Details | Last modification | View Log | RSS feed
Rev | Author | Line No. | Line |
---|---|---|---|
5361 | serge | 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 | // 114 // Total instruction count |
||
17 | // 1 // Total kernel count |
||
18 | |||
19 | .kernel PA_DN_422CP |
||
20 | .code |
||
21 | |||
22 | |||
23 | |||
24 | // FileName: DN_PA_Core.asm |
||
25 | // Author: Vivek Kumar |
||
26 | // Description: Tasks for DN only case (16x8 block) for Packed format |
||
27 | |||
28 | |||
29 | |||
30 | // FileName: DN.asm |
||
31 | // Author: Vivek Kumar |
||
32 | // Description: Tasks for DN only case (16x8 block) |
||
33 | |||
34 | |||
35 | |||
36 | |||
37 | // Module name: common.inc |
||
38 | // |
||
39 | // Common header file for all Video-Processing kernels |
||
40 | // |
||
41 | |||
42 | .default_execution_size (16) |
||
43 | .default_register_type :ub |
||
44 | |||
45 | .reg_count_total 128 |
||
46 | .reg_count_payload 7 |
||
47 | |||
48 | //========== Common constants ========== |
||
49 | |||
50 | |||
51 | //========== Macros ========== |
||
52 | |||
53 | |||
54 | //Fast Jump, For more details see "Set_Layer_N.asm" |
||
55 | |||
56 | |||
57 | //========== Defines ==================== |
||
58 | |||
59 | //========== Static Parameters (Common To All) ========== |
||
60 | //r1 |
||
61 | |||
62 | |||
63 | //r2 |
||
64 | |||
65 | // e.g. byte0 byte1 byte2 |
||
66 | // YUYV 0 1 3 |
||
67 | // YVYU 0 3 1 |
||
68 | |||
69 | //Color Pipe (IECP) parameters |
||
70 | |||
71 | |||
72 | //ByteCopy |
||
73 | |||
74 | |||
75 | //r4 |
||
76 | |||
77 | // e.g. byte0 byte1 byte2 |
||
78 | // YUYV 0 1 3 |
||
79 | // YVYU 0 3 1 |
||
80 | |||
81 | |||
82 | //========== Inline parameters (Common To All) =========== |
||
83 | |||
84 | |||
85 | //============== Binding Index Table=========== |
||
86 | //Common between DNDI and DNUV |
||
87 | |||
88 | |||
89 | //================= Common Message Descriptor ===== |
||
90 | // Message descriptor for thread spawning |
||
91 | // Message Descriptors |
||
92 | // = 000 0001 (min message len 1 ) 0,0000 (resp len 0 -add later) |
||
93 | // 0000,0000,0000 |
||
94 | // 0001(Spawn a root thread),0001 (Root thread spawn thread) |
||
95 | // = 0x02000011 |
||
96 | // Thread Spawner Message Descriptor |
||
97 | |||
98 | |||
99 | // Message descriptor for atomic operation add |
||
100 | // Message Descriptors |
||
101 | // = 000 0110 (min message len 6 ) 0,0000 (resp len 0 -add later) |
||
102 | // 1(header present)001,10(typed atomic operation)0(return enabled)0(slot group, low 8 bits),0111 (AOP_Add) |
||
103 | // 0000,0000 (Binding table index, added later) |
||
104 | // = 0x02000011 |
||
105 | |||
106 | // Atomic Operation Add Message Descriptor |
||
107 | |||
108 | |||
109 | // Message descriptor for dataport media write |
||
110 | // Message Descriptors |
||
111 | // = 000 0001 (min message len 1 - add later) 00000 (resp len 0) |
||
112 | // 1 (header present 1) 0 1010 (media block write) 000000 |
||
113 | // 00000000 (binding table index - set later) |
||
114 | // = 0x020A8000 |
||
115 | |||
116 | |||
117 | // Message Length defines |
||
118 | |||
119 | |||
120 | // Response Length defines |
||
121 | |||
122 | |||
123 | // Block Width and Height Size defines |
||
124 | |||
125 | |||
126 | // Extended Message Descriptors |
||
127 | |||
128 | |||
129 | // Common message descriptors: |
