| 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 |
* Authors: |
| 40 |
* Halley Zhao <[email protected]> |
| 41 |
*/ |
| 42 |
|
| 43 |
// Module name: YUVX_Save_RGBX_Float.asm |
| 44 |
//---------------------------------------------------------------- |
| 45 |
|
| 46 |
#include "RGBX_Load_16x8.inc" |
| 47 |
|
| 48 |
#if (0) |
| 49 |
// 8 grf reg for one row of pixel (2 pixel per grf) |
| 50 |
#define nTEMP0 34 |
| 51 |
#define nTEMP1 35 |
| 52 |
#define nTEMP2 36 |
| 53 |
#define nTEMP3 37 |
| 54 |
#define nTEMP4 38 |
| 55 |
#define nTEMP5 39 |
| 56 |
#define nTEMP6 40 |
| 57 |
#define nTEMP7 41 |
| 58 |
|
| 59 |
#define nTEMP8 42 // transformation coefficient |
| 60 |
#define nTEMP10 44 // transformation coefficient |
| 61 |
|
| 62 |
#define nTEMP12 46 // save Y/U/V in ub format |
| 63 |
#define nTEMP14 48 // save YUV in ud format |
| 64 |
#define nTEMP16 50 // dp4 result |
| 65 |
#define nTEMP17 51 |
| 66 |
#define nTEMP18 52 // temp used for repeat U/V in NV12_Load_8x4.asm |
| 67 |
|
| 68 |
#define nTEMP24 58 // it is not safe to use in my case. I try to use it for repeat U/V in NV12_Load_8x4.asm, Y data is taint in row 4/5 |
| 69 |
#endif |
| 70 |
|
| 71 |
#define ONE_ROW_DEBUG 0 |
| 72 |
|
| 73 |
#if (ONE_ROW_DEBUG) |
| 74 |
// if you want to debug a row which is not the first one, try the following: |
| 75 |
// 1. define ROW_NUM_READ to the row you want to debug |
| 76 |
// 2. ROW_NUM_WRITE can be same to DBG_ROWNUM_READ to overwrite original YUV data, or define it to a new row |
| 77 |
// 3. change (DBG_ROWNUM_BASE+?)=ROW_NUM_READ or ROW_NUM_WRITE to DBG_ROWNUM_0, to not conflict with others |
| 78 |
#define ROW_NUM_READ 0 |
| 79 |
#define ROW_NUM_WRITE 0 |
| 80 |
#define DBG_ROWNUM_BASE 1 |
| 81 |
#define DBG_ROWNUM_0 0 |
| 82 |
#else |
| 83 |
#define ROW_NUM_READ %1 |
| 84 |
#define ROW_NUM_WRITE %1 |
| 85 |
$for(0; <nY_NUM_OF_ROWS; 1) { |
| 86 |
#endif |
| 87 |
// YUVX | Y | U | V | X | |
| 88 |
// XRGB | B | G | R | X | |
| 89 |
// ###### save one row of pixel to temp grf with float format (required by dp4) |
