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 /* 24 Buffer layout after shuffle 25 _________________________________________________ 26 |_______Block0__________|_______Block1__________| 27 |_______Block2__________|_______Block3__________| 28 |_______Block4__________|_______Block5__________| 29 |_______Block6__________|_______Block7__________| 30 31 Write back buffer layout correlate to the block number#, each box stands for 1 GRF 32 _______________________________________________ 33 |____R0_________R1_____|____R2_________R3_____| 34 |____G0_________G1_____|____G2_________G3_____| 35 |____B0_________B1_____|____B2_________B3_____| 36 |____A0_________A1_____|____A2_________A3_____| 37 |____R4_________R5_____|____R6_________R7_____| 38 |____G4_________G5_____|____G6_________G7_____| 39 |____B4_________B5_____|____B6_________B7_____| 40 |____A4_________A5_____|____A6_________A7_____| 41 */ 42 43 matrix_ref<uint, 8, 8> Result = DataBuffer.format<uint, 96, 8>().select<8, 1, 8, 1>(64, 0); 44 45 SurfaceIndex Dst_Surface(MDF_FC_OUTPUT_BTI_START); 46 47 matrix_ref<uint, 4, 8> TempResult4x8_Top = Result.select<4, 1, 8, 1>(0, 0); 48 matrix_ref<uint, 4, 8> TempResult4x8_Bottom = Result.select<4, 1, 8, 1>(4, 0); 49 50 #pragma unroll 51 for (uchar i = 0; i < 2; i++, DstY += 8) 52 { 53 // First 8x16 54 { 55 // R/G/B/A channels 56 TempResult4x8_Top.format<ushort, 4, 16>().select<1, 1, 8, 1>(0, 0) = (DataBuffer.format<ushort, 96, 16>().select<1, 1, 8, 1>(4 + 32 * i, 0) >> 11) + ((DataBuffer.format<ushort, 96, 16>().select<1, 1, 8, 1>(2 + 32 * i, 0) >> 10) << 5) + (DataBuffer.format<ushort, 96, 16>().select<1, 1, 8, 1>(32 * i, 0) & 0xF800); 57 TempResult4x8_Top.format<ushort, 4, 16>().select<1, 1, 8, 1>(0, 8) = (DataBuffer.format<ushort, 96, 16>().select<1, 1, 8, 1>(12+ 32 * i, 0) >> 11) + ((DataBuffer.format<ushort, 96, 16>().select<1, 1, 8, 1>(10 + 32 * i, 0) >> 10) << 5) + (DataBuffer.format<ushort, 96, 16>().select<1, 1, 8, 1>(8 + 32 * i, 0) & 0xF800); 58 TempResult4x8_Top.format<ushort, 4, 16>().select<1, 1, 8, 1>(1, 0) = (DataBuffer.format<ushort, 96, 16>().select<1, 1, 8, 1>(4 + 32 * i, 8) >> 11) + ((DataBuffer.format<ushort, 96, 16>().select<1, 1, 8, 1>(2 + 32 * i, 8) >> 10) << 5) + (DataBuffer.format<ushort, 96, 16>().select<1, 1, 8, 1>(32 * i, 8) & 0xF800); 59 TempResult4x8_Top.format<ushort, 4, 16>().select<1, 1, 8, 1>(1, 8) = (DataBuffer.format<ushort, 96, 16>().select<1, 1, 8, 1>(12 + 32 * i, 8) >> 11) + ((DataBuffer.format<ushort, 96, 16>().select<1, 1, 8, 1>(10 + 32 * i, 8) >> 10) << 5) + (DataBuffer.format<ushort, 96, 16>().select<1, 1, 8, 1>(8 + 32 * i, 8) & 0xF800); 60 61 TempResult4x8_Top.format<ushort, 4, 16>().select<1, 1, 8, 1>(2, 0) = (DataBuffer.format<ushort, 96, 16>().select<1, 1, 8, 1>(5 + 32 * i, 0) >> 11) + ((DataBuffer.format<ushort, 96, 16>().select<1, 1, 8, 1>(3 + 32 * i, 0) >> 10) << 5) + (DataBuffer.format<ushort, 96, 16>().select<1, 1, 8, 1>(1 + 32 * i, 0) & 0xF800); 62 TempResult4x8_Top.format<ushort, 4, 16>().select<1, 1, 8, 1>(2, 8) = (DataBuffer.format<ushort, 96, 16>().select<1, 1, 8, 1>(13 + 32 * i, 0) >> 11) + ((DataBuffer.format<ushort, 96, 16>().select<1, 1, 8, 1>(11 + 32 * i, 0) >> 10) << 5) + (DataBuffer.format<ushort, 96, 16>().select<1, 1, 8, 1>(9 + 32 * i, 0) & 0xF800); 63 TempResult4x8_Top.format<ushort, 4, 16>().select<1, 1, 8, 1>(3, 