||
130 | |||
131 | |||
132 | //===================== Math Function Control =================================== |
||
133 | |||
134 | |||
135 | //============ Message Registers =============== |
||
136 | // buf4 starts from r28 |
||
137 | |||
138 | |||
139 | //#define mMSGHDR_EOT r43 // Dummy Message Register for EOT |
||
140 | |||
141 | |||
142 | .declare mubMSGPAYLOAD Base=r30 ElementSize=1 SrcRegion=<16;16,1> Type=ub |
||
143 | .declare muwMSGPAYLOAD Base=r30 ElementSize=2 SrcRegion=<16;16,1> Type=uw |
||
144 | .declare mudMSGPAYLOAD Base=r30 ElementSize=4 SrcRegion=<8;8,1> Type=ud |
||
145 | .declare mfMSGPAYLOAD Base=r30 ElementSize=4 SrcRegion=<8;8,1> Type=f |
||
146 | |||
147 | //=================== End of thread instruction =========================== |
||
148 | |||
149 | |||
150 | //=====================Pointers Used===================================== |
||
151 | |||
152 | |||
153 | //======================================================================= |
||
154 | |||
155 | |||
156 | //r9-r17 |
||
157 | // Define temp space for any usages |
||
158 | |||
159 | |||
160 | // Common Buffers |
||
161 | |||
162 | |||
163 | // temp space for rotation |
||
164 | |||
165 | .declare fROBUF Base=r9.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=f |
||
166 | |||
167 | .declare udROBUF Base=r9.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=ud |
||
168 | |||
169 | .declare uwROBUF Base=r9.0 ElementSize=2 SrcRegion=<16;16,1> DstRegion=<1> Type=uw |
||
170 | |||
171 | .declare ubROBUF Base=r9.0 ElementSize=1 SrcRegion=<16;16,1> DstRegion=<1> Type=ub |
||
172 | |||
173 | .declare ub4ROBUF Base=r9.0 ElementSize=1 SrcRegion=<32;8,4> DstRegion=<4> Type=ub |
||
174 | |||
175 | |||
176 | // End of common.inc |
||
177 | |||
178 | |||
179 | // FileName: DNDI.inc |
||
180 | // Author: Vivek Kumar |
||
181 | // Description: Include file for DN, DI and DNDI |
||
182 | // Inputs: DI_ENABLE, DN_ENABLE, DN_PLANAR, DN_PACKED |
||
183 | |||
184 | |||
185 | |||
186 | |||
187 | // End of common.inc |
||
188 | |||
189 | |||
190 | //Interface: |
||
191 | //Static Parameters: |
||
192 | //r1 |
||
193 | |||
194 | |||
195 | //====================== Binding table (Explicit To DNDI)========================================= |
||
196 | |||
197 | |||
198 | .declare mudMSGHDR_DNDI Base=r18 ElementSize=4 Type=ud |
||
199 | .declare mdMSGHDR_DNDI Base=r18 ElementSize=4 Type=d |
||
200 | .declare mwMSGHDR_DNDI Base=r18 ElementSize=2 Type=w |
||
201 | |||
202 | |||
203 | .declare mudMSGHDR_STMM Base=r20 ElementSize=4 Type=ud |
||
204 | |||
205 | |||
206 | .declare mudMSGHDR_HIST Base=r22 ElementSize=4 Type=ud |
||
207 | |||
208 | |||
209 | .declare mudMSGHDR_ENC_STATS Base=r24 ElementSize=4 Type=ud |
||
210 | .declare muwMSGHDR_ENC_STATS Base=r24 ElementSize=2 Type=uw |
||
211 | .declare mubMSGHDR_ENC_STATS Base=r24 ElementSize=1 Type=ub |
||
212 | |||
213 | |||
214 | .declare mudMSGHDR_DN_OUT Base=r31.0 ElementSize=4 Type=ud |
||
215 | .declare mdMSGHDR_DN_OUT Base=r31.0 ElementSize=4 Type=d |
||
216 | .declare mubMSGHDR_DN_OUT Base=r31.0 ElementSize=1 Type=ub |
||
217 | |||
218 | |||
219 | .declare mudMSGHDR_UVCOPY Base=r36 ElementSize=4 Type=ud |
||
220 | .declare mdMSGHDR_UVCOPY Base=r36 ElementSize=4 Type=d |
||
221 | .declare mudMSGHDR_UCOPY Base=r36 ElementSize=4 Type=ud |
||
222 | .declare mudMSGHDR_VCOPY Base=r38 ElementSize=4 Type=ud |