| 90 |
// C = Y' - 16 D = U - 128 E = V - 128 |
| 91 |
|
| 92 |
// the follow sentence doesn't work, I have to split it into two step |
| 93 |
// add (4) REG(r, nTEMP0)<1>:f r[SRC_RGBA_OFFSET_1,ROW_NUM_READ*32 + 0]<4,1>:ub REG2(r, nTEMP10, 16)<0;4,1>:b |
| 94 |
|
| 95 |
add (16) REG2(r,nTEMP12,0)<1>:w ubDEST_RGBX(0,ROW_NUM_READ*64 )<0;16,1> bYUV_OFF<0;4,1>:b |
| 96 |
add (16) REG2(r,nTEMP14,0)<1>:w ubDEST_RGBX(0,ROW_NUM_READ*64+16)<0;16,1> bYUV_OFF<0;4,1>:b |
| 97 |
add (16) REG2(r,nTEMP16,0)<1>:w ubDEST_RGBX(0,ROW_NUM_READ*64+32)<0;16,1> bYUV_OFF<0;4,1>:b |
| 98 |
add (16) REG2(r,nTEMP17,0)<1>:w ubDEST_RGBX(0,ROW_NUM_READ*64+48)<0;16,1> bYUV_OFF<0;4,1>:b |
| 99 |
|
| 100 |
mov (8) fROW_YUVA(0,0)<1> REG2(r, nTEMP12, 0)<0;8,1>:w |
| 101 |
mov (8) fROW_YUVA(1,0)<1> REG2(r, nTEMP12, 8)<0;8,1>:w |
| 102 |
mov (8) fROW_YUVA(2,0)<1> REG2(r, nTEMP14, 0)<0;8,1>:w |
| 103 |
mov (8) fROW_YUVA(3,0)<1> REG2(r, nTEMP14, 8)<0;8,1>:w |
| 104 |
mov (8) fROW_YUVA(4,0)<1> REG2(r, nTEMP16, 0)<0;8,1>:w |
| 105 |
mov (8) fROW_YUVA(5,0)<1> REG2(r, nTEMP16, 8)<0;8,1>:w |
| 106 |
mov (8) fROW_YUVA(6,0)<1> REG2(r, nTEMP17, 0)<0;8,1>:w |
| 107 |
mov (8) fROW_YUVA(7,0)<1> REG2(r, nTEMP17, 8)<0;8,1>:w |
| 108 |
|
| 109 |
#if (ONE_ROW_DEBUG) |
| 110 |
mov.sat (8) REG2(r, nTEMP14, 0)<1>:ud fROW_YUVA(0,0)<0;8,1>:f |
| 111 |
mov (8) REG2(r, nTEMP12, 0)<1>:ub REG2(r, nTEMP14, 0)<0;8,4>:ub |
| 112 |
|
| 113 |
// write Y-16, U-128, V-128 to the 2nd row of RGB (convert float to int first, write whole ud): 1st half, 2 pixels |
| 114 |
mov (16) ubDEST_RGBX(0,(DBG_ROWNUM_BASE)*64 )<1> REG2(r,nTEMP14, 0)<0;16,1>:ub |
| 115 |
mov (16) ubDEST_RGBX(0,(DBG_ROWNUM_BASE)*64+16)<1> REG2(r,nTEMP14, 16)<0;16,1>:ub |
| 116 |
|
| 117 |
mov.sat (8) REG2(r, nTEMP14, 0)<1>:ud fROW_YUVA(1,0)<0;8,1>:f |
| 118 |
mov (8) REG2(r, nTEMP12, 8)<1>:ub REG2(r, nTEMP14, 0)<0;8,4>:ub |
| 119 |
|
| 120 |
// write Y-16, U-128, V-128 to the 2nd row of RGB (convert float to int first, write whole ud): 2nd half, 2 pixels |
| 121 |
mov (16) ubDEST_RGBX(0,(DBG_ROWNUM_BASE)*64+32)<1> REG2(r,nTEMP14, 0)<0;16,1>:ub |
