { NoDDClr, NoDDChk } //U pixels
399
mov (8) r[a0.5, 256]<4> ubDNDI_RESP(6,49)<16;8,2> { NoDDClr, NoDDChk } //U pixels
401
mov (8) r[a0.6, 160]<4> ubDNDI_RESP(6,0)<16;8,2> { NoDDChk } //Vpixels
402
mov (8) r[a0.6, 192]<4> ubDNDI_RESP(6,16)<16;8,2> { NoDDChk } //Vpixels
403
mov (8) r[a0.6, 224]<4> ubDNDI_RESP(6,32)<16;8,2> { NoDDChk } //Vpixels
404
mov (8) r[a0.6, 256]<4> ubDNDI_RESP(6,48)<16;8,2> { NoDDChk } //Vpixels
406
//save the previous frame
407
send (8) null<1>:d r18.0 0x5 0xA0A801B:ud
409
//save the current frame
410
send (8) null<1>:d r23.0 0x5 0xA0A801E:ud
414
//End of Thread message
416
mov (8) r127<1>:ud r0.0<8;8,1>:ud
417
send (1) null<1>:d r127 0x27 0x02000010