| 1 |
/* |
| 2 |
* Copyright 2000-2011 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 |
*/ |
| 16 |
/* |
| 17 |
* Copyright 2000-2011 Intel Corporation All Rights Reserved |
| 18 |
* |
| 19 |
* Licensed under the Apache License, Version 2.0 (the "License"); |
| 20 |
* you may not use this file except in compliance with the License. |
| 21 |
* You may obtain a copy of the License at |
| 22 |
* |
| 23 |
* http://www.apache.org/licenses/LICENSE-2.0 |
| 24 |
* |
| 25 |
* Unless required by applicable law or agreed to in writing, software |
| 26 |
* distributed under the License is distributed on an "AS IS" BASIS, |
| 27 |
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. |
| 28 |
* See the License for the specific language governing permissions and |
| 29 |
* limitations under the License. |
| 30 |
* Authors: Zhao Yakui <[email protected]> |
| 31 |
*/ |
| 32 |
|
| 33 |
// 326 // Total instruction count |
| 34 |
// 1 // Total kernel count |
| 35 |
|
| 36 |
|
| 37 |
|
| 38 |
// Module name: common.inc |
| 39 |
// |
| 40 |
// Common header file for all Video-Processing kernels |
| 41 |
// |
| 42 |
|
| 43 |
.default_execution_size (16) |
| 44 |
.default_register_type :ub |
| 45 |
|
| 46 |
.reg_count_total 128 |
| 47 |
.reg_count_payload 7 |
| 48 |
|
| 49 |
//========== Common constants ========== |
| 50 |
|
| 51 |
|
| 52 |
//========== Macros ========== |
| 53 |
|
| 54 |
|
| 55 |
//Fast Jump, For more details see "Set_Layer_N.asm" |
| 56 |
|
| 57 |
|
| 58 |
//========== Defines ==================== |
| 59 |
|
| 60 |
//========== Static Parameters (Common To All) ========== |
| 61 |
//r1 |
| 62 |
|
| 63 |
|
| 64 |
//r2 |
| 65 |
|
| 66 |
// e.g. byte0 byte1 byte2 |
| 67 |
// YUYV 0 1 3 |
| 68 |
// YVYU 0 3 1 |
| 69 |
|
| 70 |
//Color Pipe (IECP) parameters |
| 71 |
|
| 72 |
|
| 73 |
//ByteCopy |
| 74 |
|
| 75 |
|
| 76 |
//r4 |
| 77 |
|
| 78 |
// e.g. byte0 byte1 byte2 |
| 79 |
// YUYV 0 1 3 |
| 80 |
// YVYU 0 3 1 |
| 81 |
|
| 82 |
|
| 83 |
//========== Inline parameters (Common To All) =========== |
| 84 |
|
| 85 |
|
| 86 |
//============== Binding Index Table=========== |
| 87 |
//Common between DNDI and DNUV |
| 88 |
|
| 89 |
|
| 90 |
//================= Common Message Descriptor ===== |
| 91 |
// Message descriptor for thread spawning |
| 92 |
// Message Descriptors |
| 93 |
// = 000 0001 (min message len 1 ) 0,0000 (resp len 0 -add later) |
| 94 |
// 0000,0000,0000 |
| 95 |
// 0001(Spawn a root thread),0001 (Root thread spawn thread) |
| 96 |
// = 0x02000011 |
| 97 |
// Thread Spawner Message Descriptor |
| 98 |
|
| 99 |
|
| 100 |
// Message descriptor for atomic operation add |
