Debian Codegraph @

src/shaders/post_processing/gen5_6/Common/RGB16x8_Save_Y416.asm

raw
1 /*
2 * All Video Processing kernels
3 * Copyright © <2010>, Intel Corporation.
4 *
5 * Permission is hereby granted, free of charge, to any person obtaining a
6 * copy of this software and associated documentation files (the
7 * "Software"), to deal in the Software without restriction, including
8 * without limitation the rights to use, copy, modify, merge, publish,
9 * distribute, sub license, and/or sell copies of the Software, and to
10 * permit persons to whom the Software is furnished to do so, subject to
11 * the following conditions:
12 *
13 * The above copyright notice and this permission notice (including the
14 * next paragraph) shall be included in all copies or substantial portions
15 * of the Software.
16 *
17 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
18 * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
19 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
20 * IN NO EVENT SHALL PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR
21 * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
22 * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
23 * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
24 *
25 * This file was originally licensed under the following license
26 *
27 * Licensed under the Apache License, Version 2.0 (the "License");
28 * you may not use this file except in compliance with the License.
29 * You may obtain a copy of the License at
30 *
31 * http://www.apache.org/licenses/LICENSE-2.0
32 *
33 * Unless required by applicable law or agreed to in writing, software
34 * distributed under the License is distributed on an "AS IS" BASIS,
35 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
36 * See the License for the specific language governing permissions and
37 * limitations under the License.
38 *
39 */
40
41 // Module name: RGB16x8_Save_Y416.asm
42 //
43 // Save packed ARGB 444 frame data block of size 16x8
44 //
45 // To save 16x8 block (128x8 byte layout for ARGB 16bit per component) we need 4 send instructions
46 // -----------------
47 // | 1 | 2 | 3 | 4 |
48 // -----------------
49
50 #include "RGB16x8_Save_RGB.inc"
51
52 shl (1) rMSGSRC.0<1>:d wORIX<0;1,0>:w 3:w { NoDDClr } // H. block origin need to become 8 times
53 mov (1) rMSGSRC.1<1>:d wORIY<0;1,0>:w { NoDDClr, NoDDChk } // Block origin (1st quadrant)
54 mov (1) rMSGSRC.2<1>:ud nDPW_BLOCK_SIZE_ARGB:ud { NoDDChk } // Block width and height (32x8)
55
56 mov (8) mMSGHDR<1>:ud rMSGSRC<8;8,1>:ud
57 /* Not needed for validation kernels for now -vK
58 //Use the mask to determine which pixels shouldn't be over-written
59 and (1) acc0.0<1>:ud udBLOCK_MASK<0;1,0>:ud 0x00FFFFFF:ud
60 cmp.ge.f0.0(1) dNULLREG acc0.0<0;1,0>:ud 0x00FFFFFF:ud //Check if all pixels in the block need to be modified
61 (f0.0) jmpi WriteARGBToDataPort
62
63 //If mask is not all 1's, then load the entire 64x8 block
64 //so that only those bytes may be modified that need to be (using the mask)
65
66 // Load first block 16x8 packed ARGB 444 ---------------------------------------
67 or (1) acc0.0<1>:ud udBLOCK_MASK<0;1,0>:ud 0xFF00FF00:ud //Check first block