||
223 | |||
224 | |||
225 | .declare mudMSGHDR_DI_OUT1 Base=r18.0 ElementSize=4 Type=ud |
||
226 | .declare mubMSGHDR_DI_OUT1 Base=r18.0 ElementSize=1 Type=ub |
||
227 | |||
228 | |||
229 | .declare mudMSGHDR_DI_OUT2 Base=r23.0 ElementSize=4 Type=ud |
||
230 | .declare mubMSGHDR_DI_OUT2 Base=r23.0 ElementSize=1 Type=ub |
||
231 | |||
232 | //r45 |
||
233 | //Use r45 as message header, so no need to "mov" the data. |
||
234 | |||
235 | .declare mudDN_Y_OUT Base=r45.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=ud |
||
236 | |||
237 | // Message response (Denoised & DI-ed pixels & statistics); Use buffer 5 |
||
238 | .declare udDNDI_RESP Base=r46.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=ud |
||
239 | .declare uwDNDI_RESP Base=r46.0 ElementSize=2 SrcRegion=<16;16,1> DstRegion=<1> Type=uw |
||
240 | .declare ubDNDI_RESP Base=r46.0 ElementSize=1 SrcRegion=<16;16,1> DstRegion=<1> Type=ub |
||
241 | |||
242 | // Message response (UV Copy); Use buffer 5 |
||
243 | .declare udDNDI_UV_RESP Base=r58.0 ElementSize=4 SrcRegion=<8;8,1> DstRegion=<1> Type=ud |
||
244 | .declare ubDNDI_UV_RESP Base=r58.0 ElementSize=1 SrcRegion=<16;16,1> DstRegion=<1> Type=ub |
||
245 | |||
246 | //Temp GRFs: For 42X to 422 Conversion |
||
247 | .declare uwDNDI_UVCOPY_TEMP Base=r10.0 ElementSize=2 SrcRegion=<16;16,1> DstRegion=<1> Type=uw //8 GRFs |
||
248 | .declare ubDNDI_UVCOPY_TEMP Base=r10.0 ElementSize=1 SrcRegion=<16;16,1> DstRegion=<1> Type=ub //8 GRFs |
||
249 | //--------------------------------------------------------------------------- |
||
250 | // Message descriptors |
||
251 | //--------------------------------------------------------------------------- |
||
252 | // Extended message descriptor |
||
253 | // Message descriptor for sampler read |
||
254 | // = 000 0010 (message len 2) 00000 (resp len - set later, 12 or 5 or 11) |
||
255 | // 1 (header present 1) 0 11 (SIMD32/64 mode) |
||
256 | // 1000 (message type) 0000 (DI state index) |
||
257 | // 00000000 (binding table index - set later) |
||
258 | // = 0x040b8000 |
||
259 | |||
260 | |||
261 | // Attention: The Message Length is The Number of GRFs with Data Only, without the Header |
||
262 | |||
263 | |||
264 | //--------------------------------------------------------------------------- |
||
265 | // VDI Return Data format |
||
266 | //--------------------------------------------------------------------------- |
||
267 | // Defines for DI enabled |
||
268 | |||
269 | |||
270 | // Defines for DI disabled |
||
271 | |||
272 | |||
273 | |||
274 | // FileName: DNDI_Command.asm |
||
275 | // Author: Vivek Kumar |
||
276 | // Description: Sends a message to the VDI to process one DN (16x8) or DNDI (16x4) block |
||
277 | |||
278 | // Prepare the DNDI send command |
||
279 | mov (8) mudMSGHDR_DNDI(0)<1> r0.0<8;8,1>:ud // message header |
||
280 | mov (1) mwMSGHDR_DNDI(1,4)<1> r7.0<0;1,0>:w { NoDDClr } // horizontal origin // Do we need to add offset here? -vK |
||
281 | mov (1) mwMSGHDR_DNDI(1,12)<1> r7.1<0;1,0>:w { NoDDChk } // vertical origin // Can these 2 be combined? - vK |
||
282 | |||
283 | send (8) udDNDI_RESP(0)<1> r18 0x2 0x49E8003:ud |
||
284 | |||
285 | // On Gen6, with VDI walker, use the XY pair returned rather than programmed above |
||
286 | // VDI_RETURNED_XY is ordered XY in case of walker enables and the same as programmed in case of walker disabled |