| 101 |
// Message Descriptors |
| 102 |
// = 000 0110 (min message len 6 ) 0,0000 (resp len 0 -add later) |
| 103 |
// 1(header present)001,10(typed atomic operation)0(return enabled)0(slot group, low 8 bits),0111 (AOP_Add) |
| 104 |
// 0000,0000 (Binding table index, added later) |
| 105 |
// = 0x02000011 |
| 106 |
|
| 107 |
// Atomic Operation Add Message Descriptor |
| 108 |
|
| 109 |
|
| 110 |
// Message descriptor for dataport media write |
| 111 |
// Message Descriptors |
| 112 |
// = 000 0001 (min message len 1 - add later) 00000 (resp len 0) |
| 113 |
// 1 (header present 1) 0 1010 (media block write) 000000 |
| 114 |
// 00000000 (binding table index - set later) |
| 115 |
// = 0x020A8000 |
| 116 |
|
| 117 |
|
| 118 |
// Message Length defines |
| 119 |
|
| 120 |
|
| 121 |
// Response Length defines |
| 122 |
|
| 123 |
|
| 124 |
// Block Width and Height Size defines |
| 125 |
|
| 126 |
|
| 127 |
// Extended Message Descriptors |
| 128 |
|
| 129 |
|
| 130 |
// Common message descriptors: |
| 131 |
|
| 132 |
|
| 133 |
//===================== Math Function Control =================================== |
| 134 |
|
| 135 |
|
| 136 |
//============ Message Registers =============== |
| 137 |
// buf4 starts from r28 |
| 138 |
|
| 139 |
|
| 140 |
//#define mMSGHDR_EOT r43 // Dummy Message Register for EOT |
| 141 |
|
| 142 |
|
| 143 |
.declare mubMSGPAYLOAD Base=r30 ElementSize=1 SrcRegion=<16;16,1> Type=ub |
| 144 |
.declare muwMSGPAYLOAD Base=r30 ElementSize=2 SrcRegion=<16;16,1> Type=uw |
| 145 |
.declare mudMSGPAYLOAD Base=r30 ElementSize=4 SrcRegion=<8;8,1> Type=ud |
| 146 |
.declare mfMSGPAYLOAD Base=r30 ElementSize=4 SrcRegion=<8;8,1> Type=f |
| 147 |
|
| 148 |
//=================== End of thread instruction =========================== |
| 149 |
|
| 150 |
|
| 151 |
//=====================Pointers Used===================================== |
| 152 |
|
| 153 |
|
| 154 |
//======================================================================= |
| 155 |
|
| 156 |
|
| 157 |
//r11-r17 |
| 158 |
// Define temp space for any usages |
| 159 |
|
| 160 |
|
| 161 |
// Common Buffers |
| 162 |
|
| 163 |
|
| 164 |
// temp space for rotation |
| 165 |
|
| 166 |
.declare fROBUF Base=r11.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=f |
| 167 |
|
| 168 |
.declare udROBUF Base=r11.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=ud |
| 169 |
|
| 170 |
.declare uwROBUF Base=r11.0 ElementSize=2 SrcRegion=<16;16,1> DstRegion=<1> Type=uw |
| 171 |
|
| 172 |
.declare ubROBUF Base=r11.0 ElementSize=1 SrcRegion=<16;16,1> DstRegion=<1> Type=ub |