0) = (DataBuffer.format<ushort, 96, 16>().select<1, 1, 8, 1>(5 + 32 * i, 8) >> 11) + ((DataBuffer.format<ushort, 96, 16>().select<1, 1, 8, 1>(3 + 32 * i, 8) >> 10) << 5) + (DataBuffer.format<ushort, 96, 16>().select<1, 1, 8, 1>(1 + 32 * i, 8) & 0xF800); 64 TempResult4x8_Top.format<ushort, 4, 16>().select<1, 1, 8, 1>(3, 8) = (DataBuffer.format<ushort, 96, 16>().select<1, 1, 8, 1>(13 + 32 * i, 8) >> 11) + ((DataBuffer.format<ushort, 96, 16>().select<1, 1, 8, 1>(11 + 32 * i, 8) >> 10) << 5) + (DataBuffer.format<ushort, 96, 16>().select<1, 1, 8, 1>(9 + 32 * i, 8) & 0xF800); 65 66 67 TempResult4x8_Bottom.format<ushort, 4, 16>().select<1, 1, 8, 1>(0, 0) = (DataBuffer.format<ushort, 96, 16>().select<1, 1, 8, 1>(4 + 16 + 32 * i, 0) >> 11) + ((DataBuffer.format<ushort, 96, 16>().select<1, 1, 8, 1>(2 + 16 + 32 * i, 0) >> 10) << 5) + (DataBuffer.format<ushort, 96, 16>().select<1, 1, 8, 1>(16 + 32 * i, 0) & 0xF800); 68 TempResult4x8_Bottom.format<ushort, 4, 16>().select<1, 1, 8, 1>(0, 8) = (DataBuffer.format<ushort, 96, 16>().select<1, 1, 8, 1>(12 + 16 + 32 * i, 0) >> 11) + ((DataBuffer.format<ushort, 96, 16>().select<1, 1, 8, 1>(10 + 16 + 32 * i, 0) >> 10) << 5) + (DataBuffer.format<ushort, 96, 16>().select<1, 1, 8, 1>(8 + 16 + 32 * i, 0) & 0xF800); 69 TempResult4x8_Bottom.format<ushort, 4, 16>().select<1, 1, 8, 1>(1, 0) = (DataBuffer.format<ushort, 96, 16>().select<1, 1, 8, 1>(4 + 16 + 32 * i, 8) >> 11) + ((DataBuffer.format<ushort, 96, 16>().select<1, 1, 8, 1>(2 + 16 + 32 * i, 8) >> 10) << 5) + (DataBuffer.format<ushort, 96, 16>().select<1, 1, 8, 1>(16 + 32 * i, 8) & 0xF800); 70 TempResult4x8_Bottom.format<ushort, 4, 16>().select<1, 1, 8, 1>(1, 8) = (DataBuffer.format<ushort, 96, 16>().select<1, 1, 8, 1>(12 + 16 + 32 * i, 8) >> 11) + ((DataBuffer.format<ushort, 96, 16>().select<1, 1, 8, 1>(10 + 16 + 32 * i, 8) >> 10) << 5) + (DataBuffer.format<ushort, 96, 16>().select<1, 1, 8, 1>(8 + 16 + 32 * i, 8) & 0xF800); 71 72 TempResult4x8_Bottom.format<ushort, 4, 16>().select<1, 1, 8, 1>(2, 0) = (DataBuffer.format<ushort, 96, 16>().select<1, 1, 8, 1>(5 + 16 + 32 * i, 0) >> 11) + ((DataBuffer.format<ushort, 96, 16>().select<1, 1, 8, 1>(3 + 16 + 32 * i, 0) >> 10) << 5) + (DataBuffer.format<ushort, 96, 16>().select<1, 1, 8, 1>(1 + 16 + 32 * i, 0) & 0xF800); 73 TempResult4x8_Bottom.format<ushort, 4, 16>().select<1, 1, 8, 1>(2, 8) = (DataBuffer.format<ushort, 96, 16>().select<1, 1, 8, 1>(13 + 16 + 32 * i, 0) >> 11) + ((DataBuffer.format<ushort, 96, 16>().select<1, 1, 8, 1>(11 + 16 + 32 * i, 0) >> 10) << 5) + (DataBuffer.format<ushort, 96, 16>().select<1, 1, 8, 1>(9 + 16 + 32 * i, 0) & 0xF800); 74 TempResult4x8_Bottom.format<ushort, 4, 16>().select<1, 1, 8, 1>(3, 0) = (DataBuffer.format<ushort, 96, 16>().select<1, 1, 8, 1>(5 + 16 + 32 * i, 8) >> 11) + ((DataBuffer.format<ushort, 96, 16>().select<1, 1, 8, 1>(3 + 16 + 32 * i, 8) >> 10) << 5) + (DataBuffer.format<ushort, 96, 16>().select<1, 1, 8, 1>(1 + 16 + 32 * i, 8) & 0xF800); 75 TempResult4x8_Bottom.format<ushort, 4, 16>().select<1, 1, 8, 1>(3, 8) = (DataBuffer.format<ushort, 96, 16>().select<1, 1, 8, 1>(13 + 16 + 32 * i, 8) >> 11) + ((DataBuffer.format<ushort, 96, 16>().select<1, 1, 8, 1>(11 + 16 + 32 * i, 8) >> 10) << 5) + (DataBuffer.format<ushort, 96, 16>().select<1, 1, 8, 1>(9 + 16 + 32 * i, 8) & 0xF800); 76 77 write(Dst_Surface, DstX * 2, DstY, Result); 78 } 79 } 80 }