/******************************************************************************
*
* Copyright (C) 2018 The Android Open Source Project
*
* 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.
*
*****************************************************************************
* Originally developed and contributed by Ittiam Systems Pvt. Ltd, Bangalore
*/
/**
*******************************************************************************
* @file
* ihevc_resi_trans.c
*
* @brief
* Contains function definitions for residual and forward transform
*
* @author
* 100470
*
* @par List of Functions:
* - ihevc_resi_trans_4x4_ttype1()
* - ihevc_resi_trans_4x4()
* - ihevc_resi_trans_8x8()
* - ihevc_resi_trans_16x16()
* - ihevc_resi_trans_32x32()
*
* @remarks
* None
*
*******************************************************************************
*/
#include <stdio.h>
#include <string.h>
#include <stdlib.h>
#include "ihevc_typedefs.h"
#include "ihevc_macros.h"
#include "ihevc_platform_macros.h"
#include "ihevc_defs.h"
#include "ihevc_trans_tables.h"
#include "ihevc_resi_trans.h"
#include "ihevc_func_selector.h"
#include "ihevc_trans_macros.h"
/**
*******************************************************************************
*
* @brief
* This function performs residue calculation and forward transform type 1
* on input pixels
*
* @par Description:
* Performs residue calculation by subtracting source and prediction and
* followed by forward transform
*
* @param[in] pu1_src
* Input 4x4 pixels
*
* @param[in] pu1_pred
* Prediction data
*
* @param[in] pi2_tmp
* Temporary buffer of size 4x4
*
* @param[out] pi2_dst
* Output 4x4 coefficients
*
* @param[in] src_strd
* Input stride
*
* @param[in] pred_strd
* Prediction Stride
*
* @param[in] dst_strd_chr_flag
* Output Stride and Chroma Flag packed in the MS and LS 16-bit
*
*
* @returns Void
*
* @remarks
* None
*
*******************************************************************************
*/
UWORD32 ihevc_resi_trans_4x4_ttype1(UWORD8 *pu1_src,
UWORD8 *pu1_pred,
WORD32 *pi4_temp,
WORD16 *pi2_dst,
WORD32 src_strd,
WORD32 pred_strd,
WORD32 dst_strd_chr_flag)
{
WORD32 i, c[4];
WORD32 add, shift;
WORD32 trans_size;
WORD32 *pi4_tmp_orig;
WORD16 *pi2_dst_orig;
UWORD32 u4_blk_sad = 0;
// WORD32 chroma_flag;
WORD32 dst_strd;
// chroma_flag = dst_strd_chr_flag & 1;
dst_strd = dst_strd_chr_flag >> 16;
pi2_dst_orig = pi2_dst;
pi4_tmp_orig = pi4_temp;
trans_size = TRANS_SIZE_4;
/* Residue + Forward Transform 1st stage */
shift = 1; // log2(iWidth) - 1 + g_uiBitIncrement
add = 1 << (shift - 1);
for(i = 0; i < trans_size; i++)
{
WORD32 resi_tmp_1, resi_tmp_2, resi_tmp_3;
// Intermediate Variables
resi_tmp_1 = pu1_src[0] - pu1_pred[0];
resi_tmp_2 = pu1_src[3] - pu1_pred[3];
c[0] = resi_tmp_1 + resi_tmp_2;
u4_blk_sad += abs(resi_tmp_1) + abs(resi_tmp_2);
resi_tmp_1 = pu1_src[1] - pu1_pred[1];
resi_tmp_2 = pu1_src[3] - pu1_pred[3];
c[1] = resi_tmp_1 + resi_tmp_2;
u4_blk_sad += abs(resi_tmp_1);
resi_tmp_1 = pu1_src[0] - pu1_pred[0];
resi_tmp_2 = pu1_src[1] - pu1_pred[1];
c[2] = resi_tmp_1 - resi_tmp_2;
resi_tmp_1 = pu1_src[2] - pu1_pred[2];
c[3] = 74 * resi_tmp_1;
u4_blk_sad += abs(resi_tmp_1);
pi4_temp[0] = (29 * c[0] + 55 * c[1] + c[3] + add) >> shift;
resi_tmp_1 = pu1_src[0] - pu1_pred[0];
resi_tmp_2 = pu1_src[1] - pu1_pred[1];
resi_tmp_3 = pu1_src[3] - pu1_pred[3];
pi4_temp[trans_size] =
(74 * (resi_tmp_1 + resi_tmp_2 - resi_tmp_3) + add)
>> shift;
pi4_temp[2 * trans_size] = (29 * c[2] + 55 * c[0] - c[3] + add) >> shift;
pi4_temp[3 * trans_size] = (55 * c[2] - 29 * c[1] + c[3] + add) >> shift;
pu1_src += src_strd;
pu1_pred += pred_strd;
pi4_temp++;
}
pi4_temp = pi4_tmp_orig;
/* Forward transform 2nd stage */
shift = 8; // log2(iHeight) + 6
add = 1 << (shift - 1);
for(i = 0; i < TRANS_SIZE_4; i++)
{
// Intermediate Variables
c[0] = pi4_temp[0] + pi4_temp[3];
c[1] = pi4_temp[1] + pi4_temp[3];
c[2] = pi4_temp[0] - pi4_temp[1];
c[3] = 74 * pi4_temp[2];
pi2_dst[0] = (29 * c[0] + 55 * c[1] + c[3] + add) >> shift;
pi2_dst[dst_strd] = (74 * (pi4_temp[0] + pi4_temp[1] - pi4_temp[3]) + add)
>> shift;
pi2_dst[2 * dst_strd] = (29 * c[2] + 55 * c[0] - c[3] + add) >> shift;
pi2_dst[3 * dst_strd] = (55 * c[2] - 29 * c[1] + c[3] + add) >> shift;
pi4_temp += trans_size;
pi2_dst++;
}
return u4_blk_sad;
}
/**
*******************************************************************************
*
* @brief
* This function performs residue calculation and forward transform on
* input pixels
*
* @par Description:
* Performs residue calculation by subtracting source and prediction and
* followed by forward transform
*
* @param[in] pu1_src
* Input 4x4 pixels
*
* @param[in] pu1_pred
* Prediction data
*
* @param[in] pi2_tmp
* Temporary buffer of size 4x4
*
* @param[out] pi2_dst
* Output 4x4 coefficients
*
* @param[in] src_strd