| 173 |
|
| 174 |
.declare ub4ROBUF Base=r11.0 ElementSize=1 SrcRegion=<32;8,4> DstRegion=<4> Type=ub |
| 175 |
|
| 176 |
|
| 177 |
// End of common.inc |
| 178 |
|
| 179 |
|
| 180 |
// FileName: VP_Setup.asm |
| 181 |
// Author: Vivek Kumar |
| 182 |
// Description: Sets up all parameters for the Video Processing Kernel |
| 183 |
|
| 184 |
|
| 185 |
|
| 186 |
|
| 187 |
// Description: Includes all definitions explicit to Fast Composite. |
| 188 |
|
| 189 |
|
| 190 |
|
| 191 |
|
| 192 |
// End of common.inc |
| 193 |
|
| 194 |
|
| 195 |
//========== GRF partition ========== |
| 196 |
// r0 header : r0 (1 GRF) |
| 197 |
// Static parameters : r1 - r6 (6 GRFS) |
| 198 |
// Inline parameters : r7 - r8 (2 GRFs) |
| 199 |
// MSGSRC : r27 (1 GRF) |
| 200 |
//=================================== |
| 201 |
|
| 202 |
//Interface: |
| 203 |
//========== Static Parameters (Explicit To Fast Composite) ========== |
| 204 |
//r1 |
| 205 |
//CSC Set 0 |
| 206 |
|
| 207 |
|
| 208 |
.declare udCSC_CURBE Base=r1.0 ElementSize=4 Type=ud |
| 209 |
|
| 210 |
//Constant alpha |
| 211 |
|
| 212 |
|
| 213 |
//r2 |
| 214 |
|
| 215 |
|
| 216 |
// Gen7 AVS WA |
| 217 |
|
| 218 |
|
| 219 |
// WiDi Definitions |
| 220 |
|
| 221 |
|
| 222 |
//Colorfill |
| 223 |
|
| 224 |
|
| 225 |
// 0: 0-degree, 1: 90, 2: 180, 3: 270-degree, clockwise. |
| 226 |
|
| 227 |
.declare ubCOLOR_PIXEL_VAL Base=r2.20 ElementSize=1 SrcRegion=<0;1,0> DstRegion=<1> Type=ub |
| 228 |
|
| 229 |
//r3 |
| 230 |
//Normalised Ratio of Horizontal step size with main video for all layers |
| 231 |
|
| 232 |
|
| 233 |
//Normalised Ratio of Horizontal step size with main video for all layers becomes |
| 234 |
//Normalised Horizontal step size for all layers in VP_Setup.asm |
| 235 |
|
| 236 |
|
| 237 |
//r4 |
| 238 |
//Normalised Vertical step size for all layers |
| 239 |
|
| 240 |
|
| 241 |
//r5 |
| 242 |
//Normalised Vertical Frame Origin for all layers |
| 243 |
|
| 244 |
|
| 245 |
//r6 |
| 246 |
//Normalised Horizontal Frame Origin for all layers |
| 247 |
|
| 248 |
|
| 249 |
//========== Inline Parameters (Explicit To Fast Composite) ========== |
| 250 |
|
| 251 |
|
| 252 |
//Main video Step X |
| 253 |
|
| 254 |
|
| 255 |
//====================== Binding table (Explicit To Fast Composite)========================================= |
| 256 |
|
| 257 |
|
| 258 |
//Used by Interlaced Scaling Kernels |
| 259 |
|
| 260 |
|
| 261 |
//========== Sampler State Table Index (Explicit To Fast Composite)========== |
| 262 |
//Sampler Index for AVS/IEF messages |
| 263 |
|
| 264 |
|
| 265 |
//Sampler Index for SIMD16 sampler messages |
