Subversion Repositories Kolibri OS

Compare Revisions

Regard whitespace Rev 6145 → Rev 6146

/contrib/sdk/sources/vaapi/intel-driver-1.6.2/src/shaders/h264/ildb/AVC_ILDB_Filter_Mbaff_Y_h.asm
0,0 → 1,264
/*
* 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 ILDB filter horizontal Mbaff Y ///////////////////////////////////////////////////////
//
// This filter code prepares the src data and control data for ILDB filtering on all horizontal edges of Y.
//
// It sssumes the data for horizontal de-blocking is already transposed.
//
// Luma:
//
// +-------+-------+-------+-------+ H0 Edge
// | | | | |
// | | | | |
// | | | | |
// +-------+-------+-------+-------+ H1 Edge
// | | | | |
// | | | | |
// | | | | |
// +-------+-------+-------+-------+ H2 Edge
// | | | | |
// | | | | |
// | | | | |
// +-------+-------+-------+-------+ H3 Edge
// | | | | |
// | | | | |
// | | | | |
// +-------+-------+-------+-------+
//
/////////////////////////////////////////////////////////////////////////////
 
#if defined(_DEBUG)
mov (1) EntrySignatureC:w 0xBBBB:w
#endif
 
//========== Luma deblocking ==========
 
 
//---------- Deblock Y external top edge (H0) ----------
 
// Bypass deblocking if it is the top edge of the picture.
and.z.f0.0 (1) null:w r[ECM_AddrReg, BitFlags]:ub FilterTopMbEdgeFlag:w // Check for FilterTopMbEdgeFlag
mov (1) f0.1:w DualFieldMode:w // Check for dual field mode
// Non dual field mode
 
// Get (alpha >> 2) + 2
shr (16) Mbaff_ALPHA2(0,0)<1> r[ECM_AddrReg, bAlphaTop0_Y]<0;1,0>:ub 2:w // alpha >> 2
 
mov (2) MaskA<1>:uw r[ECM_AddrReg, wEdgeCntlMapA_ExtTopHorz0]<2;2,1>:uw
 
// Ext Y
mov (16) Mbaff_ALPHA(0,0)<1> r[ECM_AddrReg, bAlphaTop0_Y]<0;1,0>:ub
mov (16) Mbaff_BETA(0,0)<1> r[ECM_AddrReg, bBetaTop0_Y]<0;1,0>:ub
mov (16) Mbaff_TC0(0,0)<1> r[ECM_AddrReg, bTc0_h00_0_Y]<1;4,0>:ub
 
add (16) Mbaff_ALPHA2(0,0)<1> Mbaff_ALPHA2(0,0)<16;16,1> 2:w // alpha2 = (alpha >> 2) + 2
 
(f0.0) jmpi H0_Y_DONE // Skip Ext Y deblocking
(f0.1) jmpi DUAL_FIELD_Y
mov (1) P_AddrReg:w PREV_MB_Y_BASE:w { NoDDClr }
mov (1) Q_AddrReg:w SRC_MB_Y_BASE:w { NoDDChk }
CALL(FILTER_Y_MBAFF, 1) // Non dual field deblocking
jmpi H0_Y_DONE
 
DUAL_FIELD_Y:
// Dual field mode, FieldModeCurrentMbFlag=0 && FieldModeAboveMbFlag=1
 
mov (1) P_AddrReg:w ABOVE_CUR_MB_BASE:w { NoDDClr }
mov (1) Q_AddrReg:w ABOVE_CUR_MB_BASE+64:w { NoDDChk }
 
// Must use PREV_MB_YW. TOP_MB_YW is not big enough.
// Get top field rows
mov (16) ABOVE_CUR_MB_YW(0)<1> PREV_MB_YW(0, 0)<16;8,1> // Copy p3, p2
mov (16) ABOVE_CUR_MB_YW(1)<1> PREV_MB_YW(2, 0)<16;8,1> // Copy p1, p0
mov (16) ABOVE_CUR_MB_YW(2)<1> SRC_YW(0, 0)<16;8,1> // Copy q0, q1
mov (16) ABOVE_CUR_MB_YW(3)<1> SRC_YW(2, 0)<16;8,1> // Copy q2, q3
 
CALL(FILTER_Y_MBAFF, 1) // Ext Y, top field
 
// Save deblocked top field rows
mov (8) PREV_MB_YW(1, 0)<1> ABOVE_CUR_MB_YW(0, 8) // Copy p2
mov (8) PREV_MB_YW(2, 0)<1> ABOVE_CUR_MB_YW(1, 0) // Copy p1
mov (8) PREV_MB_YW(3, 0)<1> ABOVE_CUR_MB_YW(1, 8) // Copy p0
mov (8) SRC_YW(0, 0)<1> ABOVE_CUR_MB_YW(2, 0) // Copy q0
mov (8) SRC_YW(1, 0)<1> ABOVE_CUR_MB_YW(2, 8) // Copy q1
mov (8) SRC_YW(2, 0)<1> ABOVE_CUR_MB_YW(3, 0) // Copy q2
 