* Input stride
*
* @param[in] pred_strd
* Prediction Stride
*
* @param[in] dst_strd_chr_flag
* Output Stride and Chroma Flag packed in the MS and LS 16-bit
*
* @returns Void
*
* @remarks
* None
*
*******************************************************************************
*/
UWORD32 ihevc_resi_trans_4x4(UWORD8 *pu1_src,
UWORD8 *pu1_pred,
WORD32 *pi4_temp,
WORD16 *pi2_dst,
WORD32 src_strd,
WORD32 pred_strd,
WORD32 dst_strd_chr_flag)
{
WORD32 i;
WORD32 e[2], o[2];
WORD32 add, shift;
WORD32 trans_size;
WORD32 *pi4_tmp_orig;
WORD16 *pi2_dst_orig;
UWORD32 u4_blk_sad=0;
WORD32 chroma_flag;
WORD32 dst_strd;
chroma_flag = dst_strd_chr_flag & 1;
dst_strd = dst_strd_chr_flag >> 16;
pi2_dst_orig = pi2_dst;
pi4_tmp_orig = pi4_temp;
trans_size = TRANS_SIZE_4;
/* Residue + Forward Transform 1st stage */
shift = 1; // log2(iWidth) - 1 + g_uiBitIncrement
add = 1 << (shift - 1);
for(i = 0; i < trans_size; i++)
{
WORD32 resi_tmp_1, resi_tmp_2;
/* e and o */
resi_tmp_1 = pu1_src[0 + 0*chroma_flag] - pu1_pred[0 + 0*chroma_flag];
resi_tmp_2 = pu1_src[3 + 3*chroma_flag] - pu1_pred[3 + 3*chroma_flag];
e[0] = resi_tmp_1 + resi_tmp_2;
o[0] = resi_tmp_1 - resi_tmp_2;
u4_blk_sad += abs(resi_tmp_1);
u4_blk_sad += abs(resi_tmp_2);
resi_tmp_1 = pu1_src[1 + 1*chroma_flag] - pu1_pred[1 + 1*chroma_flag];
resi_tmp_2 = pu1_src[2 + 2*chroma_flag] - pu1_pred[2 + 2*chroma_flag];
e[1] = resi_tmp_1 + resi_tmp_2;
o[1] = resi_tmp_1 - resi_tmp_2;
u4_blk_sad += abs(resi_tmp_1);
u4_blk_sad += abs(resi_tmp_2);
pi4_temp[0] = (g_ai2_ihevc_trans_4[0][0] * e[0]
+ g_ai2_ihevc_trans_4[0][1] * e[1]);// + add) >> shift;
pi4_temp[2 * trans_size] = (g_ai2_ihevc_trans_4[2][0] * e[0]
+ g_ai2_ihevc_trans_4[2][1] * e[1]);// + add) >> shift;
pi4_temp[trans_size] = (g_ai2_ihevc_trans_4[1][0] * o[0]
+ g_ai2_ihevc_trans_4[1][1] * o[1]);// + add) >> shift;
pi4_temp[3 * trans_size] = (g_ai2_ihevc_trans_4[3][0] * o[0]
+ g_ai2_ihevc_trans_4[3][1] * o[1]);// + add) >> shift;
pu1_src += src_strd;
pu1_pred += pred_strd;
pi4_temp++;
}
pi4_temp = pi4_tmp_orig;
/* Forward Transform 2nd stage */
shift = 9; // log2(iHeight) + 6
add = 1 << (shift - 1);
for(i = 0; i < trans_size; i++)
{
/* e and o */
e[0] = pi4_temp[0] + pi4_temp[3];
o[0] = pi4_temp[0] - pi4_temp[3];
e[1] = pi4_temp[1] + pi4_temp[2];
o[1] = pi4_temp[1] - pi4_temp[2];
pi2_dst[0] = (g_ai2_ihevc_trans_4[0][0] * e[0]
+ g_ai2_ihevc_trans_4[0][1] * e[1] + add) >> shift;
pi2_dst[2 * dst_strd] = (g_ai2_ihevc_trans_4[2][0] * e[0]
+ g_ai2_ihevc_trans_4[2][1] * e[1] + add) >> shift;
pi2_dst[dst_strd] = (g_ai2_ihevc_trans_4[1][0] * o[0]
+ g_ai2_ihevc_trans_4[1][1] * o[1] + add) >> shift;
pi2_dst[3 * dst_strd] = (g_ai2_ihevc_trans_4[3][0] * o[0]
+ g_ai2_ihevc_trans_4[3][1] * o[1] + add) >> shift;
pi4_temp += trans_size;
pi2_dst++;
}
return u4_blk_sad;
}
void ihevc_resi_trans_4x4_16bit(WORD16 *pi2_src,
UWORD8 *pu1_pred,
WORD16 *pi2_tmp,
WORD16 *pi2_dst,
WORD32 src_strd,
WORD32 pred_strd,
WORD32 dst_strd)
{
WORD32 i;
WORD32 e[2], o[2];
WORD32 add, shift;
WORD32 trans_size;
WORD16 *pi2_tmp_orig;
WORD16 *pi2_dst_orig;
pi2_dst_orig = pi2_dst;
pi2_tmp_orig = pi2_tmp;
trans_size = TRANS_SIZE_4;
/* Residue + Forward Transform 1st stage */
shift = 1; // log2(iWidth) - 1 + g_uiBitIncrement
add = 1 << (shift - 1);
for(i = 0; i < trans_size; i++)
{
WORD32 resi_tmp_1, resi_tmp_2;
/* e and o */
resi_tmp_1 = pi2_src[0] - pu1_pred[0];
resi_tmp_2 = pi2_src[3] - pu1_pred[3];
e[0] = resi_tmp_1 + resi_tmp_2;
o[0] = resi_tmp_1 - resi_tmp_2;
resi_tmp_1 = pi2_src[1] - pu1_pred[1];
resi_tmp_2 = pi2_src[2] - pu1_pred[2];
e[1] = resi_tmp_1 + resi_tmp_2;
o[1] = resi_tmp_1 - resi_tmp_2;
pi2_tmp[0] = (g_ai2_ihevc_trans_4[0][0] * e[0]
+ g_ai2_ihevc_trans_4[0][1] * e[1] + add) >> shift;
pi2_tmp[2 * trans_size] = (g_ai2_ihevc_trans_4[2][0] * e[0]
+ g_ai2_ihevc_trans_4[2][1] * e[1] + add) >> shift;
pi2_tmp[trans_size] = (g_ai2_ihevc_trans_4[1][0] * o[0]
+ g_ai2_ihevc_trans_4[1][1] * o[1] + add) >> shift;
pi2_tmp[3 * trans_size] = (g_ai2_ihevc_trans_4[3][0] * o[0]
+ g_ai2_ihevc_trans_4[3][1] * o[1] + add) >> shift;
pi2_src += src_strd;
pu1_pred += pred_strd;
pi2_tmp++;
}
pi2_tmp = pi2_tmp_orig;
/* Forward Transform 2nd stage */
shift = 8; // log2(iHeight) + 6
add = 1 << (shift - 1);
for(i = 0; i < trans_size; i++)
{
/* e and o */
e[0] = pi2_tmp[0] + pi2_tmp[3];
o[0] = pi2_tmp[0] - pi2_tmp[3];
e[1] = pi2_tmp[1] + pi2_tmp[2];
o[1] = pi2_tmp[1] - pi2_tmp[2];
pi2_dst[0] = (g_ai2_ihevc_trans_4[0][0] * e[0]
+ g_ai2_ihevc_trans_4[0][1] * e[1] + add) >> shift;
pi2_dst[2 * dst_strd] = (g_ai2_ihevc_trans_4[2][0] * e[0]
+ g_ai2_ihevc_trans_4[2][1] * e[1] + add) >> shift;
pi2_dst[dst_strd] = (g_ai2_ihevc_trans_4[1][0] * o[0]
+ g_ai2_ihevc_trans_4[1][1] * o[1] + add) >> shift;
pi2_dst[3 * dst_strd] = (g_ai2_ihevc_trans_4[3][0] * o[0]
+ g_ai2_ihevc_trans_4[3][1] * o[1] + add) >> shift;