||
287 | mov (2) r7.0<1>:w uwDNDI_RESP(4,14)<2;2,1> // horizontal/Vertial origin in W.14 and W.15 |
||
288 | |||
289 | |||
290 | |||
291 | // FileName: DN_Hist_Save.asm |
||
292 | // Author: Vivek Kumar |
||
293 | // Description: Saves DN history data to statistics surface |
||
294 | |||
295 | // Write denoise history to memory |
||
296 | mov (8) r27<1>:ud r0.0<8;8,1>:ud // message header |
||
297 | |||
298 | |||
299 | mov (2) mudMSGHDR_HIST(1)<1> udDNDI_RESP(4,0)<2;2,1> // Move denoise history to MRF (4x2) |
||
300 | |||
301 | |||
302 | shr (2) r27.0<1>:ud r7.0<2;2,1>:w 2:w // X,Y origin / 4 |
||
303 | add (1) r27.0<1>:ud r27.0<0;1,0>:ud r1.12<0;1,0>:uw { NoDDClr } // Add pitch to X origin |
||
304 | mov (1) r27.2<1>:ud 0x10003:ud { NoDDChk } // block width and height |
||
305 | |||
306 | mov (8) mudMSGHDR_HIST(0)<1> r27.0<8;8,1>:ud |
||
307 | send (8) null<1>:d r22 0x5 0x40A8021:ud |
||
308 | |||
309 | |||
310 | |||
311 | // FileName: DNDI_Enc_Stats_Save.asm |
||
312 | // Author: Vivek Kumar |
||
313 | // Description: Saves Encoder Statistics data to statistics surface in case of DI enabled (for 16x4 block) |
||
314 | |||
315 | // Write encoder statistics to memory |
||
316 | //Currently enable this only on Gen6 validation |
||
317 | mov (8) mudMSGHDR_ENC_STATS(1)<1> 0x0:ud // Init payload MRF |
||
318 | mov (8) mudMSGHDR_ENC_STATS(0)<1> r0.0<8;8,1>:ud // message header |
||
319 | |||
320 | shr (1) mudMSGHDR_ENC_STATS(0,0)<1> r7.0<0;1,0>:w 1:w { NoDDClr } //enable the flag after testing on si { NoDDClr } // X origin / 2 |
||
321 | mul (1) acc0.1<1>:ud r7.1<0;1,0>:w 3:w // Y origin * 3 |
||
322 | 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 |
||
323 | mov (1) mudMSGHDR_ENC_STATS(0,2)<1> 0x50003:ud { NoDDChk } //enable the flag after testing on si { NoDDChk } // block width and height (8x3) |
||
324 | 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 |
||
325 | |||
326 | |||
327 | //Data block for Encoder Statistics |
||
328 | //---------------------------------------------------- |
||
329 | //| 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 | Bytes |
||
330 | //---------------------------------------------------- |
||
331 | //| BNE | X | X | X | X | |
||
332 | //---------------------------------------------------- |
||
333 | //| X | SVCM | X | |
||
334 | //---------------------------------------------------- |
||
335 | //| SHCM | STAD | X | |
||
336 | //---------------------------------------------------- |
||
337 | //| X | X | |
||
338 | //---------------------------------------------------- |
||
339 | //| X | SVCM | X | |
||
340 | //---------------------------------------------------- |
||
341 | //| SHCM | STAD | X | |
||
342 | //---------------------------------------------------- |
||
343 | mov (1) mubMSGHDR_ENC_STATS(1,0)<1> ubDNDI_RESP(4,8)<0;1,0> { NoDDClr } // Move encoder statistics to MRF |
||
344 | mov (1) muwMSGHDR_ENC_STATS(1,3)<1> uwDNDI_RESP(4,11)<0;1,0> { NoDDClr, NoDDChk } // Move encoder statistics to MRF |
||
345 | mov (2) muwMSGHDR_ENC_STATS(1,4)<1> uwDNDI_RESP(4,12)<2;2,1> { NoDDClr, NoDDChk } // Move encoder statistics to MRF |
||
346 | mov (1) muwMSGHDR_ENC_STATS(1,9)<1> uwDNDI_RESP(4,8)<0;1,0> { NoDDClr, NoDDChk } // Move encoder statistics to MRF |
||
347 | mov (2) muwMSGHDR_ENC_STATS(1,10)<1> uwDNDI_RESP(4,9)<2;2,1> { NoDDChk } // Move encoder statistics to MRF |