//==================================================================================
// Bottom field
// Get (alpha >> 2) + 2
shr (16) Mbaff_ALPHA2(0,0)<1> r[ECM_AddrReg, bAlphaTop1_Y]<0;1,0>:ub 2:w // alpha >> 2
 
mov (1) P_AddrReg:w ABOVE_CUR_MB_BASE:w { NoDDClr }
mov (1) Q_AddrReg:w ABOVE_CUR_MB_BASE+64:w { NoDDChk }
// Get bot field rows
mov (16) ABOVE_CUR_MB_YW(0)<1> PREV_MB_YW(0, 8)<16;8,1> // Copy p3, p2
mov (16) ABOVE_CUR_MB_YW(1)<1> PREV_MB_YW(2, 8)<16;8,1> // Copy p1, p0
mov (16) ABOVE_CUR_MB_YW(2)<1> SRC_YW(0, 8)<16;8,1> // Copy q0, q1
mov (16) ABOVE_CUR_MB_YW(3)<1> SRC_YW(2, 8)<16;8,1> // Copy q2, q3
 
mov (2) MaskA<1>:uw r[ECM_AddrReg, wEdgeCntlMapA_ExtTopHorz1]<2;2,1>:uw
 
mov (16) Mbaff_ALPHA(0,0)<1> r[ECM_AddrReg, bAlphaTop1_Y]<0;1,0>:ub
mov (16) Mbaff_BETA(0,0)<1> r[ECM_AddrReg, bBetaTop1_Y]<0;1,0>:ub
mov (16) Mbaff_TC0(0,0)<1> r[ECM_AddrReg, bTc0_h00_1_Y]<1;4,0>:ub
 
add (16) Mbaff_ALPHA2(0,0)<1> Mbaff_ALPHA2(0,0)<16;16,1> 2:w // alpha2 = (alpha >> 2) + 2
 
CALL(FILTER_Y_MBAFF, 1) // Ext Y, bot field
 
// Save deblocked top field rows
mov (8) PREV_MB_YW(1, 8)<1> ABOVE_CUR_MB_YW(0, 8) // Copy p2
mov (8) PREV_MB_YW(2, 8)<1> ABOVE_CUR_MB_YW(1, 0) // Copy p1
mov (8) PREV_MB_YW(3, 8)<1> ABOVE_CUR_MB_YW(1, 8) // Copy p0
mov (8) SRC_YW(0, 8)<1> ABOVE_CUR_MB_YW(2, 0) // Copy q0
mov (8) SRC_YW(1, 8)<1> ABOVE_CUR_MB_YW(2, 8) // Copy q1
mov (8) SRC_YW(2, 8)<1> ABOVE_CUR_MB_YW(3, 0) // Copy q2
//==================================================================================
 
H0_Y_DONE:
 
//BYPASS_H0_Y:
//------------------------------------------------------------------
// Same alpha, alpha2, beta and MaskB for all internal edges
 
// Get (alpha >> 2) + 2
shr (16) Mbaff_ALPHA2(0,0)<1> r[ECM_AddrReg, bAlphaInternal_Y]<0;1,0>:ub 2:w // alpha >> 2
 
// alpha = bAlphaInternal_Y
// beta = bBetaInternal_Y
mov (16) Mbaff_ALPHA(0,0)<1> r[ECM_AddrReg, bAlphaInternal_Y]<0;1,0>:ub
mov (16) Mbaff_BETA(0,0)<1> r[ECM_AddrReg, bBetaInternal_Y]<0;1,0>:ub
 
mov (1) MaskB:uw 0:w // Set MaskB = 0 for all 3 edges, so it always uses bS < 4 algorithm.
 
add (16) Mbaff_ALPHA2(0,0)<1> Mbaff_ALPHA2(0,0)<16;16,1> 2:w // alpha2 = (alpha >> 2) + 2
 
//---------- Deblock Y internal top edge (H1) ----------
 
// Bypass deblocking if FilterInternal4x4EdgesFlag = 0
and.z.f0.0 (1) null:w r[ECM_AddrReg, BitFlags]:ub FilterInternal4x4EdgesFlag:w // Check for FilterInternal4x4EdgesFlag
// (f0.0) jmpi BYPASS_H1_Y
 
// p3 = Cur MB Y row 0 = r[P_AddrReg, 0]<16;16,1>
// p2 = Cur MB Y row 1 = r[P_AddrReg, 16]<16;16,1>
// p1 = Cur MB Y row 2 = r[P_AddrReg, 32]<16;16,1>
// p0 = Cur MB Y row 3 = r[P_AddrReg, 48]<16;16,1>
// q0 = Cur MB Y row 4 = r[Q_AddrReg, 0]<16;16,1>
// q1 = Cur MB Y row 5 = r[Q_AddrReg, 16]<16;16,1>
// q2 = Cur MB Y row 6 = r[Q_AddrReg, 32]<16;16,1>
// q3 = Cur MB Y row 7 = r[Q_AddrReg, 48]<16;16,1>
mov (1) P_AddrReg:w SRC_MB_Y_BASE:w { NoDDClr }
mov (1) Q_AddrReg:w 4*Y_ROW_WIDTH+SRC_MB_Y_BASE:w { NoDDChk }
 
mov (1) MaskA:uw r[ECM_AddrReg, wEdgeCntlMap_IntTopHorz]:uw
 
// tc0 has bTc0_h13_Y + bTc0_h12_Y + bTc0_h11_Y + bTc0_h10_Y
mov (16) Mbaff_TC0(0,0)<1> r[ECM_AddrReg, bTc0_h10_Y]<1;4,0>:ub
 