| 122 |
mov (16) ubDEST_RGBX(0,(DBG_ROWNUM_BASE)*64+48)<1> REG2(r,nTEMP14, 16)<0;16,1>:ub |
| 123 |
|
| 124 |
mov.sat (8) REG2(r, nTEMP14, 0)<1>:ud fROW_YUVA(2,0)<0;8,1>:f |
| 125 |
mov (8) REG2(r, nTEMP12, 16)<1>:ub REG2(r, nTEMP14, 0)<0;8,4>:ub |
| 126 |
|
| 127 |
mov.sat (8) REG2(r, nTEMP14, 0)<1>:ud fROW_YUVA(3,0)<0;8,1>:f |
| 128 |
mov (8) REG2(r, nTEMP12, 24)<1>:ub REG2(r, nTEMP14, 0)<0;8,4>:ub |
| 129 |
|
| 130 |
// write Y-16, U-128, V-128 to the 3rd row of RGB (convert float to int first, only LSB is used): 1st half, 8 pixels |
| 131 |
mov (16) ubDEST_RGBX(0,(DBG_ROWNUM_BASE+1)*64 )<1> REG2(r,nTEMP12, 0)<0;16,1>:ub |
| 132 |
mov (16) ubDEST_RGBX(0,(DBG_ROWNUM_BASE+1)*64+16)<1> REG2(r,nTEMP12, 16)<0;16,1>:ub |
| 133 |
|
| 134 |
mov.sat (8) REG2(r, nTEMP14, 0)<1>:ud fROW_YUVA(4,0)<0;8,1>:f |
| 135 |
mov (8) REG2(r, nTEMP12, 0)<1>:ub REG2(r, nTEMP14, 0)<0;8,4>:ub |
| 136 |
|
| 137 |
mov.sat (8) REG2(r, nTEMP14, 0)<1>:ud fROW_YUVA(5,0)<0;8,1>:f |
| 138 |
mov (8) REG2(r, nTEMP12, 8)<1>:ub REG2(r, nTEMP14, 0)<0;8,4>:ub |
| 139 |
|
| 140 |
mov.sat (8) REG2(r, nTEMP14, 0)<1>:ud fROW_YUVA(6,0)<0;8,1>:f |
| 141 |
mov (8) REG2(r, nTEMP12, 16)<1>:ub REG2(r, nTEMP14, 0)<0;8,4>:ub |
| 142 |
|
| 143 |
mov.sat (8) REG2(r, nTEMP14, 0)<1>:ud fROW_YUVA(7,0)<0;8,1>:f |
| 144 |
mov (8) REG2(r, nTEMP12, 24)<1>:ub REG2(r, nTEMP14, 0)<0;8,4>:ub |
| 145 |
|
| 146 |
// write Y-16, U-128, V-128 to the 3rd row of RGB (convert float to int first, only LSB is used): 2nd half, 8 pixels |
| 147 |
mov (16) ubDEST_RGBX(0,(DBG_ROWNUM_BASE+1)*64+32)<1> REG2(r,nTEMP12, 0)<0;16,1>:ub |
| 148 |
mov (16) ubDEST_RGBX(0,(DBG_ROWNUM_BASE+1)*64+48)<1> REG2(r,nTEMP12, 16)<0;16,1>:ub |
| 149 |
#endif |
| 150 |
|
| 151 |
// ######## do one row for Red ######## |
| 152 |
#define fCOEF_REG fYUV_to_RGB_CH2_Coef_Float |
| 153 |
#define CHANNEL 2 |
| 154 |
// ##### dp4(nTEMP16) and save result to uw format(nTEMP12) |
| 155 |
dp4 (8) REG2(r, nTEMP16, 0)<1>:f fROW_YUVA(0, 0)<0;8,1> fCOEF_REG<0;4,1>:f |
| 156 |