pi2_tmp += trans_size;
pi2_dst++;
}
}
/**
*******************************************************************************
*
* @brief
* This function performs residue calculation and forward transform on
* input pixels
*
* @par Description:
* Performs residue calculation by subtracting source and prediction and
* followed by forward transform
*
* @param[in] pu1_src
* Input 8x8 pixels
*
* @param[in] pu1_pred
* Prediction data
*
* @param[in] pi2_tmp
* Temporary buffer of size 8x8
*
* @param[out] pi2_dst
* Output 8x8 coefficients
*
* @param[in] src_strd
* Input stride
*
* @param[in] pred_strd
* Prediction Stride
*
* @param[in] dst_strd_chr_flag
* Output Stride and Chroma Flag packed in the MS and LS 16-bit
*
* @returns Void
*
* @remarks
* None
*
*******************************************************************************
*/
UWORD32 ihevc_resi_trans_8x8(UWORD8 *pu1_src,
UWORD8 *pu1_pred,
WORD32 *pi4_temp,
WORD16 *pi2_dst,
WORD32 src_strd,
WORD32 pred_strd,
WORD32 dst_strd_chr_flag)
{
WORD32 i, k;
WORD32 e[4], o[4];
WORD32 ee[2], eo[2];
WORD32 add, shift;
WORD32 trans_size;
WORD32 *pi4_tmp_orig;
// WORD16 *pi2_tmp;
WORD16 *pi2_dst_orig;
UWORD32 u4_blk_sad=0;
WORD32 chroma_flag;
WORD32 dst_strd;
chroma_flag = dst_strd_chr_flag & 1;
dst_strd = dst_strd_chr_flag >> 16;
pi2_dst_orig = pi2_dst;
pi4_tmp_orig = pi4_temp;
trans_size = TRANS_SIZE_8;
/* Residue + Forward Transform 1st stage */
shift = 2; // log2(iWidth) - 1 + g_uiBitIncrement
add = 1 << (shift - 1);
for(i = 0; i < trans_size; i++)
{
WORD32 resi_tmp_1, resi_tmp_2;
/* Utilizing symmetry properties to the maximum to minimize the number of multiplications */
/* e and o*/
for(k = 0; k < 4; k++)
{
resi_tmp_1 = pu1_src[k*(1+chroma_flag)] - pu1_pred[k*(1+chroma_flag)];
resi_tmp_2 = pu1_src[(7-k)*(1+chroma_flag)] - pu1_pred[(7-k)*(1+chroma_flag)];
e[k] = resi_tmp_1 + resi_tmp_2;
o[k] = resi_tmp_1 - resi_tmp_2;
u4_blk_sad += abs(resi_tmp_1) + abs(resi_tmp_2);
}
/* ee and eo */
ee[0] = e[0] + e[3];
eo[0] = e[0] - e[3];
ee[1] = e[1] + e[2];
eo[1] = e[1] - e[2];
pi4_temp[0] = (g_ai2_ihevc_trans_8[0][0] * ee[0]
+ g_ai2_ihevc_trans_8[0][1] * ee[1]);// + add) >> shift;
pi4_temp[4 * trans_size] = (g_ai2_ihevc_trans_8[4][0] * ee[0]
+ g_ai2_ihevc_trans_8[4][1] * ee[1]);// + add) >> shift;
pi4_temp[2 * trans_size] = (g_ai2_ihevc_trans_8[2][0] * eo[0]
+ g_ai2_ihevc_trans_8[2][1] * eo[1]);// + add) >> shift;
pi4_temp[6 * trans_size] = (g_ai2_ihevc_trans_8[6][0] * eo[0]
+ g_ai2_ihevc_trans_8[6][1] * eo[1]);// + add) >> shift;
pi4_temp[trans_size] = (g_ai2_ihevc_trans_8[1][0] * o[0]
+ g_ai2_ihevc_trans_8[1][1] * o[1]
+ g_ai2_ihevc_trans_8[1][2] * o[2]
+ g_ai2_ihevc_trans_8[1][3] * o[3]);// + add) >> shift;
pi4_temp[3 * trans_size] = (g_ai2_ihevc_trans_8[3][0] * o[0]
+ g_ai2_ihevc_trans_8[3][1] * o[1]
+ g_ai2_ihevc_trans_8[3][2] * o[2]
+ g_ai2_ihevc_trans_8[3][3] * o[3]);// + add) >> shift;
pi4_temp[5 * trans_size] = (g_ai2_ihevc_trans_8[5][0] * o[0]
+ g_ai2_ihevc_trans_8[5][1] * o[1]
+ g_ai2_ihevc_trans_8[5][2] * o[2]
+ g_ai2_ihevc_trans_8[5][3] * o[3]);// + add) >> shift;
pi4_temp[7 * trans_size] = (g_ai2_ihevc_trans_8[7][0] * o[0]
+ g_ai2_ihevc_trans_8[7][1] * o[1]
+ g_ai2_ihevc_trans_8[7][2] * o[2]
+ g_ai2_ihevc_trans_8[7][3] * o[3]);// + add) >> shift;
pu1_src += src_strd;
pu1_pred += pred_strd;
pi4_temp++;
}
pi4_temp = pi4_tmp_orig;
/* Forward Transform 2nd stage */
shift = 11; // log2(iHeight) + 6
add = 1 << (shift - 1);
for(i = 0; i < trans_size; i++)
{
/* Utilizing symmetry properties to the maximum to minimize the number of multiplications */
/* e and o*/
for(k = 0; k < 4; k++)
{
e[k] = pi4_temp[k] + pi4_temp[7 - k];
o[k] = pi4_temp[k] - pi4_temp[7 - k];
}
/* ee and eo */
ee[0] = e[0] + e[3];
eo[0] = e[0] - e[3];
ee[1] = e[1] + e[2];
eo[1] = e[1] - e[2];
pi2_dst[0] = (g_ai2_ihevc_trans_8[0][0] * ee[0]
+ g_ai2_ihevc_trans_8[0][1] * ee[1] + add) >> shift;
pi2_dst[4 * dst_strd] = (g_ai2_ihevc_trans_8[4][0] * ee[0]
+ g_ai2_ihevc_trans_8[4][1] * ee[1] + add) >> shift;
pi2_dst[2 * dst_strd] = (g_ai2_ihevc_trans_8[2][0] * eo[0]
+ g_ai2_ihevc_trans_8[2][1] * eo[1] + add) >> shift;
pi2_dst[6 * dst_strd] = (g_ai2_ihevc_trans_8[6][0] * eo[0]
+ g_ai2_ihevc_trans_8[6][1] * eo[1] + add) >> shift;
pi2_dst[dst_strd] = (g_ai2_ihevc_trans_8[1][0] * o[0]
+ g_ai2_ihevc_trans_8[1][1] * o[1]
+ g_ai2_ihevc_trans_8[1][2] * o[2]
+ g_ai2_ihevc_trans_8[1][3] * o[3] + add) >> shift;
pi2_dst[3 * dst_strd] = (g_ai2_ihevc_trans_8[3][0] * o[0]
+ g_ai2_ihevc_trans_8[3][1] * o[1]
+ g_ai2_ihevc_trans_8[3][2] * o[2]
+ g_ai2_ihevc_trans_8[3][3] * o[3] + add) >> shift;
pi2_dst[5 * dst_strd] = (g_ai2_ihevc_trans_8[5][0] * o[0]
+ g_ai2_ihevc_trans_8[5][1] * o[1]
+ g_ai2_ihevc_trans_8[5][2] * o[2]
+ g_ai2_ihevc_trans_8[5][3] * o[3] + add) >> shift;
pi2_dst[7 * dst_strd] = (g_ai2_ihevc_trans_8[7][0] * o[0]
