/******************************************************************************
*
* Copyright (C) 2012 Ittiam Systems Pvt Ltd, Bangalore
*
* 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.
*
******************************************************************************/
/**
*******************************************************************************
* @file
* ihevc_chroma_intra_pred_filters_x86_intr.c
*
* @brief
* Contains function Definition for intra prediction interpolation filters
*
*
* @author
* Ittiam
*
* @par List of Functions:
* ihevc_intra_pred_chroma_planar_sse42()
*
* ihevc_intra_pred_chroma_dc_sse42()
*
* @remarks
* None
*
*******************************************************************************
*/
/*****************************************************************************/
/* File Includes */
/*****************************************************************************/
#include "ihevc_typedefs.h"
#include "ihevc_macros.h"
#include "ihevc_func_selector.h"
#include "ihevc_platform_macros.h"
#include "ihevc_intra_pred.h"
#include "ihevc_chroma_intra_pred.h"
#include "ihevc_common_tables.h"
#include "ihevc_tables_x86_intr.h"
#include <mmintrin.h>
#include <xmmintrin.h>
#include <emmintrin.h>
#include <smmintrin.h>
#include <immintrin.h>
/****************************************************************************/
/* Constant Macros */
/****************************************************************************/
#define MAX_CU_SIZE 64
#define BIT_DEPTH 8
#define T32_4NT 128
#define T16_4NT 64
#define T16C_4NT 64
#define T8C_4NT 32
/****************************************************************************/
/* Function Macros */
/****************************************************************************/
#define GET_BIT(y,x) ((y) & (1 << x)) && (1 << x)
/* tables to shuffle 8-bit values */
/*****************************************************************************/
/* Function Definition */
/*****************************************************************************/
/**
*******************************************************************************
*
* @brief
* Planar Intraprediction with reference neighboring samples location
* pointed by 'pu1_ref' to the TU block location pointed by 'pu1_dst' Refer
* to section 8.4.4.2.4 in the standard
*
* @par Description:
*
*
* @param[in] pu1_src
* UWORD8 pointer to the source
*
* @param[in] pu1_dst
* UWORD8 pointer to the destination
*
* @param[in] src_strd
* integer source stride
*
* @param[in] dst_strd
* integer destination stride
*
* @param[in] nt
* integer Transform Block size
*
* @param[in] mode
* integer intraprediction mode
*
* @returns