mov.sat (2) REG2(r, nTEMP14, 0)<1>:ud REG2(r, nTEMP16, 0)<0;2,4>:f |
| 157 |
mov (2) REG2(r, nTEMP12, 0)<1>:ub REG2(r, nTEMP14, 0)<0;2,4>:ub |
| 158 |
|
| 159 |
#if (ONE_ROW_DEBUG) |
| 160 |
// write dp4 (raw float) of 2 pixel to the 4/5th row |
| 161 |
mov (8) ubDEST_RGBX(0,(DBG_ROWNUM_BASE+2)*64+CHANNEL )<4> REG2(r,nTEMP16, 0)<0;8,1>:ub |
| 162 |
mov (8) ubDEST_RGBX(0,(DBG_ROWNUM_BASE+2)*64+CHANNEL+32)<4> REG2(r,nTEMP16, 8)<0;8,1>:ub |
| 163 |
mov (8) ubDEST_RGBX(0,(DBG_ROWNUM_BASE+3)*64+CHANNEL )<4> REG2(r,nTEMP16, 16)<0;8,1>:ub |
| 164 |
mov (8) ubDEST_RGBX(0,(DBG_ROWNUM_BASE+3)*64+CHANNEL+32)<4> REG2(r,nTEMP16, 24)<0;8,1>:ub |
| 165 |
|
| 166 |
// write dp4 (convert float to ud first, write whole ud) of 2 pixel to the 6/7th row |
| 167 |
mov (8) REG2(r, nTEMP17, 0)<1>:d REG2(r, nTEMP16, 0)<0;8,1>:f |
| 168 |
mov (8) ubDEST_RGBX(0,(DBG_ROWNUM_BASE+4)*64+CHANNEL )<4> REG2(r,nTEMP17, 0)<0;8,1>:ub |
| 169 |
mov (8) ubDEST_RGBX(0,(DBG_ROWNUM_BASE+4)*64+CHANNEL+32)<4> REG2(r,nTEMP17, 8)<0;8,1>:ub |
| 170 |
mov (8) ubDEST_RGBX(0,(DBG_ROWNUM_BASE+5)*64+CHANNEL )<4> REG2(r,nTEMP17, 16)<0;8,1>:ub |
| 171 |
mov (8) ubDEST_RGBX(0,(DBG_ROWNUM_BASE+5)*64+CHANNEL+32)<4> REG2(r,nTEMP17, 24)<0;8,1>:ub |
| 172 |
#endif |
| 173 |
|
| 174 |
dp4 (8) REG2(r, nTEMP16, 0)<1>:f fROW_YUVA(1, 0)<0;8,1> fCOEF_REG<0;4,1>:f |
| 175 |
mov.sat (2) REG2(r, nTEMP14, 0)<1>:ud REG2(r, nTEMP16, 0)<0;2,4>:f |
| 176 |
mov (2) REG2(r, nTEMP12, 2)<1>:ub REG2(r, nTEMP14, 0)<0;2,4>:ub |
| 177 |
|
| 178 |
dp4 (8) REG2(r, nTEMP16, 0)<1>:f fROW_YUVA(2, 0)<0;8,1> fCOEF_REG<0;4,1>:f |
| 179 |
mov.sat (2) REG2(r, nTEMP14, 0)<1>:ud REG2(r, nTEMP16, 0)<0;2,4>:f |
| 180 |
mov (2) REG2(r, nTEMP12, 4)<1>:ub REG2(r, nTEMP14, 0)<0;2,4>:ub |
| 181 |
|
| 182 |
dp4 (8) REG2(r, nTEMP16, 0)<1>:f fROW_YUVA(3, 0)<0;8,1> fCOEF_REG<0;4,1>:f |
| 183 |
mov.sat (2) REG2(r, nTEMP14, 0)<1>:ud REG2(r, nTEMP16, 0)<0;2,4>:f |
| 184 |
mov (2) REG2(r, nTEMP12, 6)<1>:ub REG2(r, nTEMP14, 0)<0;2,4>:ub |
| 185 |
|
| 186 |
dp4 (8) REG2(r, nTEMP16, 0)<1>:f fROW_YUVA(4, 0)<0;8,1> fCOEF_REG<0;4,1>:f |
| 187 |