+ g_ai2_ihevc_trans_8[7][1] * o[1]
+ g_ai2_ihevc_trans_8[7][2] * o[2]
+ g_ai2_ihevc_trans_8[7][3] * o[3] + add) >> shift;
pi4_temp += trans_size;
pi2_dst++;
}
return u4_blk_sad;
}
void ihevc_resi_trans_8x8_16bit(WORD16 *pi2_src,
UWORD8 *pu1_pred,
WORD16 *pi2_tmp,
WORD16 *pi2_dst,
WORD32 src_strd,
WORD32 pred_strd,
WORD32 dst_strd)
{
WORD32 i, k;
WORD32 e[4], o[4];
WORD32 ee[2], eo[2];
WORD32 add, shift;
WORD32 trans_size;
WORD16 *pi2_tmp_orig;
WORD16 *pi2_dst_orig;
pi2_dst_orig = pi2_dst;
pi2_tmp_orig = pi2_tmp;
trans_size = TRANS_SIZE_8;
/* Residue + Forward Transform 1st stage */
shift = 2; // log2(iWidth) - 1 + g_uiBitIncrement
add = 1 << (shift - 1);
for(i = 0; i < trans_size; i++)
{
WORD32 resi_tmp_1, resi_tmp_2;
/* Utilizing symmetry properties to the maximum to minimize the number of multiplications */
/* e and o*/
for(k = 0; k < 4; k++)
{
resi_tmp_1 = pi2_src[k] - pu1_pred[k];
resi_tmp_2 = pi2_src[7 - k] - pu1_pred[7 - k];
e[k] = resi_tmp_1 + resi_tmp_2;
o[k] = resi_tmp_1 - resi_tmp_2;
}
/* ee and eo */
ee[0] = e[0] + e[3];
eo[0] = e[0] - e[3];
ee[1] = e[1] + e[2];
eo[1] = e[1] - e[2];
pi2_tmp[0] = (g_ai2_ihevc_trans_8[0][0] * ee[0]
+ g_ai2_ihevc_trans_8[0][1] * ee[1] + add) >> shift;
pi2_tmp[4 * trans_size] = (g_ai2_ihevc_trans_8[4][0] * ee[0]
+ g_ai2_ihevc_trans_8[4][1] * ee[1] + add) >> shift;
pi2_tmp[2 * trans_size] = (g_ai2_ihevc_trans_8[2][0] * eo[0]
+ g_ai2_ihevc_trans_8[2][1] * eo[1] + add) >> shift;
pi2_tmp[6 * trans_size] = (g_ai2_ihevc_trans_8[6][0] * eo[0]
+ g_ai2_ihevc_trans_8[6][1] * eo[1] + add) >> shift;
pi2_tmp[trans_size] = (g_ai2_ihevc_trans_8[1][0] * o[0]
+ g_ai2_ihevc_trans_8[1][1] * o[1]
+ g_ai2_ihevc_trans_8[1][2] * o[2]
+ g_ai2_ihevc_trans_8[1][3] * o[3] + add) >> shift;
pi2_tmp[3 * trans_size] = (g_ai2_ihevc_trans_8[3][0] * o[0]
+ g_ai2_ihevc_trans_8[3][1] * o[1]
+ g_ai2_ihevc_trans_8[3][2] * o[2]
+ g_ai2_ihevc_trans_8[3][3] * o[3] + add) >> shift;
pi2_tmp[5 * trans_size] = (g_ai2_ihevc_trans_8[5][0] * o[0]
+ g_ai2_ihevc_trans_8[5][1] * o[1]
+ g_ai2_ihevc_trans_8[5][2] * o[2]
+ g_ai2_ihevc_trans_8[5][3] * o[3] + add) >> shift;
pi2_tmp[7 * trans_size] = (g_ai2_ihevc_trans_8[7][0] * o[0]
+ g_ai2_ihevc_trans_8[7][1] * o[1]
+ g_ai2_ihevc_trans_8[7][2] * o[2]
+ g_ai2_ihevc_trans_8[7][3] * o[3] + add) >> shift;
pi2_src += src_strd;
pu1_pred += pred_strd;
pi2_tmp++;
}
pi2_tmp = pi2_tmp_orig;
/* Forward Transform 2nd stage */
shift = 9; // log2(iHeight) + 6
add = 1 << (shift - 1);
for(i = 0; i < trans_size; i++)
{
/* Utilizing symmetry properties to the maximum to minimize the number of multiplications */
/* e and o*/
for(k = 0; k < 4; k++)
{
e[k] = pi2_tmp[k] + pi2_tmp[7 - k];
o[k] = pi2_tmp[k] - pi2_tmp[7 - k];
}
/* ee and eo */
ee[0] = e[0] + e[3];
eo[0] = e[0] - e[3];
ee[1] = e[1] + e[2];
eo[1] = e[1] - e[2];
pi2_dst[0] = (g_ai2_ihevc_trans_8[0][0] * ee[0]
+ g_ai2_ihevc_trans_8[0][1] * ee[1] + add) >> shift;
pi2_dst[4 * dst_strd] = (g_ai2_ihevc_trans_8[4][0] * ee[0]
+ g_ai2_ihevc_trans_8[4][1] * ee[1] + add) >> shift;
pi2_dst[2 * dst_strd] = (g_ai2_ihevc_trans_8[2][0] * eo[0]
+ g_ai2_ihevc_trans_8[2][1] * eo[1] + add) >> shift;
pi2_dst[6 * dst_strd] = (g_ai2_ihevc_trans_8[6][0] * eo[0]
+ g_ai2_ihevc_trans_8[6][1] * eo[1] + add) >> shift;
pi2_dst[dst_strd] = (g_ai2_ihevc_trans_8[1][0] * o[0]
+ g_ai2_ihevc_trans_8[1][1] * o[1]
+ g_ai2_ihevc_trans_8[1][2] * o[2]
+ g_ai2_ihevc_trans_8[1][3] * o[3] + add) >> shift;
pi2_dst[3 * dst_strd] = (g_ai2_ihevc_trans_8[3][0] * o[0]
+ g_ai2_ihevc_trans_8[3][1] * o[1]
+ g_ai2_ihevc_trans_8[3][2] * o[2]
+ g_ai2_ihevc_trans_8[3][3] * o[3] + add) >> shift;
pi2_dst[5 * dst_strd] = (g_ai2_ihevc_trans_8[5][0] * o[0]
+ g_ai2_ihevc_trans_8[5][1] * o[1]
+ g_ai2_ihevc_trans_8[5][2] * o[2]
+ g_ai2_ihevc_trans_8[5][3] * o[3] + add) >> shift;
pi2_dst[7 * dst_strd] = (g_ai2_ihevc_trans_8[7][0] * o[0]
+ g_ai2_ihevc_trans_8[7][1] * o[1]
+ g_ai2_ihevc_trans_8[7][2] * o[2]
+ g_ai2_ihevc_trans_8[7][3] * o[3] + add) >> shift;
pi2_tmp += trans_size;
pi2_dst++;
}
}
/**
*******************************************************************************
*
* @brief
* This function performs residue calculation and forward transform on
* input pixels
*
* @par Description:
* Performs residue calculation by subtracting source and prediction and
* followed by forward transform
*
* @param[in] pu1_src
* Input 16x16 pixels
*
* @param[in] pu1_pred
* Prediction data
*
* @param[in] pi2_tmp
* Temporary buffer of size 16x16
*
* @param[out] pi2_dst
* Output 16x16 coefficients
*
* @param[in] src_strd
* Input stride
*
* @param[in] pred_strd
* Prediction Stride
*
* @param[in] dst_strd_chr_flag
* Output Stride and Chroma Flag packed in the MS and LS 16-bit
*
* @returns Void
*
* @remarks
* None
*
*******************************************************************************
*/