*
* @remarks
* None
*
*******************************************************************************
*/
void ihevc_intra_pred_chroma_planar_sse42(UWORD8 *pu1_ref,
WORD32 src_strd,
UWORD8 *pu1_dst,
WORD32 dst_strd,
WORD32 nt,
WORD32 mode)
{
WORD32 row, col;
WORD32 log2nt = 5;
WORD32 two_nt, three_nt;
__m128i const_temp_4x32b, const_temp1_4x32b, const_temp2_4x32b, const_temp3_4x32b, const_temp4_4x32b;
__m128i col_8x16b, const_temp5_4x32b, const_temp6_4x32b, zero_8x16b, const_temp7_4x32b;
UNUSED(src_strd);
UNUSED(mode);
switch(nt)
{
case 16:
log2nt = 4;
break;
case 8:
log2nt = 3;
break;
case 4:
log2nt = 2;
break;
default:
break;
}
two_nt = 2 * nt;
three_nt = 3 * nt;
/* Planar filtering */
/* setting vallues in registera*/
// pu1_ref[2*(two_nt - 1 - row)]
// pu1_ref[2 * (three_nt + 1)]
// pu1_ref[2 * (two_nt + 1) + col]
// pu1_ref[2 * (nt - 1)]
const_temp_4x32b = _mm_set_epi16(pu1_ref[2 * (three_nt + 1) + 1], pu1_ref[2 * (three_nt + 1)], pu1_ref[2 * (three_nt + 1) + 1],
pu1_ref[2 * (three_nt + 1)], pu1_ref[2 * (three_nt + 1) + 1], pu1_ref[2 * (three_nt + 1)],
pu1_ref[2 * (three_nt + 1) + 1], pu1_ref[2 * (three_nt + 1)]);
const_temp1_4x32b = _mm_set_epi16(pu1_ref[2 * (nt - 1) + 1], pu1_ref[2 * (nt - 1)], pu1_ref[2 * (nt - 1) + 1], pu1_ref[2 * (nt - 1)],
pu1_ref[2 * (nt - 1) + 1], pu1_ref[2 * (nt - 1)], pu1_ref[2 * (nt - 1) + 1], pu1_ref[2 * (nt - 1)]);
const_temp4_4x32b = _mm_set1_epi16(nt - 1);
const_temp6_4x32b = _mm_set1_epi16(nt);
const_temp7_4x32b = _mm_set1_epi16(4);
zero_8x16b = _mm_set1_epi32(0);
if(nt % 4 == 0)
{
const_temp7_4x32b = _mm_set1_epi16(4);
for(row = 0; row < nt; row++)
{
__m128i res_temp_8x16b, row_8x16b, res_temp1_8x16b, res_temp2_8x16b;
__m128i res_temp3_8x16b;
const_temp2_4x32b = _mm_set_epi16(pu1_ref[2 * (two_nt - 1 - row) + 1], pu1_ref[2 * (two_nt - 1 - row)], pu1_ref[2 * (two_nt - 1 - row) + 1],
pu1_ref[2 * (two_nt - 1 - row)], pu1_ref[2 * (two_nt - 1 - row) + 1], pu1_ref[2 * (two_nt - 1 - row)],
pu1_ref[2 * (two_nt - 1 - row) + 1], pu1_ref[2 * (two_nt - 1 - row)]);
const_temp3_4x32b = _mm_set1_epi16((row + 1));
row_8x16b = _mm_set1_epi16((nt - 1 - row));
const_temp5_4x32b = _mm_set_epi16(3, 3, 2, 2, 1, 1, 0, 0);
col_8x16b = _mm_set_epi16(4, 4, 3, 3, 2, 2, 1, 1);
const_temp5_4x32b = _mm_sub_epi16(const_temp4_4x32b, const_temp5_4x32b);
/*(row + 1) * pu1_ref[nt - 1]*/
res_temp_8x16b = _mm_mullo_epi16(const_temp3_4x32b, const_temp1_4x32b);
/*(row + 1) * pu1_ref[nt - 1] + nt)*/
res_temp_8x16b = _mm_add_epi16(res_temp_8x16b, const_temp6_4x32b);
for(col = 0; col < 2 * nt; col += 8)
{
__m128i src_temp_8x16b;
/* loding 8bit 16 pixles*/
src_temp_8x16b = _mm_loadu_si128((__m128i *)(pu1_ref + 2 * (two_nt + 1) + col));