mov.sat (2) REG2(r, nTEMP14, 0)<1>:ud REG2(r, nTEMP16, 0)<0;2,4>:f |
| 188 |
mov (2) REG2(r, nTEMP12, 8)<1>:ub REG2(r, nTEMP14, 0)<0;2,4>:ub |
| 189 |
|
| 190 |
dp4 (8) REG2(r, nTEMP16, 0)<1>:f fROW_YUVA(5, 0)<5;8,1> fCOEF_REG<0;4,1>:f |
| 191 |
mov.sat (2) REG2(r, nTEMP14, 0)<1>:ud REG2(r, nTEMP16, 0)<0;2,4>:f |
| 192 |
mov (2) REG2(r, nTEMP12, 10)<1>:ub REG2(r, nTEMP14, 0)<0;2,4>:ub |
| 193 |
|
| 194 |
dp4 (8) REG2(r, nTEMP16, 0)<1>:f fROW_YUVA(6, 0)<6;8,1> fCOEF_REG<0;4,1>:f |
| 195 |
mov.sat (2) REG2(r, nTEMP14, 0)<1>:ud REG2(r, nTEMP16, 0)<0;2,4>:f |
| 196 |
mov (2) REG2(r, nTEMP12, 12)<1>:ub REG2(r, nTEMP14, 0)<0;2,4>:ub |
| 197 |
|
| 198 |
dp4 (8) REG2(r, nTEMP16, 0)<1>:f fROW_YUVA(7, 0)<0;8,1> fCOEF_REG<0;4,1>:f |
| 199 |
mov.sat (2) REG2(r, nTEMP14, 0)<1>:ud REG2(r, nTEMP16, 0)<0;2,4>:f |
| 200 |
mov (2) REG2(r, nTEMP12, 14)<1>:ub REG2(r, nTEMP14, 0)<0;2,4>:ub |
| 201 |
|
| 202 |
// #### write this channel |
| 203 |
mov (8) ubDEST_RGBX(0,ROW_NUM_WRITE*64+CHANNEL )<4> REG2(r,nTEMP12, 0)<0;8,1>:ub |
| 204 |
mov (8) ubDEST_RGBX(0,ROW_NUM_WRITE*64+CHANNEL+32)<4> REG2(r,nTEMP12, 8)<0;8,1>:ub |
| 205 |
|
| 206 |
// ######## do one row for Green ######## |
| 207 |
#define fCOEF_REG fYUV_to_RGB_CH1_Coef_Float // reg for green coefficient |
| 208 |
#define CHANNEL 1 |
| 209 |
// ##### dp4(nTEMP16) and save result to uw format(nTEMP12) |
| 210 |
dp4 (8) REG2(r, nTEMP16, 0)<1>:f fROW_YUVA(0, 0)<0;8,1> fCOEF_REG<0;4,1>:f |
| 211 |
mov.sat (2) REG2(r, nTEMP14, 0)<1>:ud REG2(r, nTEMP16, 0)<0;2,4>:f |
| 212 |
mov (2) REG2(r, nTEMP12, 0)<1>:ub REG2(r, nTEMP14, 0)<0;2,4>:ub |
| 213 |
|
| 214 |
#if (ONE_ROW_DEBUG) |
| 215 |
// write dp4 (raw float) of 2 pixel to the 4/5th row |
| 216 |
mov (8) ubDEST_RGBX(0,(DBG_ROWNUM_BASE+2)*64+CHANNEL )<4> REG2(r,nTEMP16, 0)<0;8,1>:ub |
| 217 |
mov (8) ubDEST_RGBX(0,(DBG_ROWNUM_BASE+2)*64+CHANNEL+32)<4> REG2(r,nTEMP16, 8)<0;8,1>:ub |
| 218 |
mov (8) ubDEST_RGBX(0,(DBG_ROWNUM_BASE+3)*64+CHANNEL )<4> REG2(r,nTEMP16, 16)<0;8,1>:ub |
| 219 |
mov (8) ubDEST_RGBX(0,(DBG_ROWNUM_BASE+3)*64+CHANNEL+32)<4> REG2(r,nTEMP16, 24)<0;8,1>:ub |