UWORD32 ihevc_resi_trans_16x16(UWORD8 *pu1_src,
UWORD8 *pu1_pred,
WORD32 *pi4_temp,
WORD16 *pi2_dst,
WORD32 src_strd,
WORD32 pred_strd,
WORD32 dst_strd_chr_flag)
{
WORD32 i, k;
WORD32 e[8], o[8];
WORD32 ee[4], eo[4];
WORD32 eee[2], eeo[2];
WORD32 add, shift;
WORD32 trans_size;
WORD32 *pi4_tmp_orig;
WORD16 *pi2_dst_orig;
UWORD32 u4_blk_sad = 0;
WORD32 chroma_flag;
WORD32 dst_strd;
chroma_flag = dst_strd_chr_flag & 1;
dst_strd = dst_strd_chr_flag >> 16;
pi2_dst_orig = pi2_dst;
pi4_tmp_orig = pi4_temp;
trans_size = TRANS_SIZE_16;
/* Residue + Forward Transform 1st stage */
shift = 3; // log2(iWidth) - 1 + g_uiBitIncrement
add = 1 << (shift - 1);
for(i = 0; i < trans_size; i++)
{
WORD32 resi_tmp_1, resi_tmp_2;
/* e and o*/
for(k = 0; k < 8; k++)
{
resi_tmp_1 = pu1_src[k*(1+chroma_flag)] - pu1_pred[k*(1+chroma_flag)];
resi_tmp_2 = pu1_src[(15-k)*(1+chroma_flag)] - pu1_pred[(15-k)*(1+chroma_flag)];
e[k] = resi_tmp_1 + resi_tmp_2;
o[k] = resi_tmp_1 - resi_tmp_2;
u4_blk_sad += abs(resi_tmp_1) + abs(resi_tmp_2);
}
/* ee and eo */
for(k = 0; k < 4; k++)
{
ee[k] = e[k] + e[7 - k];
eo[k] = e[k] - e[7 - k];
}
/* eee and eeo */
eee[0] = ee[0] + ee[3];
eeo[0] = ee[0] - ee[3];
eee[1] = ee[1] + ee[2];
eeo[1] = ee[1] - ee[2];
pi4_temp[0] = (g_ai2_ihevc_trans_16[0][0] * eee[0]
+ g_ai2_ihevc_trans_16[0][1] * eee[1]);// + add) >> shift;
pi4_temp[8 * trans_size] = (g_ai2_ihevc_trans_16[8][0] * eee[0]
+ g_ai2_ihevc_trans_16[8][1] * eee[1]);// + add) >> shift;
pi4_temp[4 * trans_size] = (g_ai2_ihevc_trans_16[4][0] * eeo[0]
+ g_ai2_ihevc_trans_16[4][1] * eeo[1]);// + add) >> shift;
pi4_temp[12 * trans_size] = (g_ai2_ihevc_trans_16[12][0] * eeo[0]
+ g_ai2_ihevc_trans_16[12][1] * eeo[1]);// + add) >> shift;
for(k = 2; k < 16; k += 4)
{
pi4_temp[k * trans_size] = (g_ai2_ihevc_trans_16[k][0] * eo[0]
+ g_ai2_ihevc_trans_16[k][1] * eo[1]
+ g_ai2_ihevc_trans_16[k][2] * eo[2]
+ g_ai2_ihevc_trans_16[k][3] * eo[3]);// + add)>> shift;
}
for(k = 1; k < 16; k += 2)
{
pi4_temp[k * trans_size] = (g_ai2_ihevc_trans_16[k][0] * o[0]
+ g_ai2_ihevc_trans_16[k][1] * o[1]
+ g_ai2_ihevc_trans_16[k][2] * o[2]
+ g_ai2_ihevc_trans_16[k][3] * o[3]
+ g_ai2_ihevc_trans_16[k][4] * o[4]
+ g_ai2_ihevc_trans_16[k][5] * o[5]
+ g_ai2_ihevc_trans_16[k][6] * o[6]
+ g_ai2_ihevc_trans_16[k][7] * o[7]);// + add) >> shift;
}
pu1_src += src_strd;
pu1_pred += pred_strd;
pi4_temp++;
}
pi4_temp = pi4_tmp_orig;
/* Forward Transform 2nd stage */
shift = 13; // log2(iHeight) + 6
add = 1 << (shift - 1);
for(i = 0; i < TRANS_SIZE_16; i++)
{
/* e and o*/
for(k = 0; k < 8; k++)
{
e[k] = pi4_temp[k] + pi4_temp[15 - k];
o[k] = pi4_temp[k] - pi4_temp[15 - k];
}
/* ee and eo */
for(k = 0; k < 4; k++)
{
ee[k] = e[k] + e[7 - k];
eo[k] = e[k] - e[7 - k];
}
/* eee and eeo */
eee[0] = ee[0] + ee[3];
eeo[0] = ee[0] - ee[3];
eee[1] = ee[1] + ee[2];
eeo[1] = ee[1] - ee[2];
pi2_dst[0] = (g_ai2_ihevc_trans_16[0][0] * eee[0]
+ g_ai2_ihevc_trans_16[0][1] * eee[1] + add) >> shift;
pi2_dst[8 * dst_strd] = (g_ai2_ihevc_trans_16[8][0] * eee[0]
+ g_ai2_ihevc_trans_16[8][1] * eee[1] + add) >> shift;
pi2_dst[4 * dst_strd] = (g_ai2_ihevc_trans_16[4][0] * eeo[0]
+ g_ai2_ihevc_trans_16[4][1] * eeo[1] + add) >> shift;
pi2_dst[12 * dst_strd] = (g_ai2_ihevc_trans_16[12][0] * eeo[0]
+ g_ai2_ihevc_trans_16[12][1] * eeo[1] + add) >> shift;
for(k = 2; k < 16; k += 4)
{
pi2_dst[k * dst_strd] = (g_ai2_ihevc_trans_16[k][0] * eo[0]
+ g_ai2_ihevc_trans_16[k][1] * eo[1]
+ g_ai2_ihevc_trans_16[k][2] * eo[2]
+ g_ai2_ihevc_trans_16[k][3] * eo[3] + add)
>> shift;
}
for(k = 1; k < 16; k += 2)
{
pi2_dst[k * dst_strd] = (g_ai2_ihevc_trans_16[k][0] * o[0]
+ g_ai2_ihevc_trans_16[k][1] * o[1]
+ g_ai2_ihevc_trans_16[k][2] * o[2]
+ g_ai2_ihevc_trans_16[k][3] * o[3]
+ g_ai2_ihevc_trans_16[k][4] * o[4]
+ g_ai2_ihevc_trans_16[k][5] * o[5]
+ g_ai2_ihevc_trans_16[k][6] * o[6]
+ g_ai2_ihevc_trans_16[k][7] * o[7] + add) >> shift;
}
pi4_temp += trans_size;
pi2_dst++;
}
return u4_blk_sad;
}
void ihevc_resi_trans_16x16_16bit(WORD16 *pi2_src,
UWORD8 *pu1_pred,
WORD16 *pi2_tmp,
WORD16 *pi2_dst,
WORD32 src_strd,
WORD32 pred_strd,
WORD32 dst_strd)
{
WORD32 i, k;
WORD32 e[8], o[8];
WORD32 ee[4], eo[4];
WORD32 eee[2], eeo[2];
WORD32 add, shift;
WORD32 trans_size;
WORD16 *pi2_tmp_orig;
WORD16 *pi2_dst_orig;
pi2_dst_orig = pi2_dst;
pi2_tmp_orig = pi2_tmp;
trans_size = TRANS_SIZE_16;
/* Residue + Forward Transform 1st stage */
shift = 3; // log2(iWidth) - 1 + g_uiBitIncrement
add = 1 << (shift - 1);
for(i = 0; i < trans_size; i++)
{
WORD32 resi_tmp_1, resi_tmp_2;
/* e and o*/
for(k = 0; k < 8; k++)
{
resi_tmp_1 = pi2_src[k] - pu1_pred[k];
resi_tmp_2 = pi2_src[15 - k] - pu1_pred[15 - k];
e[k] = resi_tmp_1 + resi_tmp_2;
o[k] = resi_tmp_1 - resi_tmp_2;
}
/* ee and eo */
for(k = 0; k < 4; k++)
{
ee[k] = e[k] + e[7 - k];
eo[k] = e[k] - e[7 - k];
