1 /* 2 * Copyright (c) 2019, Intel Corporation 3 * 4 * Permission is hereby granted, free of charge, to any person obtaining a 5 * copy of this software and associated documentation files (the "Software"), 6 * to deal in the Software without restriction, including without limitation 7 * the rights to use, copy, modify, merge, publish, distribute, sublicense, 8 * and/or sell copies of the Software, and to permit persons to whom the 9 * Software is furnished to do so, subject to the following conditions: 10 * 11 * The above copyright notice and this permission notice shall be included 12 * in all copies or substantial portions of the Software. 13 * 14 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS 15 * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, 16 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL 17 * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR 18 * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, 19 * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR 20 * OTHER DEALINGS IN THE SOFTWARE. 21 */ 22 { 23 matrix<ushort, 1, 16> Temp; 24 25 #pragma unroll 26 for (int i = 0; i < 4; i++) 27 { 28 /* 29 Buffer layout after shuffle 30 _________________________________________________ 31 |_______Block0__________|_______Block1__________| 32 |_______Block2__________|_______Block3__________| 33 |_______Block4__________|_______Block5__________| 34 |_______Block6__________|_______Block7__________| 35 36 Write back buffer layout correlate to the block number#, each box stands for 1 GRF 37 _______________________________________________ 38 |____R0_________R1_____|____R2_________R3_____| 39 |____G0_________G1_____|____G2_________G3_____| 40 |____B0_________B1_____|____B2_________B3_____| 41 |____A0_________A1_____|____A2_________A3_____| 42 |____R4_________R5_____|____R6_________R7_____| 43 |____G4_________G5_____|____G6_________G7_____| 44 |____B4_________B5_____|____B6_________B7_____| 45 |____A4_________A5_____|____A6_________A7_____| 46 */ 47 { 48 matrix_ref<ushort, 1, 16> Alpha = DataBuffer.format<ushort, 96, 16>().select<1, 1, 16, 1>(16 * i + 6, 0); 49 50 // R0/G0/B0/A0 51 Temp = (((DataBuffer.format<ushort, 96, 16>().select<1, 1, 16, 1>(16 * i + Channel_Offset_RV_0, 0) * (cm_add<ushort>(Alpha, 256, SAT)))) + (((ColorFill(0) << 8) * (cm_add<ushort>(0xFF00, -Alpha, SAT))))) >> 16; 52 DataBuffer.format<ushort, 96, 16>().select<1, 1, 16, 1>(16 * i + Channel_Offset_RV_0, 0).merge(Temp, (ColorFill(0) << 8), TempMask0[0][4 * i]); 53 Temp = (((DataBuffer.format<ushort, 96, 16>().select<1, 1, 16, 1>(16 * i + Channel_Offset_GY_0, 0) * (cm_add<ushort>(Alpha, 256, SAT)))) + (((ColorFill(1) << 8) * (cm_add<ushort>(0xFF00, -Alpha, SAT))))) >> 16; 54 DataBuffer.format<ushort, 96, 16>().select<1, 1, 16, 1>(16 * i + Channel_Offset_GY_0, 