| 220 |
|
| 221 |
// write dp4 (convert float to ud first, write whole ud) of 2 pixel to the 6/7th row |
| 222 |
mov (8) REG2(r, nTEMP17, 0)<1>:d REG2(r, nTEMP16, 0)<0;8,1>:f |
| 223 |
mov (8) ubDEST_RGBX(0,(DBG_ROWNUM_BASE+4)*64+CHANNEL )<4> REG2(r,nTEMP17, 0)<0;8,1>:ub |
| 224 |
mov (8) ubDEST_RGBX(0,(DBG_ROWNUM_BASE+4)*64+CHANNEL+32)<4> REG2(r,nTEMP17, 8)<0;8,1>:ub |
| 225 |
mov (8) ubDEST_RGBX(0,(DBG_ROWNUM_BASE+5)*64+CHANNEL )<4> REG2(r,nTEMP17, 16)<0;8,1>:ub |
| 226 |
mov (8) ubDEST_RGBX(0,(DBG_ROWNUM_BASE+5)*64+CHANNEL+32)<4> REG2(r,nTEMP17, 24)<0;8,1>:ub |
| 227 |
#endif |
| 228 |
|
| 229 |
dp4 (8) REG2(r, nTEMP16, 0)<1>:f fROW_YUVA(1, 0)<0;8,1> fCOEF_REG<0;4,1>:f |
| 230 |
mov.sat (2) REG2(r, nTEMP14, 0)<1>:ud REG2(r, nTEMP16, 0)<0;2,4>:f |
| 231 |
mov (2) REG2(r, nTEMP12, 2)<1>:ub REG2(r, nTEMP14, 0)<0;2,4>:ub |
| 232 |
|
| 233 |
dp4 (8) REG2(r, nTEMP16, 0)<1>:f fROW_YUVA(2, 0)<0;8,1> fCOEF_REG<0;4,1>:f |
| 234 |
mov.sat (2) REG2(r, nTEMP14, 0)<1>:ud REG2(r, nTEMP16, 0)<0;2,4>:f |
| 235 |
mov (2) REG2(r, nTEMP12, 4)<1>:ub REG2(r, nTEMP14, 0)<0;2,4>:ub |
| 236 |
|
| 237 |
dp4 (8) REG2(r, nTEMP16, 0)<1>:f fROW_YUVA(3, 0)<0;8,1> fCOEF_REG<0;4,1>:f |
| 238 |
mov.sat (2) REG2(r, nTEMP14, 0)<1>:ud REG2(r, nTEMP16, 0)<0;2,4>:f |
| 239 |
mov (2) REG2(r, nTEMP12, 6)<1>:ub REG2(r, nTEMP14, 0)<0;2,4>:ub |
| 240 |
|
| 241 |
dp4 (8) REG2(r, nTEMP16, 0)<1>:f fROW_YUVA(4, 0)<0;8,1> fCOEF_REG<0;4,1>:f |
| 242 |
mov.sat (2) REG2(r, nTEMP14, 0)<1>:ud REG2(r, nTEMP16, 0)<0;2,4>:f |
| 243 |
mov (2) REG2(r, nTEMP12, 8)<1>:ub REG2(r, nTEMP14, 0)<0;2,4>:ub |
| 244 |
|
| 245 |
dp4 (8) REG2(r, nTEMP16, 0)<1>:f fROW_YUVA(5, 0)<5;8,1> fCOEF_REG<0;4,1>:f |
| 246 |
mov.sat (2) REG2(r, nTEMP14, 0)<1>:ud REG2(r, nTEMP16, 0)<0;2,4>:f |
| 247 |
mov (2) REG2(r, nTEMP12, 10)<1>:ub REG2(r, nTEMP14, 0)<0;2,4>:ub |
| 248 |
|
| 249 |
dp4 (8) REG2(r, nTEMP16, 0)<1>:f fROW_YUVA(6, 0)<6;8,1> fCOEF_REG<0;4,1>:f |
| 250 |
mov.sat (2) REG2(r, nTEMP14, 0)<1>:ud REG2(r, nTEMP16, 0)<0;2,4>:f |
| 251 |