||
348 | |||
349 | |||
350 | send (8) null<1>:d r24 0x5 0x40A8021:ud |
||
351 | |||
352 | |||
353 | |||
354 | // FileName: DN_Save_PA.asm |
||
355 | // Author: Vivek Kumar |
||
356 | // Description: Save one 16x8 blocks of DN output in Packed format |
||
357 | |||
358 | |||
359 | add (4) a0.4<1>:uw r2.28<4;4,1>:ub 1024:w // Initial Y,U,V offset in YUV422 block; it starts at m14 |
||
360 | |||
361 | mov (8) mudMSGHDR_DN_OUT(0)<1> r0<8;8,1>:ud // message header |
||
362 | shl (1) mdMSGHDR_DN_OUT(0,0)<1> r7.0<0;1,0>:w 1:w { NoDDClr } // X origin * 2 (422 output) |
||
363 | mov (1) mdMSGHDR_DN_OUT(0,1)<1> r7.1<0;1,0>:w { NoDDClr, NoDDChk } // Y origin |
||
364 | mov (1) mudMSGHDR_DN_OUT(0,2)<1> 0x7001F:ud { NoDDChk } // block width and height (32x8) |
||
365 | |||
366 | mov (16) r[a0.4,0]<2>:ub ubDNDI_RESP(0,0)<16;16,1> { NoDDClr } // copy line of Y directly to memory as optimization |
||
367 | mov (16) r[a0.4,32]<2>:ub ubDNDI_RESP(0,16)<16;16,1> { NoDDClr } // copy line of Y directly to memory as optimization |
||
368 | mov (16) r[a0.4,64]<2>:ub ubDNDI_RESP(0,32)<16;16,1> { NoDDClr } // copy line of Y directly to memory as optimization |
||
369 | mov (16) r[a0.4,96]<2>:ub ubDNDI_RESP(0,48)<16;16,1> { NoDDClr } // copy line of Y directly to memory as optimization |
||
370 | mov (16) r[a0.4,128]<2>:ub ubDNDI_RESP(0,64)<16;16,1> { NoDDClr } // copy line of Y directly to memory as optimization |
||
371 | mov (16) r[a0.4,160]<2>:ub ubDNDI_RESP(0,80)<16;16,1> { NoDDClr } // copy line of Y directly to memory as optimization |
||
372 | mov (16) r[a0.4,192]<2>:ub ubDNDI_RESP(0,96)<16;16,1> { NoDDClr } // copy line of Y directly to memory as optimization |
||
373 | mov (16) r[a0.4,224]<2>:ub ubDNDI_RESP(0,112)<16;16,1> { NoDDClr } // copy line of Y directly to memory as optimization |
||
374 | mov (8) r[a0.5,0]<4>:ub ubDNDI_RESP(5,1)<16;8,2> { NoDDClr, NoDDChk } // copy line of U directly to memory as optimization |
||
375 | mov (8) r[a0.6,0]<4>:ub ubDNDI_RESP(5,0)<16;8,2> { NoDDChk } // copy line of V directly to memory as optimization |
||
376 | mov (8) r[a0.5,32]<4>:ub ubDNDI_RESP(5,17)<16;8,2> { NoDDClr, NoDDChk } // copy line of U directly to memory as optimization |
||
377 | mov (8) r[a0.6,32]<4>:ub ubDNDI_RESP(5,16)<16;8,2> { NoDDChk } // copy line of V directly to memory as optimization |
||
378 | mov (8) r[a0.5,64]<4>:ub ubDNDI_RESP(5,33)<16;8,2> { NoDDClr, NoDDChk } // copy line of U directly to memory as optimization |
||
379 | mov (8) r[a0.6,64]<4>:ub ubDNDI_RESP(5,32)<16;8,2> { NoDDChk } // copy line of V directly to memory as optimization |
||
380 | mov (8) r[a0.5,96]<4>:ub ubDNDI_RESP(5,49)<16;8,2> { NoDDClr, NoDDChk } // copy line of U directly to memory as optimization |
||
381 | mov (8) r[a0.6,96]<4>:ub ubDNDI_RESP(5,48)<16;8,2> { NoDDChk } // copy line of V directly to memory as optimization |
||
382 | mov (8) r[a0.5,128]<4>:ub ubDNDI_RESP(5,65)<16;8,2> { NoDDClr, NoDDChk } // copy line of U directly to memory as optimization |
||
383 | mov (8) r[a0.6,128]<4>:ub ubDNDI_RESP(5,64)<16;8,2> { NoDDChk } // copy line of V directly to memory as optimization |
||
384 | mov (8) r[a0.5,160]<4>:ub ubDNDI_RESP(5,81)<16;8,2> { NoDDClr, NoDDChk } // copy line of U directly to memory as optimization |
||
385 | mov (8) r[a0.6,160]<4>:ub ubDNDI_RESP(5,80)<16;8,2> { NoDDChk } // copy line of V directly to memory as optimization |