src_temp_8x16b = _mm_cvtepu8_epi16(src_temp_8x16b); /* row=0*/
/* (nt - 1 - row) * pu1_ref[two_nt + 1 + col] */
res_temp1_8x16b = _mm_mullo_epi16(src_temp_8x16b, row_8x16b);
/*(col + 1) * pu1_ref[three_nt + 1]*/
res_temp2_8x16b = _mm_mullo_epi16(const_temp_4x32b, col_8x16b);
/*(nt - 1 - col)* pu1_ref[two_nt - 1 - row]*/
res_temp3_8x16b = _mm_mullo_epi16(const_temp2_4x32b, const_temp5_4x32b);
res_temp1_8x16b = _mm_add_epi16(res_temp_8x16b, res_temp1_8x16b);
res_temp1_8x16b = _mm_add_epi16(res_temp1_8x16b, res_temp2_8x16b);
res_temp1_8x16b = _mm_add_epi16(res_temp1_8x16b, res_temp3_8x16b);
res_temp1_8x16b = _mm_srli_epi16(res_temp1_8x16b, (log2nt + 1));
res_temp1_8x16b = _mm_packus_epi16(res_temp1_8x16b, zero_8x16b);
_mm_storel_epi64((__m128i *)(pu1_dst + (row * dst_strd) + col), res_temp1_8x16b);
const_temp5_4x32b = _mm_sub_epi16(const_temp5_4x32b, const_temp7_4x32b);
col_8x16b = _mm_add_epi16(col_8x16b, const_temp7_4x32b);
} /* inner loop ends here */
}
}
}
/**
*******************************************************************************
*
* @brief
* Intraprediction for DC mode with reference neighboring samples location
* pointed by 'pu1_ref' to the TU block location pointed by 'pu1_dst' Refer
* to section 8.4.4.2.5 in the standard
*
* @par Description:
*
*
* @param[in] pu1_src
* UWORD8 pointer to the source
*
* @param[in] pu1_dst
* UWORD8 pointer to the destination
*
* @param[in] src_strd
* integer source stride
*
* @param[in] dst_strd
* integer destination stride
*
* @param[in] nt
* integer Transform Block size (Chroma)
*
* @param[in] mode
* integer intraprediction mode
*
* @returns
*
* @remarks
* None
*
*******************************************************************************
*/
void ihevc_intra_pred_chroma_dc_sse42(UWORD8 *pu1_ref,
WORD32 src_strd,
UWORD8 *pu1_dst,
WORD32 dst_strd,
WORD32 nt,
WORD32 mode)
{
WORD32 acc_dc_u, acc_dc_v;
WORD32 dc_val_u, dc_val_v;
WORD32 row;
WORD32 log2nt = 5;
__m128i src_temp1, src_temp3, src_temp4, src_temp5, src_temp6, m_mask;
__m128i src_temp7, src_temp8, src_temp9, src_temp10;
__m128i m_zero = _mm_set1_epi32(0);
UNUSED(src_strd);
UNUSED(mode);
switch(nt)
{
case 32:
log2nt = 5;
break;
case 16:
log2nt = 4;
break;
case 8:
log2nt = 3;
break;
case 4:
log2nt = 2;
break;
default:
break;
}
acc_dc_u = 0;
acc_dc_v = 0;
/* Calculate DC value for the transform block */
m_mask = _mm_loadu_si128((__m128i *)&IHEVCE_SHUFFLEMASKY9[0]);
if(nt == 16)
{
__m128i temp_sad;
src_temp3 = _mm_loadu_si128((__m128i *)(pu1_ref + (2 * nt)));
src_temp4 = _mm_loadu_si128((__m128i *)(pu1_ref + (2 * nt) + 16));
src_temp7 = _mm_loadu_si128((__m128i *)(pu1_ref + (2 * nt) + 32));