0).merge(Temp, (ColorFill(1) << 8), TempMask0[0][4 * i]); 55 Temp = (((DataBuffer.format<ushort, 96, 16>().select<1, 1, 16, 1>(16 * i + Channel_Offset_BU_0, 0) * (cm_add<ushort>(Alpha, 256, SAT)))) + (((ColorFill(2) << 8) * (cm_add<ushort>(0xFF00, -Alpha, SAT))))) >> 16; 56 DataBuffer.format<ushort, 96, 16>().select<1, 1, 16, 1>(16 * i + Channel_Offset_BU_0, 0).merge(Temp, (ColorFill(2) << 8), TempMask0[0][4 * i]); 57 Temp = (((DataBuffer.format<ushort, 96, 16>().select<1, 1, 16, 1>(16 * i + Channel_Offset_A_0, 0) * (cm_add<ushort>(Alpha, 256, SAT)))) + (((ColorFill(3) << 8) * (cm_add<ushort>(0xFF00, -Alpha, SAT))))) >> 16; 58 DataBuffer.format<ushort, 96, 16>().select<1, 1, 16, 1>(16 * i + Channel_Offset_A_0, 0).merge(Temp, (ColorFill(3) << 8), TempMask0[0][4 * i]); 59 } 60 61 { 62 matrix_ref<ushort, 1, 16> Alpha = DataBuffer.format<ushort, 96, 16>().select<1, 1, 16, 1>(16 * i + 7, 0); 63 64 // R1/G1/B1/A1 65 Temp = (((DataBuffer.format<ushort, 96, 16>().select<1, 1, 16, 1>(16 * i + Channel_Offset_RV_1, 0) * (cm_add<ushort>(Alpha, 256, SAT)))) + (((ColorFill(0) << 8) * (cm_add<ushort>(0xFF00, -Alpha, SAT))))) >> 16; 66 DataBuffer.format<ushort, 96, 16>().select<1, 1, 16, 1>(16 * i + Channel_Offset_RV_1, 0).merge(Temp, (ColorFill(0) << 8), TempMask0[0][4 * i + 1]); 67 Temp = (((DataBuffer.format<ushort, 96, 16>().select<1, 1, 16, 1>(16 * i + Channel_Offset_GY_1, 0) * (cm_add<ushort>(Alpha, 256, SAT)))) + (((ColorFill(1) << 8) * (cm_add<ushort>(0xFF00, -Alpha, SAT))))) >> 16; 68 DataBuffer.format<ushort, 96, 16>().select<1, 1, 16, 1>(16 * i + Channel_Offset_GY_1, 0).merge(Temp, (ColorFill(1) << 8), TempMask0[0][4 * i + 1]); 69 Temp = (((DataBuffer.format<ushort, 96, 16>().select<1, 1, 16, 1>(16 * i + Channel_Offset_BU_1, 0) * (cm_add<ushort>(Alpha, 256, SAT)))) + (((ColorFill(2) << 8) * (cm_add<ushort>(0xFF00, -Alpha, SAT))))) >> 16; 70 DataBuffer.format<ushort, 96, 16>().select<1, 1, 16, 1>(16 * i + Channel_Offset_BU_1, 0).merge(Temp, (ColorFill(2) << 8), TempMask0[0][4 * i + 1]); 71 Temp = (((DataBuffer.format<ushort, 96, 16>().select<1, 1, 16, 1>(16 * i + Channel_Offset_A_1, 0) * (cm_add<ushort>(Alpha, 256, SAT)))) + (((ColorFill(3) << 8) * (cm_add<ushort>(0xFF00, -Alpha, SAT))))) >> 16; 72 DataBuffer.format<ushort, 96, 16>().select<1, 1, 16, 1>(16 * i + Channel_Offset_A_1, 0).merge(Temp, (ColorFill(3) << 8), TempMask0[0][4 * i + 1]); 73 } 74 75 { 76 matrix_ref<ushort, 1, 16> Alpha = DataBuffer.format<ushort, 96, 16>().select<1, 1, 16, 1>(16 * i + 14, 0); 77 78 // R2/G2/B2/A2 79 Temp = (((DataBuffer.format<ushort, 96, 16>().select<1, 1, 16, 1>(16 * i + Channel_Offset_RV_2, 0) * (cm_add<ushort>(Alpha, 256, SAT)))) + (((ColorFill(0) << 8) * (cm_add<ushort>(0xFF00, -Alpha, SAT))))) >> 16; 80 DataBuffer.format<ushort, 96, 16>().select<1, 1, 16, 1>(16 * i + Channel_Offset_RV_2, 