mov (2) REG2(r, nTEMP12, 12)<1>:ub REG2(r, nTEMP14, 0)<0;2,4>:ub |
| 252 |
|
| 253 |
dp4 (8) REG2(r, nTEMP16, 0)<1>:f fROW_YUVA(7, 0)<0;8,1> fCOEF_REG<0;4,1>:f |
| 254 |
mov.sat (2) REG2(r, nTEMP14, 0)<1>:ud REG2(r, nTEMP16, 0)<0;2,4>:f |
| 255 |
mov (2) REG2(r, nTEMP12, 14)<1>:ub REG2(r, nTEMP14, 0)<0;2,4>:ub |
| 256 |
|
| 257 |
// #### write this channel |
| 258 |
mov (8) ubDEST_RGBX(0,ROW_NUM_WRITE*64+CHANNEL )<4> REG2(r,nTEMP12, 0)<0;8,1>:ub |
| 259 |
mov (8) ubDEST_RGBX(0,ROW_NUM_WRITE*64+CHANNEL+32)<4> REG2(r,nTEMP12, 8)<0;8,1>:ub |
| 260 |
|
| 261 |
// ###### do one row for Blue channel |
| 262 |
#define fCOEF_REG fYUV_to_RGB_CH0_Coef_Float // reg for Blue coefficient |
| 263 |
#define CHANNEL 0 |
| 264 |
// ##### dp4(nTEMP16) and save result to uw format(nTEMP12) |
| 265 |
dp4 (8) REG2(r, nTEMP16, 0)<1>:f fROW_YUVA(0, 0)<0;8,1> fCOEF_REG<0;4,1>:f |
| 266 |
mov.sat (2) REG2(r, nTEMP14, 0)<1>:ud REG2(r, nTEMP16, 0)<0;2,4>:f |
| 267 |
mov (2) REG2(r, nTEMP12, 0)<1>:ub REG2(r, nTEMP14, 0)<0;2,4>:ub |
| 268 |
|
| 269 |
#if (ONE_ROW_DEBUG) |
| 270 |
// write dp4 (raw float) of 2 pixel to the 4/5th row |
| 271 |
mov (8) ubDEST_RGBX(0,(DBG_ROWNUM_BASE+2)*64+CHANNEL )<4> REG2(r,nTEMP16, 0)<0;8,1>:ub |
| 272 |
mov (8) ubDEST_RGBX(0,(DBG_ROWNUM_BASE+2)*64+CHANNEL+32)<4> REG2(r,nTEMP16, 8)<0;8,1>:ub |
| 273 |
mov (8) ubDEST_RGBX(0,(DBG_ROWNUM_BASE+3)*64+CHANNEL )<4> REG2(r,nTEMP16, 16)<0;8,1>:ub |
| 274 |
mov (8) ubDEST_RGBX(0,(DBG_ROWNUM_BASE+3)*64+CHANNEL+32)<4> REG2(r,nTEMP16, 24)<0;8,1>:ub |
| 275 |
|
| 276 |
// write dp4 (convert float to ud first, write whole ud) of 2 pixel to the 6/7th row |
| 277 |
mov (8) REG2(r, nTEMP17, 0)<1>:d REG2(r, nTEMP16, 0)<0;8,1>:f |
| 278 |
mov (8) ubDEST_RGBX(0,(DBG_ROWNUM_BASE+4)*64+CHANNEL )<4> REG2(r,nTEMP17, 0)<0;8,1>:ub |
| 279 |
mov (8) ubDEST_RGBX(0,(DBG_ROWNUM_BASE+4)*64+CHANNEL+32)<4> REG2(r,nTEMP17, 8)<0;8,1>:ub |
| 280 |
mov (8) ubDEST_RGBX(0,(DBG_ROWNUM_BASE+5)*64+CHANNEL )<4> REG2(r,nTEMP17, 16)<0;8,1>:ub |
| 281 |
mov (8) ubDEST_RGBX(0,(DBG_ROWNUM_BASE+5)*64+CHANNEL+32)<4> REG2(r,nTEMP17, 24)<0;8,1>:ub |
