/* * Copyright © <2010>, Intel Corporation. * * Permission is hereby granted, free of charge, to any person obtaining a * copy of this software and associated documentation files (the * "Software"), to deal in the Software without restriction, including * without limitation the rights to use, copy, modify, merge, publish, * distribute, sub license, and/or sell copies of the Software, and to * permit persons to whom the Software is furnished to do so, subject to * the following conditions: * * The above copyright notice and this permission notice (including the * next paragraph) shall be included in all copies or substantial portions * of the Software. * * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. * IN NO EVENT SHALL PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. * * This file was originally licensed under the following license * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. * */ ////////////////////////////////////////////////////////////////////////////////////////////////////////////// // AVC Child Kernel (Vertical and horizontal de-block a 4:2:0 MB Y comp) // // First, de-block vertical edges from left to right. // Second, de-block horizontal edge from top to bottom. // // If transform_size_8x8_flag = 1, luma is de-blocked at 8x8. Otherwise, luma is de-blocked at 4x4. // ////////////////////////////////////////////////////////////////////////////////////////////////////////////// #define AVC_ILDB .kernel AVC_ILDB_CHILD_Y #if defined(COMBINED_KERNEL) ILDB_LABEL(AVC_ILDB_CHILD_Y): #endif #include "SetupVPKernel.asm" #include "AVC_ILDB.inc" #if defined(_DEBUG) mov (1) EntrySignatureC:w 0x9998:w #endif // Init local variables shl (8) ORIX_CUR<1>:w ORIX<0;2,1>:w 4:w // Expand addr to bytes, repeat (x,y) 4 times // Init addr register for vertical control data mov (1) ECM_AddrReg<1>:w CNTRL_DATA_BASE:w // Init edge control map AddrReg //=== Null Kernel =============================================================== // jmpi ILDB_LABEL(POST_ILDB_Y) //=============================================================================== mul (1) URBOffsetC:uw ORIY:uw 4:w #if !defined(DEV_CL) //==================================================================================== // For BearLake-C, 64 bytes are stored in memory and dataport expands to 256 bytes. Need to use a special read command on BL-C. // MB_offset = MBsCntX * CurRow + CurCol // MBCntrlDataOffsetY = globel_byte_offset = MB_offset * 64 mul (1) CntrlDataOffsetY:ud MBsCntX:w ORIY:w add (1) CntrlDataOffsetY:ud CntrlDataOffsetY:ud ORIX:w // Assign to MSGSRC.2:ud for memory access // mul (1) CntrlDataOffsetY:ud CntrlDataOffsetY:ud 64:uw mul (1) MSGSRC.2:ud CntrlDataOffsetY:ud 64:uw #endif // Load current MB control data #if defined(DEV_CL) #if defined(_APPLE) #include "Load_ILDB_Cntrl_Data_22DW.asm" // Crestline for Apple, progressive only #else #include "Load_ILDB_Cntrl_Data_64DW.asm" // Crestline #endif #else #include "Load_ILDB_Cntrl_Data_16DW.asm" // Cantiga and beyond #endif // Check loaded control data #if defined(_APPLE) and.z.f0.1 (8) null<1>:uw r[ECM_AddrReg, wEdgeCntlMap_IntLeftVert]<8;8,1>:uw 0xFFFF:uw // Skip ILDB? (f0.1) and.z.f0.1 (2) null<1>:uw r[ECM_AddrReg, wEdgeCntlMapA_ExtTopHorz0]<2;2,1>:uw 0xFFFF:uw // Skip ILDB? #else and.z.f0.1 (16) null<1>:uw r[ECM_AddrReg, wEdgeCntlMap_IntLeftVert]<16;16,1>:uw 0xFFFF:uw // Skip ILDB? #endif and.nz.f0.0 (1) null:w r[ECM_AddrReg, ExtBitFlags]:ub DISABLE_ILDB_FLAG:w // Skip ILDB? // Use free cycles here add (1) ORIX_LEFT:w ORIX_LEFT:w -4:w // add (1) ORIY_TOP:w ORIY_TOP:w -4:w mov (1) GateWayOffsetC:uw ORIY:uw // Use row # as Gateway offset #if defined(_APPLE) (f0.1.all8h) jmpi ILDB_LABEL(READ_FOR_URB_Y) // Skip ILDB #else (f0.1.all16h) jmpi ILDB_LABEL(READ_FOR_URB_Y) // Skip ILDB #endif (f0.0) jmpi ILDB_LABEL(READ_FOR_URB_Y) // Skip ILDB add (1) ORIY_TOP:w ORIY_TOP:w -4:w // Bettr performance is observed if boundary MBs are not checked and skipped. #include "load_Cur_Y_16x16T.asm" // Load cur MB Y, 16x16, transpose // #include "load_Left_Y_4x16T.asm" // Load left MB (4x16) Y data from memory #include "load_Top_Y_16x4.asm" // Load top MB (16x4) Y data from memory #include "Transpose_Cur_Y_16x16.asm" // #include "Transpose_Left_Y_4x16.asm" //---------- Perform vertical ILDB filting on Y --------- #include "AVC_ILDB_Filter_Y_v.asm" //------------------------------------------------------- #include "save_Left_Y_16x4T.asm" // Write left MB (4x16) Y data to memory #include "Transpose_Cur_Y_16x16.asm" // Transpose a MB for horizontal edge de-blocking //---------- Perform horizontal ILDB filting on Y ------- #include "AVC_ILDB_Filter_Y_h.asm" //------------------------------------------------------- #include "save_Cur_Y_16x16.asm" // Write cur MB (16x16) #include "save_Top_Y_16x4.asm" // Write top MB (16x4) //---------- Write right most 4 columns of cur MB to URB ---------- // Transpose the right most 4 cols 4x16 in GRF to 16x4 in LEFT_TEMP_B. It is 4 left most cols in cur MB. #include "Transpose_Cur_Y_4x16.asm" ILDB_LABEL(WRITE_URB_Y): // Note: LEFT_TEMP_B(2) = TOP_TEMP_B(0), TOP_TEMP_B must be avail mov (16) m1<1>:ud LEFT_TEMP_D(2)<8;8,1> // Copy 2 GRFs to 2 URB entries (Y) #include "writeURB_Y_Child.asm" //----------------------------------------------------------------- //=========== Check write commit of the last write ============ mov (8) WritebackResponse(0)<1> WritebackResponse(0) ILDB_LABEL(POST_ILDB_Y): // Send notification thru Gateway to root thread, update luma Status[CurRow] #include "AVC_ILDB_ForwardMsg.asm" #if !defined(GW_DCN) // For non-ILK chipsets //child send EOT : Request type = 1 END_CHILD_THREAD #endif // !defined(DEV_ILK) // The thread finishs here //------------------------------------------------------------------------------ ILDB_LABEL(READ_FOR_URB_Y): // Still need to prepare URB data for the right neighbor MB #include "load_Cur_Y_Right_Most_4x16.asm" // Load cur MB ( right most 4x16) Y data from memory #include "Transpose_Cur_Y_Right_Most_4x16.asm" // jmpi ILDB_LABEL(WRITE_URB_Y) // Note: LEFT_TEMP_B(2) = TOP_TEMP_B(0), TOP_TEMP_B must be avail mov (16) m1<1>:ud LEFT_TEMP_D(2)<8;8,1> // Copy 2 GRFs to 2 URB entries (Y) #include "writeURB_Y_Child.asm" //----------------------------------------------------------------- // Send notification thru Gateway to root thread, update luma Status[CurRow] #include "AVC_ILDB_ForwardMsg.asm" #if !defined(GW_DCN) // For non-ILK chipsets //child send EOT : Request type = 1 END_CHILD_THREAD #endif // !defined(DEV_ILK) // The thread finishs here //------------------------------------------------------------------------------ //////////////////////////////////////////////////////////////////////////////// // Include other subrutines being called #include "AVC_ILDB_Luma_Core.asm" // #include "AVC_ILDB_Chroma_Core.asm" #if !defined(COMBINED_KERNEL) // For standalone kernel only .end_code .end_kernel #endif