||
386 | mov (8) r[a0.5,192]<4>:ub ubDNDI_RESP(5,97)<16;8,2> { NoDDClr, NoDDChk } // copy line of U directly to memory as optimization |
||
387 | mov (8) r[a0.6,192]<4>:ub ubDNDI_RESP(5,96)<16;8,2> { NoDDChk } // copy line of V directly to memory as optimization |
||
388 | mov (8) r[a0.5,224]<4>:ub ubDNDI_RESP(5,113)<16;8,2> { NoDDClr, NoDDChk } // copy line of U directly to memory as optimization |
||
389 | mov (8) r[a0.6,224]<4>:ub ubDNDI_RESP(5,112)<16;8,2> { NoDDChk } // copy line of V directly to memory as optimization |
||
390 | |||
391 | //send out data through data port |
||
392 | send (8) null<1>:d r31.0 0x5 0x120A8018:ud |
||
393 | |||
394 | |||
395 | |||
396 | // FileName: DN_Save_422CP_16x8.asm |
||
397 | // Author: Vivek Kumar |
||
398 | // Description: Save one 16x8 blocks of DN output to the color pipe in 4-2-2 format |
||
399 | |||
400 | |||
401 | .declare mubMSGHDR_DN_OUT_2 Base=r36.0 ElementSize=1 Type=ub |
||
402 | |||
403 | |||
404 | mov (8) mudMSGHDR_DN_OUT(0)<1> r0<8;8,1>:ud // message header |
||
405 | shl (1) mdMSGHDR_DN_OUT(0,0)<1> r7.0<0;1,0>:w 1:w { NoDDClr } // X origin * 2 (422 output) |
||
406 | mov (1) mdMSGHDR_DN_OUT(0,1)<1> r7.1<0;1,0>:w { NoDDClr, NoDDChk } // Y origin |
||
407 | mov (1) mudMSGHDR_DN_OUT(0,2)<1> 0x7000F:ud { NoDDClr, NoDDChk } // block width and height (16x8) |
||
408 | |||
409 | //M0.3 - 0 - CP Enable, 1 - Area of Interest, 3:2 Message Format(TBD), 4:3 - Ignored, 31:5 CP state pointer |
||
410 | //Compose area-of-interest bit + color pipe state pointer |
||
411 | or (1) mudMSGHDR_DN_OUT(0,3)<1> r2.4<0;1,0>:ud r7.26<0;1,0>:b { NoDDChk } |
||
412 | |||
413 | // First 8 x 8 Block |
||
414 | mov (8) mubMSGHDR_DN_OUT(1)<2> ubDNDI_RESP(0,0)<8;8,1> { NoDDClr } // copy line of Y directly to memory as optimization |
||
415 | mov (8) mubMSGHDR_DN_OUT(1,16)<2> ubDNDI_RESP(0,16)<8;8,1> { NoDDClr, NoDDChk } // copy line of Y directly to memory as optimization |
||
416 | mov (8) mubMSGHDR_DN_OUT(2)<2> ubDNDI_RESP(0,32)<8;8,1> { NoDDClr } // copy line of Y directly to memory as optimization |
||
417 | mov (8) mubMSGHDR_DN_OUT(2,16)<2> ubDNDI_RESP(0,48)<8;8,1> { NoDDClr, NoDDChk } // copy line of Y directly to memory as optimization |
||
418 | mov (8) mubMSGHDR_DN_OUT(3)<2> ubDNDI_RESP(0,64)<8;8,1> { NoDDClr } // copy line of Y directly to memory as optimization |
||
419 | mov (8) mubMSGHDR_DN_OUT(3,16)<2> ubDNDI_RESP(0,80)<8;8,1> { NoDDClr, NoDDChk } // copy line of Y directly to memory as optimization |
||
420 | mov (8) mubMSGHDR_DN_OUT(4)<2> ubDNDI_RESP(0,96)<8;8,1> { NoDDClr } // copy line of Y directly to memory as optimization |
||
421 | mov (8) mubMSGHDR_DN_OUT(4,16)<2> ubDNDI_RESP(0,112)<8;8,1> { NoDDClr, NoDDChk } // copy line of Y directly to memory as optimization |
||
422 | |||
423 | mov (4) mubMSGHDR_DN_OUT(1,1)<4> ubDNDI_RESP(5,1)<8;4,2> { NoDDClr, NoDDChk } // copy line of U directly to memory as optimization |
||
424 | mov (4) mubMSGHDR_DN_OUT(1,17)<4> ubDNDI_RESP(5,17)<8;4,2> { NoDDClr, NoDDChk } // copy line of U directly to memory as optimization |
||
425 | |||
426 | mov (4) mubMSGHDR_DN_OUT(1,3)<4> ubDNDI_RESP(5,0)<8;4,2> { NoDDChk } // copy line of V directly to memory as optimization |
||
427 | mov (4) mubMSGHDR_DN_OUT(1,19)<4> ubDNDI_RESP(5,16)<8;4,2> { NoDDChk } // copy line of V directly to memory as optimization |