| 282 |
#endif |
| 283 |
|
| 284 |
dp4 (8) REG2(r, nTEMP16, 0)<1>:f fROW_YUVA(1, 0)<0;8,1> fCOEF_REG<0;4,1>:f |
| 285 |
mov.sat (2) REG2(r, nTEMP14, 0)<1>:ud REG2(r, nTEMP16, 0)<0;2,4>:f |
| 286 |
mov (2) REG2(r, nTEMP12, 2)<1>:ub REG2(r, nTEMP14, 0)<0;2,4>:ub |
| 287 |
|
| 288 |
dp4 (8) REG2(r, nTEMP16, 0)<1>:f fROW_YUVA(2, 0)<0;8,1> fCOEF_REG<0;4,1>:f |
| 289 |
mov.sat (2) REG2(r, nTEMP14, 0)<1>:ud REG2(r, nTEMP16, 0)<0;2,4>:f |
| 290 |
mov (2) REG2(r, nTEMP12, 4)<1>:ub REG2(r, nTEMP14, 0)<0;2,4>:ub |
| 291 |
|
| 292 |
dp4 (8) REG2(r, nTEMP16, 0)<1>:f fROW_YUVA(3, 0)<0;8,1> fCOEF_REG<0;4,1>:f |
| 293 |
mov.sat (2) REG2(r, nTEMP14, 0)<1>:ud REG2(r, nTEMP16, 0)<0;2,4>:f |
| 294 |
mov (2) REG2(r, nTEMP12, 6)<1>:ub REG2(r, nTEMP14, 0)<0;2,4>:ub |
| 295 |
|
| 296 |
dp4 (8) REG2(r, nTEMP16, 0)<1>:f fROW_YUVA(4, 0)<0;8,1> fCOEF_REG<0;4,1>:f |
| 297 |
mov.sat (2) REG2(r, nTEMP14, 0)<1>:ud REG2(r, nTEMP16, 0)<0;2,4>:f |
| 298 |
mov (2) REG2(r, nTEMP12, 8)<1>:ub REG2(r, nTEMP14, 0)<0;2,4>:ub |
| 299 |
|
| 300 |
dp4 (8) REG2(r, nTEMP16, 0)<1>:f fROW_YUVA(5, 0)<5;8,1> fCOEF_REG<0;4,1>:f |
| 301 |
mov.sat (2) REG2(r, nTEMP14, 0)<1>:ud REG2(r, nTEMP16, 0)<0;2,4>:f |
| 302 |
mov (2) REG2(r, nTEMP12, 10)<1>:ub REG2(r, nTEMP14, 0)<0;2,4>:ub |
| 303 |
|
| 304 |
dp4 (8) REG2(r, nTEMP16, 0)<1>:f fROW_YUVA(6, 0)<6;8,1> fCOEF_REG<0;4,1>:f |
| 305 |
mov.sat (2) REG2(r, nTEMP14, 0)<1>:ud REG2(r, nTEMP16, 0)<0;2,4>:f |
| 306 |
mov (2) REG2(r, nTEMP12, 12)<1>:ub REG2(r, nTEMP14, 0)<0;2,4>:ub |
| 307 |
|
| 308 |
dp4 (8) REG2(r, nTEMP16, 0)<1>:f fROW_YUVA(7, 0)<0;8,1> fCOEF_REG<0;4,1>:f |
| 309 |
mov.sat (2) REG2(r, nTEMP14, 0)<1>:ud REG2(r, nTEMP16, 0)<0;2,4>:f |
| 310 |
mov (2) REG2(r, nTEMP12, 14)<1>:ub REG2(r, nTEMP14, 0)<0;2,4>:ub |
| 311 |
|
| 312 |
// #### write this channel |
| 313 |
mov (8) ubDEST_RGBX(0,ROW_NUM_WRITE*64+CHANNEL )<4> REG2(r,nTEMP12, 0)<0;8,1>:ub |
| 314 |
mov (8) ubDEST_RGBX(0,ROW_NUM_WRITE*64+CHANNEL+32)<4> REG2(r,nTEMP12, 8)<0;8,1>:ub |
| 315 |
#if (!ONE_ROW_DEBUG) |
| 316 |
} |
| 317 |
#endif |