src_temp8 = _mm_loadu_si128((__m128i *)(pu1_ref + (2 * nt) + 48));
src_temp5 = _mm_cvtepu8_epi16(src_temp3);
src_temp6 = _mm_cvtepu8_epi16(src_temp4);
src_temp9 = _mm_cvtepu8_epi16(src_temp7);
src_temp10 = _mm_cvtepu8_epi16(src_temp8);
src_temp3 = _mm_srli_si128(src_temp3, 8);
src_temp4 = _mm_srli_si128(src_temp4, 8);
src_temp7 = _mm_srli_si128(src_temp7, 8);
src_temp8 = _mm_srli_si128(src_temp8, 8);
src_temp3 = _mm_cvtepu8_epi16(src_temp3);
src_temp4 = _mm_cvtepu8_epi16(src_temp4);
src_temp7 = _mm_cvtepu8_epi16(src_temp7);
src_temp8 = _mm_cvtepu8_epi16(src_temp8);
src_temp4 = _mm_add_epi16(src_temp4, src_temp6);
src_temp6 = _mm_add_epi16(src_temp3, src_temp5);
src_temp8 = _mm_add_epi16(src_temp7, src_temp8);
src_temp10 = _mm_add_epi16(src_temp9, src_temp10);
src_temp4 = _mm_add_epi16(src_temp4, src_temp6);
src_temp8 = _mm_add_epi16(src_temp8, src_temp10);
src_temp4 = _mm_add_epi16(src_temp4, src_temp8);
src_temp4 = _mm_shuffle_epi8(src_temp4, m_mask);
src_temp4 = _mm_hadd_epi16(src_temp4, m_zero);
src_temp4 = _mm_hadd_epi16(src_temp4, m_zero);
src_temp4 = _mm_cvtepi16_epi32(src_temp4);
temp_sad = _mm_srli_si128(src_temp4, 4); /* Next 32 bits */
acc_dc_u = _mm_cvtsi128_si32(src_temp4);
acc_dc_v = _mm_cvtsi128_si32(temp_sad);
}
else if(nt == 8)
{
__m128i temp_sad;
src_temp3 = _mm_loadu_si128((__m128i *)(pu1_ref + (2 * nt)));
src_temp4 = _mm_loadu_si128((__m128i *)(pu1_ref + (2 * nt) + 16));
src_temp5 = _mm_cvtepu8_epi16(src_temp3);
src_temp6 = _mm_cvtepu8_epi16(src_temp4);
src_temp3 = _mm_srli_si128(src_temp3, 8);
src_temp4 = _mm_srli_si128(src_temp4, 8);
src_temp3 = _mm_cvtepu8_epi16(src_temp3);
src_temp4 = _mm_cvtepu8_epi16(src_temp4);
src_temp4 = _mm_add_epi16(src_temp4, src_temp6);
src_temp6 = _mm_add_epi16(src_temp3, src_temp5);
src_temp4 = _mm_add_epi16(src_temp4, src_temp6);
src_temp4 = _mm_shuffle_epi8(src_temp4, m_mask);
src_temp4 = _mm_hadd_epi16(src_temp4, m_zero);
src_temp4 = _mm_hadd_epi16(src_temp4, m_zero);
src_temp4 = _mm_cvtepi16_epi32(src_temp4);
temp_sad = _mm_srli_si128(src_temp4, 4); /* Next 32 bits */
acc_dc_u = _mm_cvtsi128_si32(src_temp4);
acc_dc_v = _mm_cvtsi128_si32(temp_sad);
}
else if(nt == 4)
{
__m128i temp_sad;
src_temp3 = _mm_loadu_si128((__m128i *)(pu1_ref + (2 * nt)));
src_temp5 = _mm_cvtepu8_epi16(src_temp3);
src_temp4 = _mm_srli_si128(src_temp3, 8);
src_temp4 = _mm_cvtepu8_epi16(src_temp4);
src_temp4 = _mm_add_epi16(src_temp4, src_temp5);
src_temp4 = _mm_shuffle_epi8(src_temp4, m_mask);
src_temp4 = _mm_hadd_epi16(src_temp4, m_zero);
src_temp4 = _mm_hadd_epi16(src_temp4, m_zero);
src_temp4 = _mm_cvtepi16_epi32(src_temp4);
temp_sad = _mm_srli_si128(src_temp4, 4); /* Next 32 bits */
acc_dc_u = _mm_cvtsi128_si32(src_temp4);