}
/* eee and eeo */
eee[0] = ee[0] + ee[3];
eeo[0] = ee[0] - ee[3];
eee[1] = ee[1] + ee[2];
eeo[1] = ee[1] - ee[2];
pi2_tmp[0] = (g_ai2_ihevc_trans_16[0][0] * eee[0]
+ g_ai2_ihevc_trans_16[0][1] * eee[1] + add) >> shift;
pi2_tmp[8 * trans_size] = (g_ai2_ihevc_trans_16[8][0] * eee[0]
+ g_ai2_ihevc_trans_16[8][1] * eee[1] + add) >> shift;
pi2_tmp[4 * trans_size] = (g_ai2_ihevc_trans_16[4][0] * eeo[0]
+ g_ai2_ihevc_trans_16[4][1] * eeo[1] + add) >> shift;
pi2_tmp[12 * trans_size] = (g_ai2_ihevc_trans_16[12][0] * eeo[0]
+ g_ai2_ihevc_trans_16[12][1] * eeo[1] + add) >> shift;
for(k = 2; k < 16; k += 4)
{
pi2_tmp[k * trans_size] = (g_ai2_ihevc_trans_16[k][0] * eo[0]
+ g_ai2_ihevc_trans_16[k][1] * eo[1]
+ g_ai2_ihevc_trans_16[k][2] * eo[2]
+ g_ai2_ihevc_trans_16[k][3] * eo[3] + add)
>> shift;
}
for(k = 1; k < 16; k += 2)
{
pi2_tmp[k * trans_size] = (g_ai2_ihevc_trans_16[k][0] * o[0]
+ g_ai2_ihevc_trans_16[k][1] * o[1]
+ g_ai2_ihevc_trans_16[k][2] * o[2]
+ g_ai2_ihevc_trans_16[k][3] * o[3]
+ g_ai2_ihevc_trans_16[k][4] * o[4]
+ g_ai2_ihevc_trans_16[k][5] * o[5]
+ g_ai2_ihevc_trans_16[k][6] * o[6]
+ g_ai2_ihevc_trans_16[k][7] * o[7] + add) >> shift;
}
pi2_src += src_strd;
pu1_pred += pred_strd;
pi2_tmp++;
}
pi2_tmp = pi2_tmp_orig;
/* Forward Transform 2nd stage */
shift = 10; // log2(iHeight) + 6
add = 1 << (shift - 1);
for(i = 0; i < TRANS_SIZE_16; i++)
{
/* e and o*/
for(k = 0; k < 8; k++)
{
e[k] = pi2_tmp[k] + pi2_tmp[15 - k];
o[k] = pi2_tmp[k] - pi2_tmp[15 - k];
}
/* ee and eo */
for(k = 0; k < 4; k++)
{
ee[k] = e[k] + e[7 - k];
eo[k] = e[k] - e[7 - k];
}
/* eee and eeo */
eee[0] = ee[0] + ee[3];
eeo[0] = ee[0] - ee[3];
eee[1] = ee[1] + ee[2];
eeo[1] = ee[1] - ee[2];
pi2_dst[0] = (g_ai2_ihevc_trans_16[0][0] * eee[0]
+ g_ai2_ihevc_trans_16[0][1] * eee[1] + add) >> shift;
pi2_dst[8 * dst_strd] = (g_ai2_ihevc_trans_16[8][0] * eee[0]
+ g_ai2_ihevc_trans_16[8][1] * eee[1] + add) >> shift;
pi2_dst[4 * dst_strd] = (g_ai2_ihevc_trans_16[4][0] * eeo[0]
+ g_ai2_ihevc_trans_16[4][1] * eeo[1] + add) >> shift;
pi2_dst[12 * dst_strd] = (g_ai2_ihevc_trans_16[12][0] * eeo[0]
+ g_ai2_ihevc_trans_16[12][1] * eeo[1] + add) >> shift;
for(k = 2; k < 16; k += 4)
{
pi2_dst[k * dst_strd] = (g_ai2_ihevc_trans_16[k][0] * eo[0]
+ g_ai2_ihevc_trans_16[k][1] * eo[1]
+ g_ai2_ihevc_trans_16[k][2] * eo[2]
+ g_ai2_ihevc_trans_16[k][3] * eo[3] + add)
>> shift;
}
for(k = 1; k < 16; k += 2)
{
pi2_dst[k * dst_strd] = (g_ai2_ihevc_trans_16[k][0] * o[0]
+ g_ai2_ihevc_trans_16[k][1] * o[1]
+ g_ai2_ihevc_trans_16[k][2] * o[2]
+ g_ai2_ihevc_trans_16[k][3] * o[3]
+ g_ai2_ihevc_trans_16[k][4] * o[4]
+ g_ai2_ihevc_trans_16[k][5] * o[5]
+ g_ai2_ihevc_trans_16[k][6] * o[6]
+ g_ai2_ihevc_trans_16[k][7] * o[7] + add) >> shift;
}
pi2_tmp += trans_size;
pi2_dst++;
}
}
/**
*******************************************************************************
*
* @brief
* This function performs residue calculation and forward transform on
* input pixels
*
* @par Description:
* Performs residue calculation by subtracting source and prediction and
* followed by forward transform
*
* @param[in] pu1_src
* Input 32x32 pixels
*
* @param[in] pu1_pred
* Prediction data
*
* @param[in] pi2_tmp
* Temporary buffer of size 32x32
*
* @param[out] pi2_dst
* Output 32x32 coefficients
*
* @param[in] src_strd
* Input stride
*
* @param[in] pred_strd
* Prediction Stride
*
* @param[in] dst_strd_chr_flag
* Output Stride and Chroma Flag packed in the MS and LS 16-bit
*
* @returns Void
*
* @remarks
* None
*
*******************************************************************************
*/
UWORD32 ihevc_resi_trans_32x32(UWORD8 *pu1_src,
UWORD8 *pu1_pred,
WORD32 *pi4_temp,
WORD16 *pi2_dst,
WORD32 src_strd,
WORD32 pred_strd,
WORD32 dst_strd_chr_flag)
{
WORD32 i, k;
WORD32 e[16], o[16];
WORD32 ee[8], eo[8];
WORD32 eee[4], eeo[4];
WORD32 eeee[2], eeeo[2];
WORD32 add, shift;
WORD32 trans_size;
WORD32 *pi4_tmp_orig;
WORD16 *pi2_dst_orig;
UWORD32 u4_blk_sad = 0 ;
WORD32 chroma_flag;
WORD32 dst_strd;
chroma_flag = dst_strd_chr_flag & 1;
dst_strd = dst_strd_chr_flag >> 16;
pi2_dst_orig = pi2_dst;
pi4_tmp_orig = pi4_temp;
trans_size = TRANS_SIZE_32;
/* Residue + Forward Transform 1st stage */
/* Made to zero to match with intrinsics */
shift = 0; // 4 : log2(iWidth) - 1 + g_uiBitIncrement
add = 0 ; //1 << (shift - 1);
for(i = 0; i < trans_size; i++)
{
WORD32 resi_tmp_1, resi_tmp_2;
/* e and o*/
for(k = 0; k < 16; k++)
{
resi_tmp_1 = pu1_src[k] - pu1_pred[k];
resi_tmp_2 = pu1_src[31 - k] - pu1_pred[31 - k];
e[k] = resi_tmp_1 + resi_tmp_2;
o[k] = resi_tmp_1 - resi_tmp_2;
u4_blk_sad += abs(resi_tmp_1) + abs(resi_tmp_2);
}
/* ee and eo */
for(k = 0; k < 8; k++)
{
ee[k] = e[k] + e[15 - k];
eo[k] = e[k] - e[15 - k];
}
/* eee and eeo */
for(k = 0; k < 4; k++)
{
eee[k] = ee[k] + ee[7 - k];