||
428 | mov (4) mubMSGHDR_DN_OUT(2,1)<4> ubDNDI_RESP(5,33)<8;4,2> { NoDDClr, NoDDChk } // copy line of U directly to memory as optimization |
||
429 | mov (4) mubMSGHDR_DN_OUT(2,17)<4> ubDNDI_RESP(5,49)<8;4,2> { NoDDClr, NoDDChk } // copy line of U directly to memory as optimization |
||
430 | |||
431 | mov (4) mubMSGHDR_DN_OUT(2,3)<4> ubDNDI_RESP(5,32)<8;4,2> { NoDDChk } // copy line of V directly to memory as optimization |
||
432 | mov (4) mubMSGHDR_DN_OUT(2,19)<4> ubDNDI_RESP(5,48)<8;4,2> { NoDDChk } // copy line of V directly to memory as optimization |
||
433 | mov (4) mubMSGHDR_DN_OUT(3,1)<4> ubDNDI_RESP(5,65)<8;4,2> { NoDDClr, NoDDChk } // copy line of U directly to memory as optimization |
||
434 | mov (4) mubMSGHDR_DN_OUT(3,17)<4> ubDNDI_RESP(5,81)<8;4,2> { NoDDClr, NoDDChk } // copy line of U directly to memory as optimization |
||
435 | |||
436 | mov (4) mubMSGHDR_DN_OUT(3,3)<4> ubDNDI_RESP(5,64)<8;4,2> { NoDDChk } // copy line of V directly to memory as optimization |
||
437 | mov (4) mubMSGHDR_DN_OUT(3,19)<4> ubDNDI_RESP(5,80)<8;4,2> { NoDDChk } // copy line of V directly to memory as optimization |
||
438 | mov (4) mubMSGHDR_DN_OUT(4,1)<4> ubDNDI_RESP(5,97)<8;4,2> { NoDDClr, NoDDChk } // copy line of U directly to memory as optimization |
||
439 | mov (4) mubMSGHDR_DN_OUT(4,17)<4> ubDNDI_RESP(5,113)<8;4,2> { NoDDClr, NoDDChk } // copy line of U directly to memory as optimization |
||
440 | |||
441 | mov (4) mubMSGHDR_DN_OUT(4,3)<4> ubDNDI_RESP(5,96)<8;4,2> { NoDDChk } // copy line of V directly to memory as optimization |
||
442 | mov (4) mubMSGHDR_DN_OUT(4,19)<4> ubDNDI_RESP(5,112)<8;4,2> { NoDDChk } // copy line of V directly to memory as optimization |
||
443 | |||
444 | // Second 8 x 8 Block |
||
445 | mov (8) r36.0<1>:ud r31.0<8;8,1>:ud |
||
446 | add (1) r36.0<1>:ud r36.0<0;1,0>:w 0x10:w |
||
447 | |||
448 | mov (8) mubMSGHDR_DN_OUT_2(1)<2> ubDNDI_RESP(0,8)<8;8,1> { NoDDClr } // copy line of Y directly to memory as optimization |
||
449 | mov (8) mubMSGHDR_DN_OUT_2(1,16)<2> ubDNDI_RESP(0,24)<8;8,1> { NoDDClr, NoDDChk } // copy line of Y directly to memory as optimization |
||
450 | mov (8) mubMSGHDR_DN_OUT_2(2)<2> ubDNDI_RESP(0,40)<8;8,1> { NoDDClr } // copy line of Y directly to memory as optimization |
||
451 | mov (8) mubMSGHDR_DN_OUT_2(2,16)<2> ubDNDI_RESP(0,56)<8;8,1> { NoDDClr, NoDDChk } // copy line of Y directly to memory as optimization |
||
452 | mov (8) mubMSGHDR_DN_OUT_2(3)<2> ubDNDI_RESP(0,72)<8;8,1> { NoDDClr } // copy line of Y directly to memory as optimization |
||
453 | mov (8) mubMSGHDR_DN_OUT_2(3,16)<2> ubDNDI_RESP(0,88)<8;8,1> { NoDDClr, NoDDChk } // copy line of Y directly to memory as optimization |
||
454 | mov (8) mubMSGHDR_DN_OUT_2(4)<2> ubDNDI_RESP(0,104)<8;8,1> { NoDDClr } // copy line of Y directly to memory as optimization |
||
455 | mov (8) mubMSGHDR_DN_OUT_2(4,16)<2> ubDNDI_RESP(0,120)<8;8,1> { NoDDClr, NoDDChk } // copy line of Y directly to memory as optimization |
||
456 | |||
457 | mov (4) mubMSGHDR_DN_OUT_2(1,1)<4> ubDNDI_RESP(5,9)<8;4,2> { NoDDClr, NoDDChk } // copy line of U directly to memory as optimization |
||
458 | mov (4) mubMSGHDR_DN_OUT_2(1,17)<4> ubDNDI_RESP(5,25)<8;4,2> { NoDDClr, NoDDChk } // copy line of U directly to memory as optimization |
||
459 | |||