acc_dc_v = _mm_cvtsi128_si32(temp_sad);
}
acc_dc_u += pu1_ref[6 * nt];
acc_dc_v += pu1_ref[6 * nt + 1];
acc_dc_u -= pu1_ref[4 * nt];
acc_dc_v -= pu1_ref[4 * nt + 1];
dc_val_u = (acc_dc_u + nt) >> (log2nt + 1);
dc_val_v = (acc_dc_v + nt) >> (log2nt + 1);
dc_val_u = dc_val_u | (dc_val_v << 8);
/* Fill the remaining rows with DC value*/
if(nt == 4)
{
src_temp1 = _mm_set1_epi16(dc_val_u);
/* pu1_dst[(row * dst_strd) + col] = dc_val;*/
_mm_storel_epi64((__m128i *)(pu1_dst + (0 * dst_strd)), src_temp1);
_mm_storel_epi64((__m128i *)(pu1_dst + (1 * dst_strd)), src_temp1);
_mm_storel_epi64((__m128i *)(pu1_dst + (2 * dst_strd)), src_temp1);
_mm_storel_epi64((__m128i *)(pu1_dst + (3 * dst_strd)), src_temp1);
}
else if(nt == 8)
{
src_temp1 = _mm_set1_epi16(dc_val_u);
/* pu1_dst[(row * dst_strd) + col] = dc_val;*/
_mm_storeu_si128((__m128i *)(pu1_dst + (0 * dst_strd)), src_temp1);
_mm_storeu_si128((__m128i *)(pu1_dst + (1 * dst_strd)), src_temp1);
_mm_storeu_si128((__m128i *)(pu1_dst + (2 * dst_strd)), src_temp1);
_mm_storeu_si128((__m128i *)(pu1_dst + (3 * dst_strd)), src_temp1);
_mm_storeu_si128((__m128i *)(pu1_dst + (4 * dst_strd)), src_temp1);
_mm_storeu_si128((__m128i *)(pu1_dst + (5 * dst_strd)), src_temp1);
_mm_storeu_si128((__m128i *)(pu1_dst + (6 * dst_strd)), src_temp1);
_mm_storeu_si128((__m128i *)(pu1_dst + (7 * dst_strd)), src_temp1);
}
else /* nt == 16 */
{
src_temp1 = _mm_set1_epi16(dc_val_u);
for(row = 0; row < nt; row += 8)
{
/* pu1_dst[(row * dst_strd) + col] = dc_val;*/
_mm_storeu_si128((__m128i *)(pu1_dst + (0 * dst_strd)), src_temp1);
_mm_storeu_si128((__m128i *)(pu1_dst + (1 * dst_strd)), src_temp1);
_mm_storeu_si128((__m128i *)(pu1_dst + (2 * dst_strd)), src_temp1);
_mm_storeu_si128((__m128i *)(pu1_dst + (3 * dst_strd)), src_temp1);
_mm_storeu_si128((__m128i *)(pu1_dst + 16 + (0 * dst_strd)), src_temp1);
_mm_storeu_si128((__m128i *)(pu1_dst + 16 + (1 * dst_strd)), src_temp1);
_mm_storeu_si128((__m128i *)(pu1_dst + 16 + (2 * dst_strd)), src_temp1);
_mm_storeu_si128((__m128i *)(pu1_dst + 16 + (3 * dst_strd)), src_temp1);
_mm_storeu_si128((__m128i *)(pu1_dst + (4 * dst_strd)), src_temp1);
_mm_storeu_si128((__m128i *)(pu1_dst + (5 * dst_strd)), src_temp1);
_mm_storeu_si128((__m128i *)(pu1_dst + (6 * dst_strd)), src_temp1);
_mm_storeu_si128((__m128i *)(pu1_dst + (7 * dst_strd)), src_temp1);
_mm_storeu_si128((__m128i *)(pu1_dst + 16 + (4 * dst_strd)), src_temp1);
_mm_storeu_si128((__m128i *)(pu1_dst + 16 + (5 * dst_strd)), src_temp1);
_mm_storeu_si128((__m128i *)(pu1_dst + 16 + (6 * dst_strd)), src_temp1);
_mm_storeu_si128((__m128i *)(pu1_dst + 16 + (7 * dst_strd)), src_temp1);
pu1_dst += 8 * dst_strd;
}
}
}