| 266 |
|
| 267 |
|
| 268 |
//============================================================================= |
| 269 |
|
| 270 |
.declare fBUFFER_0 Base=r64.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=f |
| 271 |
.declare fBUFFER_1 Base=r80.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=f |
| 272 |
.declare fBUFFER_2 Base=r96.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=f |
| 273 |
.declare fBUFFER_3 Base=r112.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=f |
| 274 |
.declare fBUFFER_4 Base=r28.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=f |
| 275 |
.declare fBUFFER_5 Base=r46.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=f |
| 276 |
|
| 277 |
.declare udBUFFER_0 Base=r64.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=ud |
| 278 |
.declare udBUFFER_1 Base=r80.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=ud |
| 279 |
.declare udBUFFER_2 Base=r96.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=ud |
| 280 |
.declare udBUFFER_3 Base=r112.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=ud |
| 281 |
.declare udBUFFER_4 Base=r28.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=ud |
| 282 |
.declare udBUFFER_5 Base=r46.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=ud |
| 283 |
|
| 284 |
.declare uwBUFFER_0 Base=r64.0 ElementSize=2 SrcRegion=<16;16,1> DstRegion=<1> Type=uw |
| 285 |
.declare uwBUFFER_1 Base=r80.0 ElementSize=2 SrcRegion=<16;16,1> DstRegion=<1> Type=uw |
| 286 |
.declare uwBUFFER_2 Base=r96.0 ElementSize=2 SrcRegion=<16;16,1> DstRegion=<1> Type=uw |
| 287 |
.declare uwBUFFER_3 Base=r112.0 ElementSize=2 SrcRegion=<16;16,1> DstRegion=<1> Type=uw |
| 288 |
.declare uwBUFFER_4 Base=r28.0 ElementSize=2 SrcRegion=<16;16,1> DstRegion=<1> Type=uw |
| 289 |
.declare uwBUFFER_5 Base=r46.0 ElementSize=2 SrcRegion=<16;16,1> DstRegion=<1> Type=uw |
| 290 |
|
| 291 |
.declare ubBUFFER_0 Base=r64.0 ElementSize=1 SrcRegion=<16;16,1> DstRegion=<1> Type=ub |
| 292 |
.declare ubBUFFER_1 Base=r80.0 ElementSize=1 SrcRegion=<16;16,1> DstRegion=<1> Type=ub |
| 293 |
.declare ubBUFFER_2 Base=r96.0 ElementSize=1 SrcRegion=<16;16,1> DstRegion=<1> Type=ub |
| 294 |
.declare ubBUFFER_3 Base=r112.0 ElementSize=1 SrcRegion=<16;16,1> DstRegion=<1> Type=ub |
| 295 |
.declare ubBUFFER_4 Base=r28.0 ElementSize=1 SrcRegion=<16;16,1> DstRegion=<1> Type=ub |
| 296 |
.declare ubBUFFER_5 Base=r46.0 ElementSize=1 SrcRegion=<16;16,1> DstRegion=<1> Type=ub |
| 297 |
|
| 298 |
.declare ub4BUFFER_0 Base=r64.0 ElementSize=1 SrcRegion=<32;8,4> DstRegion=<4> Type=ub |
| 299 |