0).merge(Temp, (ColorFill(0) << 8), TempMask0[0][4 * i + 2]); 81 Temp = (((DataBuffer.format<ushort, 96, 16>().select<1, 1, 16, 1>(16 * i + Channel_Offset_GY_2, 0) * (cm_add<ushort>(Alpha, 256, SAT)))) + (((ColorFill(1) << 8) * (cm_add<ushort>(0xFF00, -Alpha, SAT))))) >> 16; 82 DataBuffer.format<ushort, 96, 16>().select<1, 1, 16, 1>(16 * i + Channel_Offset_GY_2, 0).merge(Temp, (ColorFill(1) << 8), TempMask0[0][4 * i + 2]); 83 Temp = (((DataBuffer.format<ushort, 96, 16>().select<1, 1, 16, 1>(16 * i + Channel_Offset_BU_2, 0) * (cm_add<ushort>(Alpha, 256, SAT)))) + (((ColorFill(2) << 8) * (cm_add<ushort>(0xFF00, -Alpha, SAT))))) >> 16; 84 DataBuffer.format<ushort, 96, 16>().select<1, 1, 16, 1>(16 * i + Channel_Offset_BU_2, 0).merge(Temp, (ColorFill(2) << 8), TempMask0[0][4 * i + 2]); 85 Temp = (((DataBuffer.format<ushort, 96, 16>().select<1, 1, 16, 1>(16 * i + Channel_Offset_A_2, 0) * (cm_add<ushort>(Alpha, 256, SAT)))) + (((ColorFill(3) << 8) * (cm_add<ushort>(0xFF00, -Alpha, SAT))))) >> 16; 86 DataBuffer.format<ushort, 96, 16>().select<1, 1, 16, 1>(16 * i + Channel_Offset_A_2, 0).merge(Temp, (ColorFill(3) << 8), TempMask0[0][4 * i + 2]); 87 } 88 89 { 90 matrix_ref<ushort, 1, 16> Alpha = DataBuffer.format<ushort, 96, 16>().select<1, 1, 16, 1>(16 * i + 15, 0); 91 92 // R3/G3/B3/A3 93 Temp = (((DataBuffer.format<ushort, 96, 16>().select<1, 1, 16, 1>(16 * i + Channel_Offset_RV_3, 0) * (cm_add<ushort>(Alpha, 256, SAT)))) + (((ColorFill(0) << 8) * (cm_add<ushort>(0xFF00, -Alpha, SAT))))) >> 16; 94 DataBuffer.format<ushort, 96, 16>().select<1, 1, 16, 1>(16 * i + Channel_Offset_RV_3, 0).merge(Temp, (ColorFill(0) << 8), TempMask0[0][4 * i + 3]); 95 Temp = (((DataBuffer.format<ushort, 96, 16>().select<1, 1, 16, 1>(16 * i + Channel_Offset_GY_3, 0) * (cm_add<ushort>(Alpha, 256, SAT)))) + (((ColorFill(1) << 8) * (cm_add<ushort>(0xFF00, -Alpha, SAT))))) >> 16; 96 DataBuffer.format<ushort, 96, 16>().select<1, 1, 16, 1>(16 * i + Channel_Offset_GY_3, 0).merge(Temp, (ColorFill(1) << 8), TempMask0[0][4 * i + 3]); 97 Temp = (((DataBuffer.format<ushort, 96, 16>().select<1, 1, 16, 1>(16 * i + Channel_Offset_BU_3, 0) * (cm_add<ushort>(Alpha, 256, SAT)))) + (((ColorFill(2) << 8) * (cm_add<ushort>(0xFF00, -Alpha, SAT))))) >> 16; 98 DataBuffer.format<ushort, 96, 16>().select<1, 1, 16, 1>(16 * i + Channel_Offset_BU_3, 0).merge(Temp, (ColorFill(2) << 8), TempMask0[0][4 * i + 3]); 99 Temp = (((DataBuffer.format<ushort, 96, 16>().select<1, 1, 16, 1>(16 * i + Channel_Offset_A_3, 0) * (cm_add<ushort>(Alpha, 256, SAT)))) + (((ColorFill(3) << 8) * (cm_add<ushort>(0xFF00, -Alpha, SAT))))) >> 16; 100 DataBuffer.format<ushort, 96, 16>().select<1, 1, 16, 1>(16 * i + Channel_Offset_A_3, 0).merge(Temp, (ColorFill(3) << 8), TempMask0[0][4 * i + 3]); 101 } 102 } 103 TempMask0 = 0xFFFFFFFF; 104 }