460 | mov (4) mubMSGHDR_DN_OUT_2(1,3)<4> ubDNDI_RESP(5,8)<8;4,2> { NoDDChk } // copy line of V directly to memory as optimization |
||
461 | mov (4) mubMSGHDR_DN_OUT_2(1,19)<4> ubDNDI_RESP(5,24)<8;4,2> { NoDDChk } // copy line of V directly to memory as optimization |
||
462 | mov (4) mubMSGHDR_DN_OUT_2(2,1)<4> ubDNDI_RESP(5,41)<8;4,2> { NoDDClr, NoDDChk } // copy line of U directly to memory as optimization |
||
463 | mov (4) mubMSGHDR_DN_OUT_2(2,17)<4> ubDNDI_RESP(5,57)<8;4,2> { NoDDClr, NoDDChk } // copy line of U directly to memory as optimization |
||
464 | |||
465 | mov (4) mubMSGHDR_DN_OUT_2(2,3)<4> ubDNDI_RESP(5,40)<8;4,2> { NoDDChk } // copy line of V directly to memory as optimization |
||
466 | mov (4) mubMSGHDR_DN_OUT_2(2,19)<4> ubDNDI_RESP(5,56)<8;4,2> { NoDDChk } // copy line of V directly to memory as optimization |
||
467 | mov (4) mubMSGHDR_DN_OUT_2(3,1)<4> ubDNDI_RESP(5,73)<8;4,2> { NoDDClr, NoDDChk } // copy line of U directly to memory as optimization |
||
468 | mov (4) mubMSGHDR_DN_OUT_2(3,17)<4> ubDNDI_RESP(5,89)<8;4,2> { NoDDClr, NoDDChk } // copy line of U directly to memory as optimization |
||
469 | |||
470 | mov (4) mubMSGHDR_DN_OUT_2(3,3)<4> ubDNDI_RESP(5,72)<8;4,2> { NoDDChk } // copy line of V directly to memory as optimization |
||
471 | mov (4) mubMSGHDR_DN_OUT_2(3,19)<4> ubDNDI_RESP(5,88)<8;4,2> { NoDDChk } // copy line of V directly to memory as optimization |
||
472 | mov (4) mubMSGHDR_DN_OUT_2(4,1)<4> ubDNDI_RESP(5,105)<8;4,2> { NoDDClr, NoDDChk } // copy line of U directly to memory as optimization |
||
473 | mov (4) mubMSGHDR_DN_OUT_2(4,17)<4> ubDNDI_RESP(5,121)<8;4,2> { NoDDClr, NoDDChk } // copy line of U directly to memory as optimization |
||
474 | |||
475 | mov (4) mubMSGHDR_DN_OUT_2(4,3)<4> ubDNDI_RESP(5,104)<8;4,2> { NoDDChk } // copy line of V directly to memory as optimization |
||
476 | mov (4) mubMSGHDR_DN_OUT_2(4,19)<4> ubDNDI_RESP(5,120)<8;4,2> { NoDDChk } // copy line of V directly to memory as optimization |
||
477 | |||
478 | //send out data through data port |
||
479 | send (8) null<1>:d r31.0 0x5 0xA0A801B:ud |
||
480 | send (8) null<1>:d r36.0 0x5 0xA0A801B:ud |
||
481 | |||
482 | |||
483 | |||
484 | //End of Thread message |
||
485 | |||
486 | mov (8) r127<1>:ud r0.0<8;8,1>:ud |
||
487 | send (1) null<1>:d r127 0x27 0x02000010 |
||
488 | |||
489 | |||
490 | .end_code |
||
491 | .end_kernel1>8;8,1>1>1>1>8;4,2>4>8;4,2>4>8;4,2>4>8;4,2>4>8;4,2>4>8;4,2>4>8;4,2>4>8;4,2>4>8;4,2>4>8;4,2>4>8;4,2>4>8;4,2>4>8;4,2>4>8;4,2>4>8;4,2>4>8;4,2>4>8;8,1>2>8;8,1>2>8;8,1>2>8;8,1>2>8;8,1>2>8;8,1>2>8;8,1>2>8;8,1>2>0;1,0>1>8;8,1>1>8;4,2>4>8;4,2>4>8;4,2>4>8;4,2>4>8;4,2>4>8;4,2>4>8;4,2>4>8;4,2>4>8;4,2>4>8;4,2>4>8;4,2>4>8;4,2>4>8;4,2>4>8;4,2>4>8;4,2>4>8;4,2>4>8;8,1>2>8;8,1>2>8;8,1>2>8;8,1>2>8;8,1>2>8;8,1>2>8;8,1>2>8;8,1>2>0;1,0>0;1,0>1>1>0;1,0>1>0;1,0>1>8;8,1>1>1>16;8,2>4>16;8,2>4>16;8,2>4>16;8,2>4>16;8,2>4>16;8,2>4>16;8,2>4>16;8,2>4>16;8,2>4>16;8,2>4>16;8,2>4>16;8,2>4>16;8,2>4>16;8,2>4>16;8,2>4>16;8,2>4>16;16,1>2>16;16,1>2>16;16,1>2>16;16,1>2>16;16,1>2>16;16,1>2>16;16,1>2>16;16,1>2>1>0;1,0>1>0;1,0>1>8;8,1>1>4;4,1>1>1>2;2,1>1>0;1,0>1>2;2,1>1>0;1,0>1>0;1,0>1>2;2,1>2;2,1>1>1>0;1,0>1>0;1,0>1>0;1,0>1>8;8,1>1>1>1>8;8,1>1>1>0;1,0>0;1,0>1>2;2,1>1>2;2,1>1>8;8,1>1>2;2,1>1>1>0;1,0>1>0;1,0>1>8;8,1>1>1>16;16,1>1>16;16,1>1>16;16,1>1>8;8,1>1>16;16,1>1>16;16,1>1>8;8,1>1>8;8,1>4>32;8,4>1>16;16,1>1>16;16,1>1>8;8,1>1>8;8,1>8;8,1>8;8,1>16;16,1>16;16,1> |