.declare ub4BUFFER_1 Base=r80.0 ElementSize=1 SrcRegion=<32;8,4> DstRegion=<4> Type=ub |
| 300 |
.declare ub4BUFFER_2 Base=r96.0 ElementSize=1 SrcRegion=<32;8,4> DstRegion=<4> Type=ub |
| 301 |
.declare ub4BUFFER_3 Base=r112.0 ElementSize=1 SrcRegion=<32;8,4> DstRegion=<4> Type=ub |
| 302 |
.declare ub4BUFFER_4 Base=r28.0 ElementSize=1 SrcRegion=<32;8,4> DstRegion=<4> Type=ub |
| 303 |
.declare ub4BUFFER_5 Base=r46.0 ElementSize=1 SrcRegion=<32;8,4> DstRegion=<4> Type=ub |
| 304 |
|
| 305 |
//Pointer to mask reg |
| 306 |
|
| 307 |
|
| 308 |
//r18 |
| 309 |
|
| 310 |
|
| 311 |
//Always keep Cannel Pointers and Offsets in same GRF, so that we can use |
| 312 |
// NODDCLR, NODDCHK flags. -rT |
| 313 |
|
| 314 |
|
| 315 |
.declare udCSC_COEFF_0 Base=r18.0 ElementSize=4 Type=ud // 1 GRF |
| 316 |
|
| 317 |
//r19 |
| 318 |
|
| 319 |
|
| 320 |
.declare udCSC_COEFF_1 Base=r19.0 ElementSize=4 Type=ud // 1 GRF |
| 321 |
|
| 322 |
|
| 323 |
//r20 |
| 324 |
|
| 325 |
.declare uwALPHA_MASK_REG_TEMP Base=r20.0 ElementSize=2 SrcRegion=<16;16,1> Type=uw // 1 GRF |
| 326 |
|
| 327 |
//r21 |
| 328 |
|
| 329 |
.declare uwALPHA_MASK_REG Base=r21.0 ElementSize=2 SrcRegion=<16;16,1> Type=uw // 1 GRF |
| 330 |
|
| 331 |
//r22 |
| 332 |
|
| 333 |
|
| 334 |
//Always keep Cannel Pointers and Offsets in same GRF, so that we can use |
| 335 |
// NODDCLR, NODDCHK flags. -rT |
| 336 |
|
| 337 |
|
| 338 |
//Keep fORIGIN_X_NLAS, fY_OFFSET_2ND_BLOCK, fSTEP_X_NLAS, pMSGDSC_COPY, ubCONST_ALPHA_COPY as |
| 339 |
//sub registers of same GRF to enable using NODDCLR NODDCHK. -rT |
| 340 |
|
| 341 |
//r23 |
| 342 |
|
| 343 |
|
| 344 |
//Lumakey |
| 345 |
|
| 346 |
|
| 347 |
//r24 |
| 348 |
|
| 349 |
|
| 350 |
//r25 |
| 351 |
|
| 352 |
|
| 353 |
//r26 |
| 354 |
|
| 355 |
|
| 356 |
//defines to generate LABELS during compile time. |
| 357 |
|
| 358 |
|
| 359 |
//Setup pointer to the inline parameter |
| 360 |
|
| 361 |
// Copy MSG HDR |
| 362 |
mov (8) r27.0<1>:ud r0.0<8;8,1>:ud // Initialize message payload header with R0 |
| 363 |
|
| 364 |
// Only one layer is enough |
| 365 |
|
| 366 |
//temp; remove it once unread msg warnings are resolved -vK |
| 367 |
mov (8) r25<1>:ud r0.0<8;8,1>:ud |
| 368 |
mov (8) r26<1>:ud r0.0<8;8,1>:ud |
| 369 |
|
| 370 |
// Calculate StepX for all layers and overwrite it on the ratio |
| 371 |
mul (8) r3.0<1>:f r3.0<8;8,1>:f r9.4<0;1,0>:f //StepX_ratio = StepX / VideoStepX |
| 372 |
|
| 373 |
//Normalised Ratio of Horizontal step size with main video for all layers now becomes |
| 374 |
//Normalised Horizontal step size for all layers |
| 375 |
|
| 376 |
// Calculate block origin for all layers and overwrite it on the frame origin |