// CALL(FILTER_Y_MBAFF, 1)
PRED_CALL(-f0.0, FILTER_Y_MBAFF, 1)
 
//BYPASS_H1_Y:
//------------------------------------------------------------------
 
 
//---------- Deblock Y internal mid horizontal edge (H2) ----------
 
// Bypass deblocking if FilterInternal8x8EdgesFlag = 0
and.z.f0.0 (1) null:w r[ECM_AddrReg, BitFlags]:ub FilterInternal8x8EdgesFlag:w // Check for FilterInternal4x4EdgesFlag
// (f0.0) jmpi BYPASS_H2_Y
 
// p3 = Cur MB Y row 4 = r[P_AddrReg, 0]<16;16,1>
// p2 = Cur MB Y row 5 = r[P_AddrReg, 16]<16;16,1>
// p1 = Cur MB Y row 6 = r[P_AddrReg, 32]<16;16,1>
// p0 = Cur MB Y row 7 = r[P_AddrReg, 48]<16;16,1>
// q0 = Cur MB Y row 8 = r[Q_AddrReg, 0]<16;16,1>
// q1 = Cur MB Y row 9 = r[Q_AddrReg, 16]<16;16,1>
// q2 = Cur MB Y row 10 = r[Q_AddrReg, 32]<16;16,1>
// q3 = Cur MB Y row 11 = r[Q_AddrReg, 48]<16;16,1>
mov (1) P_AddrReg:w 4*Y_ROW_WIDTH+SRC_MB_Y_BASE:w { NoDDClr }
mov (1) Q_AddrReg:w 8*Y_ROW_WIDTH+SRC_MB_Y_BASE:w { NoDDChk }
 
mov (1) MaskA:uw r[ECM_AddrReg, wEdgeCntlMap_IntMidHorz]:uw
 
// tc0 has bTc0_h23_Y + bTc0_h22_Y + bTc0_h21_Y + bTc0_h20_Y
mov (16) Mbaff_TC0(0,0)<1> r[ECM_AddrReg, bTc0_h20_Y]<1;4,0>:ub
 
// CALL(FILTER_Y_MBAFF, 1)
PRED_CALL(-f0.0, FILTER_Y_MBAFF, 1)
 
//BYPASS_H2_Y:
//-----------------------------------------------
 
 
//---------- Deblock Y internal bottom edge (H3) ----------
 
// Bypass deblocking if FilterInternal4x4EdgesFlag = 0
and.z.f0.0 (1) null:w r[ECM_AddrReg, BitFlags]:ub FilterInternal4x4EdgesFlag:w // Check for FilterInternal4x4EdgesFlag
// (f0.0) jmpi BYPASS_H3_Y
 
// p3 = Cur MB Y row 8 = r[P_AddrReg, 0]<16;16,1>
// p2 = Cur MB Y row 9 = r[P_AddrReg, 16]<16;16,1>
// p1 = Cur MB Y row 10 = r[P_AddrReg, 32]<16;16,1>
// p0 = Cur MB Y row 11 = r[P_AddrReg, 48]<16;16,1>
// q0 = Cur MB Y row 12 = r[Q_AddrReg, 0]<16;16,1>
// q1 = Cur MB Y row 13 = r[Q_AddrReg, 16]<16;16,1>
// q2 = Cur MB Y row 14 = r[Q_AddrReg, 32]<16;16,1>
// q3 = Cur MB Y row 15 = r[Q_AddrReg, 48]<16;16,1>
mov (1) P_AddrReg:w 8*Y_ROW_WIDTH+SRC_MB_Y_BASE:w { NoDDClr }
mov (1) Q_AddrReg:w 12*Y_ROW_WIDTH+SRC_MB_Y_BASE:w { NoDDChk }
mov (1) MaskA:uw r[ECM_AddrReg, wEdgeCntlMap_IntBotHorz]:uw
 
// tc0 has bTc0_h33_Y + bTc0_h32_Y + bTc0_h31_Y + bTc0_h30_Y
mov (16) Mbaff_TC0(0,0)<1> r[ECM_AddrReg, bTc0_h30_Y]<1;4,0>:ub
 
// CALL(FILTER_Y_MBAFF, 1)
PRED_CALL(-f0.0, FILTER_Y_MBAFF, 1)
 
//BYPASS_H3_Y:
//-----------------------------------------------