eeo[k] = ee[k] - ee[7 - k];
}
/* eeee and eeeo */
eeee[0] = eee[0] + eee[3];
eeeo[0] = eee[0] - eee[3];
eeee[1] = eee[1] + eee[2];
eeeo[1] = eee[1] - eee[2];
pi4_temp[0] = (g_ai2_ihevc_trans_32[0][0] * eeee[0]
+ g_ai2_ihevc_trans_32[0][1] * eeee[1]);// + add) >> shift;
pi4_temp[16 * trans_size] = (g_ai2_ihevc_trans_32[16][0] * eeee[0]
+ g_ai2_ihevc_trans_32[16][1] * eeee[1]);// + add) >> shift;
pi4_temp[8 * trans_size] = (g_ai2_ihevc_trans_32[8][0] * eeeo[0]
+ g_ai2_ihevc_trans_32[8][1] * eeeo[1]);// + add) >> shift;
pi4_temp[24 * trans_size] = (g_ai2_ihevc_trans_32[24][0] * eeeo[0]
+ g_ai2_ihevc_trans_32[24][1] * eeeo[1]);// + add) >> shift;
for(k = 4; k < 32; k += 8)
{
pi4_temp[k * trans_size] = (g_ai2_ihevc_trans_32[k][0] * eeo[0]
+ g_ai2_ihevc_trans_32[k][1] * eeo[1]
+ g_ai2_ihevc_trans_32[k][2] * eeo[2]
+ g_ai2_ihevc_trans_32[k][3] * eeo[3]);// + add)>> shift;
}
for(k = 2; k < 32; k += 4)
{
pi4_temp[k * trans_size] = (g_ai2_ihevc_trans_32[k][0] * eo[0]
+ g_ai2_ihevc_trans_32[k][1] * eo[1]
+ g_ai2_ihevc_trans_32[k][2] * eo[2]
+ g_ai2_ihevc_trans_32[k][3] * eo[3]
+ g_ai2_ihevc_trans_32[k][4] * eo[4]
+ g_ai2_ihevc_trans_32[k][5] * eo[5]
+ g_ai2_ihevc_trans_32[k][6] * eo[6]
+ g_ai2_ihevc_trans_32[k][7] * eo[7]);// + add)>> shift;
}
for(k = 1; k < 32; k += 2)
{
pi4_temp[k * trans_size] = (g_ai2_ihevc_trans_32[k][0] * o[0]
+ g_ai2_ihevc_trans_32[k][1] * o[1]
+ g_ai2_ihevc_trans_32[k][2] * o[2]
+ g_ai2_ihevc_trans_32[k][3] * o[3]
+ g_ai2_ihevc_trans_32[k][4] * o[4]
+ g_ai2_ihevc_trans_32[k][5] * o[5]
+ g_ai2_ihevc_trans_32[k][6] * o[6]
+ g_ai2_ihevc_trans_32[k][7] * o[7]
+ g_ai2_ihevc_trans_32[k][8] * o[8]
+ g_ai2_ihevc_trans_32[k][9] * o[9]
+ g_ai2_ihevc_trans_32[k][10] * o[10]
+ g_ai2_ihevc_trans_32[k][11] * o[11]
+ g_ai2_ihevc_trans_32[k][12] * o[12]
+ g_ai2_ihevc_trans_32[k][13] * o[13]
+ g_ai2_ihevc_trans_32[k][14] * o[14]
+ g_ai2_ihevc_trans_32[k][15] * o[15]);// + add) >> shift;
}
pu1_src += src_strd;
pu1_pred += pred_strd;
pi4_temp++;
}
pi4_temp = pi4_tmp_orig;
/* Forward Transform 2nd stage */
shift = 15; // log2(iHeight) + 6
add = 1 << (shift - 1);
for(i = 0; i < TRANS_SIZE_32; i++)
{
/* e and o*/
for(k = 0; k < 16; k++)
{
e[k] = pi4_temp[k] + pi4_temp[31 - k];
o[k] = pi4_temp[k] - pi4_temp[31 - k];
}
/* ee and eo */
for(k = 0; k < 8; k++)
{
ee[k] = e[k] + e[15 - k];
eo[k] = e[k] - e[15 - k];
}
/* eee and eeo */
for(k = 0; k < 4; k++)
{
eee[k] = ee[k] + ee[7 - k];
eeo[k] = ee[k] - ee[7 - k];
}
/* eeee and eeeo */
eeee[0] = eee[0] + eee[3];
eeeo[0] = eee[0] - eee[3];
eeee[1] = eee[1] + eee[2];
eeeo[1] = eee[1] - eee[2];
pi2_dst[0] = (g_ai2_ihevc_trans_32[0][0] * eeee[0]
+ g_ai2_ihevc_trans_32[0][1] * eeee[1] + add) >> shift;
pi2_dst[16 * dst_strd] = (g_ai2_ihevc_trans_32[16][0] * eeee[0]
+ g_ai2_ihevc_trans_32[16][1] * eeee[1] + add) >> shift;
pi2_dst[8 * dst_strd] = (g_ai2_ihevc_trans_32[8][0] * eeeo[0]
+ g_ai2_ihevc_trans_32[8][1] * eeeo[1] + add) >> shift;
pi2_dst[24 * dst_strd] = (g_ai2_ihevc_trans_32[24][0] * eeeo[0]
+ g_ai2_ihevc_trans_32[24][1] * eeeo[1] + add) >> shift;
for(k = 4; k < 32; k += 8)
{
pi2_dst[k * dst_strd] = (g_ai2_ihevc_trans_32[k][0] * eeo[0]
+ g_ai2_ihevc_trans_32[k][1] * eeo[1]
+ g_ai2_ihevc_trans_32[k][2] * eeo[2]
+ g_ai2_ihevc_trans_32[k][3] * eeo[3] + add)
>> shift;
}
for(k = 2; k < 32; k += 4)
{
pi2_dst[k * dst_strd] = (g_ai2_ihevc_trans_32[k][0] * eo[0]
+ g_ai2_ihevc_trans_32[k][1] * eo[1]
+ g_ai2_ihevc_trans_32[k][2] * eo[2]
+ g_ai2_ihevc_trans_32[k][3] * eo[3]
+ g_ai2_ihevc_trans_32[k][4] * eo[4]
+ g_ai2_ihevc_trans_32[k][5] * eo[5]
+ g_ai2_ihevc_trans_32[k][6] * eo[6]
+ g_ai2_ihevc_trans_32[k][7] * eo[7] + add)
>> shift;
}
for(k = 1; k < 32; k += 2)
{
pi2_dst[k * dst_strd] = (g_ai2_ihevc_trans_32[k][0] * o[0]
+ g_ai2_ihevc_trans_32[k][1] * o[1]
+ g_ai2_ihevc_trans_32[k][2] * o[2]
+ g_ai2_ihevc_trans_32[k][3] * o[3]
+ g_ai2_ihevc_trans_32[k][4] * o[4]
+ g_ai2_ihevc_trans_32[k][5] * o[5]
+ g_ai2_ihevc_trans_32[k][6] * o[6]
+ g_ai2_ihevc_trans_32[k][7] * o[7]
+ g_ai2_ihevc_trans_32[k][8] * o[8]
+ g_ai2_ihevc_trans_32[k][9] * o[9]
+ g_ai2_ihevc_trans_32[k][10] * o[10]
+ g_ai2_ihevc_trans_32[k][11] * o[11]
+ g_ai2_ihevc_trans_32[k][12] * o[12]
+ g_ai2_ihevc_trans_32[k][13] * o[13]
+ g_ai2_ihevc_trans_32[k][14] * o[14]
+ g_ai2_ihevc_trans_32[k][15] * o[15] + add)
>> shift;
}
pi4_temp += trans_size;
pi2_dst++;
}
return u4_blk_sad;
}
void ihevc_resi_trans_32x32_16bit(WORD16 *pi2_src,
UWORD8 *pu1_pred,
WORD16 *pi2_tmp,
WORD16 *pi2_dst,
WORD32 src_strd,
WORD32 pred_strd,
WORD32 dst_strd)
{
WORD32 i, k;
WORD32 e[16], o[16];
WORD32 ee[8], eo[8];
WORD32 eee[4], eeo[4];
WORD32 eeee[2], eeeo[2];
WORD32 add, shift;
WORD32 trans_size;
WORD16 *pi2_tmp_orig;
WORD16 *pi2_dst_orig;
pi2_dst_orig = pi2_dst;
pi2_tmp_orig = pi2_tmp;
trans_size = TRANS_SIZE_32;
/* Residue + Forward Transform 1st stage */