| 377 |
mov (2) r10.5<1>:f r9.0<2;2,1>:w //Convert origin from word to float |
| 378 |
|
| 379 |
cmp.e.f0.0 (1) null<1>:d r2.26<0;1,0>:ub 1:uw |
| 380 |
|
| 381 |
|
| 382 |
shr (1) r17.0<1>:uw r2.2<0;1,0>:uw 0:uw |
| 383 |
and (1) r17.0<1>:uw r17.0<0;1,0>:uw 3:uw |
| 384 |
cmp.e.f0.1 (1) null<1>:w r17.0<0;1,0>:uw 1:uw |
| 385 |
(f0.1) jmpi (1) ROTATE_90_L0 |
| 386 |
cmp.e.f0.1 (1) null<1>:w r17.0<0;1,0>:uw 2:uw |
| 387 |
(f0.1) jmpi (1) ROTATE_180_L0 |
| 388 |
cmp.e.f0.1 (1) null<1>:w r17.0<0;1,0>:uw 3:uw |
| 389 |
(f0.1) jmpi (1) ROTATE_270_L0 |
| 390 |
|
| 391 |
// rotate 0 degree |
| 392 |
ROTATE_0_L0: |
| 393 |
(-f0.0)mov (1) acc0.0<1>:f r6.0<0;1,0>:f |
| 394 |
(-f0.0)mac (1) r6.0<1>:f r3.0<0;1,0>:f r10.5<0;1,0>:f |
| 395 |
|
| 396 |
mov (1) acc0.0<1>:f r5.0<0;1,0>:f |
| 397 |
mac (1) r5.0<1>:f r4.0<0;1,0>:f r10.6<0;1,0>:f |
| 398 |
jmpi (1) END_SRC_BLOCK_ORIG_COMP_L0 |
| 399 |
|
| 400 |
// rotate 90 degree |
| 401 |
ROTATE_90_L0: |
| 402 |
(-f0.0)mov (1) acc0.0<1>:f r6.0<0;1,0>:f |
| 403 |
(-f0.0)mac (1) r6.0<1>:f r3.0<0;1,0>:f r10.6<0;1,0>:f |
| 404 |
|
| 405 |
mov (1) r16.0<1>:f r2.0<0;1,0>:uw |
| 406 |
add (1) r17.0<1>:f -r10.5<0;1,0>:f r16.0<0;1,0>:f |
| 407 |
add (1) r17.0<1>:f r17.0<0;1,0>:f -16.0:f |
| 408 |
|
| 409 |
mov (1) acc0.0<1>:f r5.0<0;1,0>:f |
| 410 |
mac (1) r5.0<1>:f r4.0<0;1,0>:f r17.0<0;1,0>:f |
| 411 |
jmpi (1) END_SRC_BLOCK_ORIG_COMP_L0 |
| 412 |
|
| 413 |
// rotate 180 degree |
| 414 |
ROTATE_180_L0: |
| 415 |
(-f0.0)mov (1) r16.0<1>:f r2.0<0;1,0>:uw |
| 416 |
(-f0.0)add (1) r17.0<1>:f -r10.5<0;1,0>:f r16.0<0;1,0>:f |
| 417 |
(-f0.0)add (1) r17.0<1>:f r17.0<0;1,0>:f -16.0:f |
| 418 |
(-f0.0)mov (1) acc0.0<1>:f r6.0<0;1,0>:f |
| 419 |
(-f0.0)mac (1) r6.0<1>:f r3.0<0;1,0>:f r17.0<0;1,0>:f |
| 420 |
|
| 421 |
mov (1) r16.0<1>:f r2.1<0;1,0>:uw |
| 422 |
add (1) r17.0<1>:f -r10.6<0;1,0>:f r16.0<0;1,0>:f |
| 423 |
add (1) r17.0<1>:f r17.0<0;1,0>:f -16.0:f |
| 424 |
mov (1) acc0.0<1>:f r5.0<0;1,0>:f |
| 425 |
mac (1) r5.0<1>:f r4.0<0;1,0>:f r17.0<0;1,0>:f |
| 426 |
jmpi (1) END_SRC_BLOCK_ORIG_COMP_L0 |
| 427 |
|
| 428 |
// rotate 270 degree |
| 429 |
ROTATE_270_L0: |
| 430 |
(-f0.0)mov (1) r16.0<1>:f r2.1<0;1,0>:uw |
| 431 |
(-f0.0)add (1) r17.0<1>:f -r10.6<0;1,0>:f r16.0<0;1,0>:f |
| 432 |
(-f0.0)add (1) r17.0<1>:f r17.0<0;1,0>:f -16.0:f |
| 433 |
(-f0.0)mov (1) acc0.0<1>:f r6.0<0;1,0>:f |
| 434 |
(-f0.0)mac (1) r6.0<1>:f r3.0<0;1,0>:f r17.0<0;1,0>:f |
| 435 |
|
| 436 |
mov (1) acc0.0<1>:f r5.0<0;1,0>:f |
| 437 |
mac (1) r5.0<1>:f r4.0<0;1,0>:f r10.5<0;1,0>:f |
| 438 |
|
| 439 |
END_SRC_BLOCK_ORIG_COMP_L0: |
| 440 |
nop |