68 cmp.e.f0.0 (1) dNULLREG acc0.0<0;1,0>:ud 0xFFFFFFFF:ud
69 (f0.0) jmpi SkipFirstBlockMerge //If full mask then skip this block
70
71 send (8) udSRC_ARGB(0)<1> mMSGHDR udDUMMY_NULL nDATAPORT_READ nDPMR_MSGDSC+nDPR_MSG_SIZE_ARGB+nBI_DESTINATION_RGB:ud
72 mov (8) mMSGHDR<1>:ud rMSGSRC<8;8,1>:ud
73
74 //Merge the data
75 mov (1) f0.0:uw ubBLOCK_MASK_V:ub //Load the mask on flag reg
76 (f0.0) mov (8) rMASK_TEMP<1>:uw uwBLOCK_MASK_H:uw //use sel instruction - vK
77 (-f0.0) mov (8) rMASK_TEMP<1>:uw 0:uw
78
79 $for(0, 0; <nY_NUM_OF_ROWS; 1, 2) { //take care of the lines in the block, they are different in the src and dest
80 mov (1) f0.1:uw uwMASK_TEMP(0,%1)<0;1,0>
81 (-f0.1) mov (8) udDEST_ARGB(%2)<1> udSRC_ARGB(%1)
82 }
83
84 SkipFirstBlockMerge:
85 // Load second block 16x8 packed ARGB 444 ---------------------------------------
86 or (1) acc0.0<1>:ud udBLOCK_MASK<0;1,0>:ud 0xFF0000FF:ud //Check second block
87 cmp.e.f0.0 (1) dNULLREG acc0.0<0;1,0>:ud 0xFFFFFFFF:ud
88 (f0.0) jmpi WriteARGBToDataPort //If full mask then skip this block
89
90 add (1) mMSGHDR.0<1>:d rMSGSRC.0<0;1,0>:d 32:d // Point to 2nd part
91 send (8) udSRC_ARGB(0)<1> mMSGHDR udDUMMY_NULL nDATAPORT_READ nDPMR_MSGDSC+nDPR_MSG_SIZE_ARGB+nBI_DESTINATION_RGB:ud
92 mov (8) mMSGHDR<1>:ud rMSGSRC<8;8,1>:ud // Point to 1st part again
93
94 //Merge the data
95 mov (1) f0.0:uw ubBLOCK_MASK_V:ub //Load the mask on flag reg
96 (f0.0) shr (8) rMASK_TEMP<1>:uw uwBLOCK_MASK_H:uw 8:uw //load the mask for second block
97 (-f0.0) mov (8) rMASK_TEMP<1>:uw 0:uw
98
99 $for(0, 1; <nY_NUM_OF_ROWS; 1, 2) { //take care of the lines in the block, they are different in the src and dest
100 mov (1) f0.1:uw uwMASK_TEMP(0,%1)<0;1,0>
101 (-f0.1) mov (8) udDEST_ARGB(%2)<1> udSRC_ARGB(%1)
102 }
103 */
104 WriteARGBToDataPort:
105 // Move packed data to MRF and output
106
107 //Write 1st 4X8 pixels
108 $for(0; <nY_NUM_OF_ROWS; 1) {
109 mov (8) mudMSGPAYLOAD(%1)<1> udDEST_ARGB(%1*4)
110 }
111 send (8) dNULLREG mMSGHDR udDUMMY_NULL nDATAPORT_WRITE nDPMW_MSGDSC+nDPW_MSG_SIZE_ARGB+nBI_DESTINATION_RGB:ud
112
113 //Write 2nd 4X8 pixels
114 mov (8) mMSGHDR<1>:ud rMSGSRC<8;8,1>:ud
115 add (1) mMSGHDR.0<1>:d rMSGSRC.0<0;1,0>:d 32:d // Point to 2nd part
116 $for(0; <nY_NUM_OF_ROWS; 1) {
117 mov (8) mudMSGPAYLOAD(%1)<1> udDEST_ARGB(%1*4+1)
118 }
119 send (8) dNULLREG mMSGHDR udDUMMY_NULL nDATAPORT_WRITE nDPMW_MSGDSC+nDPW_MSG_SIZE_ARGB+nBI_DESTINATION_RGB:ud
120
121 //Write 3rd 4X8 pixels
122 mov (8) mMSGHDR<1>:ud rMSGSRC<8;8,1>:ud
123 add (1) mMSGHDR.0<1>:d rMSGSRC.0<0;1,0>:d 64:d // Point to 2nd part
124 $for(0; <nY_NUM_OF_ROWS; 1) {
125 mov (8) mudMSGPAYLOAD(%1)<1> udDEST_ARGB(%1*4+2)
126 }
127 send (8) dNULLREG mMSGHDR udDUMMY_NULL nDATAPORT_WRITE nDPMW_MSGDSC+nDPW_MSG_SIZE_ARGB+nBI_DESTINATION_RGB:ud
128
129 //Write 4th 4X8 pixels
130 mov (8) mMSGHDR<1>:ud rMSGSRC<8;8,1>:ud
131 add (1) mMSGHDR.0<1>:d rMSGSRC.0<0;1,0>:d 96:d // Point to 2nd part
132 $for(0; <nY_NUM_OF_ROWS; 1) {
133 mov (8) mudMSGPAYLOAD(%1)<1> udDEST_ARGB(%1*4+3)
134 }
135 send (8) dNULLREG mMSGHDR udDUMMY_NULL nDATAPORT_WRITE nDPMW_MSGDSC+nDPW_MSG_SIZE_ARGB+nBI_DESTINATION_RGB:ud
136
137 // End of RGB16x8_Save_Y416