shift = 4; // log2(iWidth) - 1 + g_uiBitIncrement
add = 1 << (shift - 1);
for(i = 0; i < trans_size; i++)
{
WORD32 resi_tmp_1, resi_tmp_2;
/* e and o*/
for(k = 0; k < 16; k++)
{
resi_tmp_1 = pi2_src[k] - pu1_pred[k];
resi_tmp_2 = pi2_src[31 - k] - pu1_pred[31 - k];
e[k] = resi_tmp_1 + resi_tmp_2;
o[k] = resi_tmp_1 - resi_tmp_2;
}
/* ee and eo */
for(k = 0; k < 8; k++)
{
ee[k] = e[k] + e[15 - k];
eo[k] = e[k] - e[15 - k];
}
/* eee and eeo */
for(k = 0; k < 4; k++)
{
eee[k] = ee[k] + ee[7 - k];
eeo[k] = ee[k] - ee[7 - k];
}
/* eeee and eeeo */
eeee[0] = eee[0] + eee[3];
eeeo[0] = eee[0] - eee[3];
eeee[1] = eee[1] + eee[2];
eeeo[1] = eee[1] - eee[2];
pi2_tmp[0] = (g_ai2_ihevc_trans_32[0][0] * eeee[0]
+ g_ai2_ihevc_trans_32[0][1] * eeee[1] + add) >> shift;
pi2_tmp[16 * trans_size] = (g_ai2_ihevc_trans_32[16][0] * eeee[0]
+ g_ai2_ihevc_trans_32[16][1] * eeee[1] + add) >> shift;
pi2_tmp[8 * trans_size] = (g_ai2_ihevc_trans_32[8][0] * eeeo[0]
+ g_ai2_ihevc_trans_32[8][1] * eeeo[1] + add) >> shift;
pi2_tmp[24 * trans_size] = (g_ai2_ihevc_trans_32[24][0] * eeeo[0]
+ g_ai2_ihevc_trans_32[24][1] * eeeo[1] + add) >> shift;
for(k = 4; k < 32; k += 8)
{
pi2_tmp[k * trans_size] = (g_ai2_ihevc_trans_32[k][0] * eeo[0]
+ g_ai2_ihevc_trans_32[k][1] * eeo[1]
+ g_ai2_ihevc_trans_32[k][2] * eeo[2]
+ g_ai2_ihevc_trans_32[k][3] * eeo[3] + add)
>> shift;
}
for(k = 2; k < 32; k += 4)
{
pi2_tmp[k * trans_size] = (g_ai2_ihevc_trans_32[k][0] * eo[0]
+ g_ai2_ihevc_trans_32[k][1] * eo[1]
+ g_ai2_ihevc_trans_32[k][2] * eo[2]
+ g_ai2_ihevc_trans_32[k][3] * eo[3]
+ g_ai2_ihevc_trans_32[k][4] * eo[4]
+ g_ai2_ihevc_trans_32[k][5] * eo[5]
+ g_ai2_ihevc_trans_32[k][6] * eo[6]
+ g_ai2_ihevc_trans_32[k][7] * eo[7] + add)
>> shift;
}
for(k = 1; k < 32; k += 2)
{
pi2_tmp[k * trans_size] = (g_ai2_ihevc_trans_32[k][0] * o[0]
+ g_ai2_ihevc_trans_32[k][1] * o[1]
+ g_ai2_ihevc_trans_32[k][2] * o[2]
+ g_ai2_ihevc_trans_32[k][3] * o[3]
+ g_ai2_ihevc_trans_32[k][4] * o[4]
+ g_ai2_ihevc_trans_32[k][5] * o[5]
+ g_ai2_ihevc_trans_32[k][6] * o[6]
+ g_ai2_ihevc_trans_32[k][7] * o[7]
+ g_ai2_ihevc_trans_32[k][8] * o[8]
+ g_ai2_ihevc_trans_32[k][9] * o[9]
+ g_ai2_ihevc_trans_32[k][10] * o[10]
+ g_ai2_ihevc_trans_32[k][11] * o[11]
+ g_ai2_ihevc_trans_32[k][12] * o[12]
+ g_ai2_ihevc_trans_32[k][13] * o[13]
+ g_ai2_ihevc_trans_32[k][14] * o[14]
+ g_ai2_ihevc_trans_32[k][15] * o[15] + add)
>> shift;
}
pi2_src += src_strd;
pu1_pred += pred_strd;
pi2_tmp++;
}
pi2_tmp = pi2_tmp_orig;
/* Forward Transform 2nd stage */
shift = 11; // log2(iHeight) + 6
add = 1 << (shift - 1);
for(i = 0; i < TRANS_SIZE_32; i++)
{
/* e and o*/
for(k = 0; k < 16; k++)
{
e[k] = pi2_tmp[k] + pi2_tmp[31 - k];
o[k] = pi2_tmp[k] - pi2_tmp[31 - k];
}
/* ee and eo */
for(k = 0; k < 8; k++)
{
ee[k] = e[k] + e[15 - k];
eo[k] = e[k] - e[15 - k];
}
/* eee and eeo */
for(k = 0; k < 4; k++)
{
eee[k] = ee[k] + ee[7 - k];
eeo[k] = ee[k] - ee[7 - k];
}
/* eeee and eeeo */
eeee[0] = eee[0] + eee[3];
eeeo[0] = eee[0] - eee[3];
eeee[1] = eee[1] + eee[2];
eeeo[1] = eee[1] - eee[2];
pi2_dst[0] = (g_ai2_ihevc_trans_32[0][0] * eeee[0]
+ g_ai2_ihevc_trans_32[0][1] * eeee[1] + add) >> shift;
pi2_dst[16 * dst_strd] = (g_ai2_ihevc_trans_32[16][0] * eeee[0]
+ g_ai2_ihevc_trans_32[16][1] * eeee[1] + add) >> shift;
pi2_dst[8 * dst_strd] = (g_ai2_ihevc_trans_32[8][0] * eeeo[0]
+ g_ai2_ihevc_trans_32[8][1] * eeeo[1] + add) >> shift;
pi2_dst[24 * dst_strd] = (g_ai2_ihevc_trans_32[24][0] * eeeo[0]
+ g_ai2_ihevc_trans_32[24][1] * eeeo[1] + add) >> shift;
for(k = 4; k < 32; k += 8)
{
pi2_dst[k * dst_strd] = (g_ai2_ihevc_trans_32[k][0] * eeo[0]
+ g_ai2_ihevc_trans_32[k][1] * eeo[1]
+ g_ai2_ihevc_trans_32[k][2] * eeo[2]
+ g_ai2_ihevc_trans_32[k][3] * eeo[3] + add)
>> shift;
}
for(k = 2; k < 32; k += 4)
{
pi2_dst[k * dst_strd] = (g_ai2_ihevc_trans_32[k][0] * eo[0]
+ g_ai2_ihevc_trans_32[k][1] * eo[1]
+ g_ai2_ihevc_trans_32[k][2] * eo[2]
+ g_ai2_ihevc_trans_32[k][3] * eo[3]
+ g_ai2_ihevc_trans_32[k][4] * eo[4]
+ g_ai2_ihevc_trans_32[k][5] * eo[5]
+ g_ai2_ihevc_trans_32[k][6] * eo[6]
+ g_ai2_ihevc_trans_32[k][7] * eo[7] + add)
>> shift;
}
for(k = 1; k < 32; k += 2)
{
pi2_dst[k * dst_strd] = (g_ai2_ihevc_trans_32[k][0] * o[0]
+ g_ai2_ihevc_trans_32[k][1] * o[1]
+ g_ai2_ihevc_trans_32[k][2] * o[2]
+ g_ai2_ihevc_trans_32[k][3] * o[3]
+ g_ai2_ihevc_trans_32[k][4] * o[4]
+ g_ai2_ihevc_trans_32[k][5] * o[5]
+ g_ai2_ihevc_trans_32[k][6] * o[6]
+ g_ai2_ihevc_trans_32[k][7] * o[7]
+ g_ai2_ihevc_trans_32[k][8] * o[8]
+ g_ai2_ihevc_trans_32[k][9] * o[9]
+ g_ai2_ihevc_trans_32[k][10] * o[10]
+ g_ai2_ihevc_trans_32[k][11] * o[11]
+ g_ai2_ihevc_trans_32[k][12] * o[12]
+ g_ai2_ihevc_trans_32[k][13] * o[13]
+ g_ai2_ihevc_trans_32[k][14] * o[14]
+ g_ai2_ihevc_trans_32[k][15] * o[15] + add)
>> shift;
}
pi2_tmp += trans_size;
pi2_dst++;
}
}