/****************************************************************************** * * Copyright (C) 2015 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 Name : ih264_weighted_pred_intr_sse42.c */ /* */ /* Description : Contains function definitions for weighted */ /* prediction functions in x86 sse4 intrinsics */ /* */ /* List of Functions : ih264_default_weighted_pred_luma_sse42() */ /* ih264_default_weighted_pred_chroma_sse42() */ /* ih264_weighted_pred_luma_sse42() */ /* ih264_weighted_pred_chroma_sse42() */ /* ih264_weighted_bipred_luma_sse42() */ /* ih264_weighted_bipred_chroma_sse42() */ /* */ /* Issues / Problems : None */ /* */ /* Revision History : */ /* */ /* DD MM YYYY Author(s) Changes */ /* 30 01 2015 Kaushik Initial version */ /* Senthoor */ /* */ /*****************************************************************************/ /*****************************************************************************/ /* File Includes */ /*****************************************************************************/ #include #include "ih264_typedefs.h" #include "ih264_macros.h" #include "ih264_platform_macros.h" #include "ih264_weighted_pred.h" /*****************************************************************************/ /* Function definitions . */ /*****************************************************************************/ /*****************************************************************************/ /* */ /* Function Name : ih264_default_weighted_pred_luma_sse42 */ /* */ /* Description : This function performs the default weighted prediction */ /* as described in sec 8.4.2.3.1 titled "Default weighted */ /* sample prediction process" for luma. The function gets */ /* two ht x wd blocks, calculates their rounded-average and */ /* stores it in the destination block. (ht,wd) can be */ /* (4,4), (8,4), (4,8), (8,8), (16,8), (8,16) or (16,16). */ /* */ /* Inputs : pu1_src1 - Pointer to source 1 */ /* pu1_src2 - Pointer to source 2 */ /* pu1_dst - Pointer to destination */ /* src_strd1 - stride for source 1 */ /* src_strd1 - stride for source 2 */ /* dst_strd - stride for destination */ /* ht - height of the block */ /* wd - width of the block */ /* */ /* Issues : None */ /* */ /* Revision History: */ /* */ /* DD MM YYYY Author(s) Changes */ /* 04 02 2015 Kaushik Initial Version */ /* Senthoor */ /* */ /*****************************************************************************/ void ih264_default_weighted_pred_luma_sse42(UWORD8 *pu1_src1, UWORD8 *pu1_src2, UWORD8 *pu1_dst, WORD32 src_strd1, WORD32 src_strd2, WORD32 dst_strd, WORD32 ht, WORD32 wd) { __m128i y0_0_16x8b, y0_1_16x8b, y0_2_16x8b, y0_3_16x8b; __m128i y1_0_16x8b, y1_1_16x8b, y1_2_16x8b, y1_3_16x8b; if(wd == 4) { do { y0_0_16x8b = _mm_loadl_epi64((__m128i *)pu1_src1); y0_1_16x8b = _mm_loadl_epi64((__m128i *)(pu1_src1 + src_strd1)); y0_2_16x8b = _mm_loadl_epi64( (__m128i *)(pu1_src1 + (src_strd1 << 1))); y0_3_16x8b = _mm_loadl_epi64((__m128i *)(pu1_src1 + src_strd1 * 3)); y1_0_16x8b = _mm_loadl_epi64((__m128i *)pu1_src2); y1_1_16x8b = _mm_loadl_epi64((__m128i *)(pu1_src2 + src_strd2)); y1_2_16x8b = _mm_loadl_epi64( (__m128i *)(pu1_src2 + (src_strd2 << 1))); y1_3_16x8b = _mm_loadl_epi64((__m128i *)(pu1_src2 + src_strd2 * 3)); y0_0_16x8b = _mm_avg_epu8(y0_0_16x8b, y1_0_16x8b); y0_1_16x8b = _mm_avg_epu8(y0_1_16x8b, y1_1_16x8b); y0_2_16x8b = _mm_avg_epu8(y0_2_16x8b, y1_2_16x8b); y0_3_16x8b = _mm_avg_epu8(y0_3_16x8b, y1_3_16x8b); *((WORD32 *)(pu1_dst)) = _mm_cvtsi128_si32(y0_0_16x8b); *((WORD32 *)(pu1_dst + dst_strd)) = _mm_cvtsi128_si32(y0_1_16x8b); *((WORD32 *)(pu1_dst + (dst_strd << 1))) = _mm_cvtsi128_si32(y0_2_16x8b); *((WORD32 *)(pu1_dst + dst_strd * 3)) = _mm_cvtsi128_si32(y0_3_16x8b); ht -= 4; pu1_src1 += src_strd1 << 2; pu1_src2 += src_strd2 << 2; pu1_dst += dst_strd << 2; } while(ht > 0); } else if(wd == 8) { do { y0_0_16x8b = _mm_loadl_epi64((__m128i *)pu1_src1); y0_1_16x8b = _mm_loadl_epi64((__m128i *)(pu1_src1 + src_strd1)); y0_2_16x8b = _mm_loadl_epi64( (__m128i *)(pu1_src1 + (src_strd1 << 1))); y0_3_16x8b = _mm_loadl_epi64((__m128i *)(pu1_src1 + src_strd1 * 3)); y1_0_16x8b = _mm_loadl_epi64((__m128i *)pu1_src2); y1_1_16x8b = _mm_loadl_epi64((__m128i *)(pu1_src2 + src_strd2)); y1_2_16x8b = _mm_loadl_epi64( (__m128i *)(pu1_src2 + (src_strd2 << 1))); y1_3_16x8b = _mm_loadl_epi64((__m128i *)(pu1_src2 + src_strd2 * 3)); y0_0_16x8b = _mm_avg_epu8(y0_0_16x8b, y1_0_16x8b); y0_1_16x8b = _mm_avg_epu8(y0_1_16x8b, y1_1_16x8b); y0_2_16x8b = _mm_avg_epu8(y0_2_16x8b, y1_2_16x8b); y0_3_16x8b = _mm_avg_epu8(y0_3_16x8b, y1_3_16x8b); _mm_storel_epi64((__m128i *)pu1_dst, y0_0_16x8b); _mm_storel_epi64((__m128i *)(pu1_dst + dst_strd), y0_1_16x8b); _mm_storel_epi64((__m128i *)(pu1_dst + (dst_strd << 1)), y0_2_16x8b); _mm_storel_epi64((__m128i *)(pu1_dst + dst_strd * 3), y0_3_16x8b); ht -= 4; pu1_src1 += src_strd1 << 2; pu1_src2 += src_strd2 << 2; pu1_dst += dst_strd << 2; } while(ht > 0); } else // wd == 16 { __m128i y0_4_16x8b, y0_5_16x8b, y0_6_16x8b, y0_7_16x8b; __m128i y1_4_16x8b, y1_5_16x8b, y1_6_16x8b, y1_7_16x8b; do { y0_0_16x8b = _mm_loadu_si128((__m128i *)pu1_src1); y0_1_16x8b = _mm_loadu_si128((__m128i *)(pu1_src1 + src_strd1)); y0_2_16x8b = _mm_loadu_si128( (__m128i *)(pu1_src1 + (src_strd1 << 1))); y0_3_16x8b = _mm_loadu_si128((__m128i *)(pu1_src1 + src_strd1 * 3)); y0_4_16x8b = _mm_loadu_si128( (__m128i *)(pu1_src1 + (src_strd1 << 2))); y0_5_16x8b = _mm_loadu_si128((__m128i *)(pu1_src1 + src_strd1 * 5)); y0_6_16x8b = _mm_loadu_si128((__m128i *)(pu1_src1 + src_strd1 * 6)); y0_7_16x8b = _mm_loadu_si128((__m128i *)(pu1_src1 + src_strd1 * 7)); y1_0_16x8b = _mm_loadu_si128((__m128i *)pu1_src2); y1_1_16x8b = _mm_loadu_si128((__m128i *)(pu1_src2 + src_strd2)); y1_2_16x8b = _mm_loadu_si128( (__m128i *)(pu1_src2 + (src_strd2 << 1))); y1_3_16x8b = _mm_loadu_si128((__m128i *)(pu1_src2 + src_strd2 * 3)); y1_4_16x8b = _mm_loadu_si128( (__m128i *)(pu1_src2 + (src_strd2 << 2))); y1_5_16x8b = _mm_loadu_si128((__m128i *)(pu1_src2 + src_strd2 * 5)); y1_6_16x8b = _mm_loadu_si128((__m128i *)(pu1_src2 + src_strd2 * 6)); y1_7_16x8b = _mm_loadu_si128((__m128i *)(pu1_src2 + src_strd2 * 7)); y0_0_16x8b = _mm_avg_epu8(y0_0_16x8b, y1_0_16x8b); y0_1_16x8b = _mm_avg_epu8(y0_1_16x8b, y1_1_16x8b); y0_2_16x8b = _mm_avg_epu8(y0_2_16x8b, y1_2_16x8b); y0_3_16x8b = _mm_avg_epu8(y0_3_16x8b, y1_3_16x8b); y0_4_16x8b = _mm_avg_epu8(y0_4_16x8b, y1_4_16x8b); y0_5_16x8b = _mm_avg_epu8(y0_5_16x8b, y1_5_16x8b); y0_6_16x8b = _mm_avg_epu8(y0_6_16x8b, y1_6_16x8b); y0_7_16x8b = _mm_avg_epu8(y0_7_16x8b, y1_7_16x8b); _mm_storeu_si128((__m128i *)pu1_dst, y0_0_16x8b); _mm_storeu_si128((__m128i *)(pu1_dst + dst_strd), y0_1_16x8b); _mm_storeu_si128((__m128i *)(pu1_dst + (dst_strd << 1)), y0_2_16x8b); _mm_storeu_si128((__m128i *)(pu1_dst + dst_strd * 3), y0_3_16x8b); _mm_storeu_si128((__m128i *)(pu1_dst + (dst_strd << 2)), y0_4_16x8b); _mm_storeu_si128((__m128i *)(pu1_dst + dst_strd * 5), y0_5_16x8b); _mm_storeu_si128((__m128i *)(pu1_dst + dst_strd * 6), y0_6_16x8b); _mm_storeu_si128((__m128i *)(pu1_dst + dst_strd * 7), y0_7_16x8b); ht -= 8; pu1_src1 += src_strd1 << 3; pu1_src2 += src_strd2 << 3; pu1_dst += dst_strd << 3; } while(ht > 0); } } /*****************************************************************************/ /* */ /* Function Name : ih264_default_weighted_pred_chroma_sse42 */ /* */ /* Description : This function performs the default weighted prediction */ /* as described in sec 8.4.2.3.1 titled "Default weighted */ /* sample prediction process" for chroma. The function gets */ /* two ht x wd blocks, calculates their rounded-average and */ /* stores it in the destination block. (ht,wd) can be */ /* (2,2), (4,2) , (2,4), (4,4), (8,4), (4,8) or (8,8). */ /* */ /* Inputs : pu1_src1 - Pointer to source 1 */ /* pu1_src2 - Pointer to source 2 */ /* pu1_dst - Pointer to destination */ /* src_strd1 - stride for source 1 */ /* src_strd1 - stride for source 2 */ /* dst_strd - stride for destination */ /* ht - height of the block */ /* wd - width of the block */ /* */ /* Issues : None */ /* */ /* Revision History: */ /* */ /* DD MM YYYY Author(s) Changes */ /* 04 02 2015 Kaushik Initial Version */ /* Senthoor */ /* */ /*****************************************************************************/ void ih264_default_weighted_pred_chroma_sse42(UWORD8 *pu1_src1, UWORD8 *pu1_src2, UWORD8 *pu1_dst, WORD32 src_strd1, WORD32 src_strd2, WORD32 dst_strd, WORD32 ht, WORD32 wd) { __m128i uv0_0_16x8b, uv0_1_16x8b; __m128i uv1_0_16x8b, uv1_1_16x8b; if(wd == 2) { do { uv0_0_16x8b = _mm_loadl_epi64((__m128i *)pu1_src1); uv0_1_16x8b = _mm_loadl_epi64((__m128i *)(pu1_src1 + src_strd1)); uv1_0_16x8b = _mm_loadl_epi64((__m128i *)pu1_src2); uv1_1_16x8b = _mm_loadl_epi64((__m128i *)(pu1_src2 + src_strd2)); uv0_0_16x8b = _mm_avg_epu8(uv0_0_16x8b, uv1_0_16x8b); uv0_1_16x8b = _mm_avg_epu8(uv0_1_16x8b, uv1_1_16x8b); *((WORD32 *)(pu1_dst)) = _mm_cvtsi128_si32(uv0_0_16x8b); *((WORD32 *)(pu1_dst + dst_strd)) = _mm_cvtsi128_si32(uv0_1_16x8b); ht -= 2; pu1_src1 += src_strd1 << 1; pu1_src2 += src_strd2 << 1; pu1_dst += dst_strd << 1; } while(ht > 0); } else if(wd == 4) { do { uv0_0_16x8b = _mm_loadl_epi64((__m128i *)pu1_src1); uv0_1_16x8b = _mm_loadl_epi64((__m128i *)(pu1_src1 + src_strd1)); uv1_0_16x8b = _mm_loadl_epi64((__m128i *)pu1_src2); uv1_1_16x8b = _mm_loadl_epi64((__m128i *)(pu1_src2 + src_strd2)); uv0_0_16x8b = _mm_avg_epu8(uv0_0_16x8b, uv1_0_16x8b); uv0_1_16x8b = _mm_avg_epu8(uv0_1_16x8b, uv1_1_16x8b); _mm_storel_epi64((__m128i *)pu1_dst, uv0_0_16x8b); _mm_storel_epi64((__m128i *)(pu1_dst + dst_strd), uv0_1_16x8b); ht -= 2; pu1_src1 += src_strd1 << 1; pu1_src2 += src_strd2 << 1; pu1_dst += dst_strd << 1; } while(ht > 0); } else // wd == 8 { __m128i uv0_2_16x8b, uv0_3_16x8b; __m128i uv1_2_16x8b, uv1_3_16x8b; do { uv0_0_16x8b = _mm_loadu_si128((__m128i *)pu1_src1); uv0_1_16x8b = _mm_loadu_si128((__m128i *)(pu1_src1 + src_strd1)); uv0_2_16x8b = _mm_loadu_si128( (__m128i *)(pu1_src1 + (src_strd1 << 1))); uv0_3_16x8b = _mm_loadu_si128( (__m128i *)(pu1_src1 + src_strd1 * 3)); uv1_0_16x8b = _mm_loadu_si128((__m128i *)pu1_src2); uv1_1_16x8b = _mm_loadu_si128((__m128i *)(pu1_src2 + src_strd2)); uv1_2_16x8b = _mm_loadu_si128( (__m128i *)(pu1_src2 + (src_strd2 << 1))); uv1_3_16x8b = _mm_loadu_si128( (__m128i *)(pu1_src2 + src_strd2 * 3)); uv0_0_16x8b = _mm_avg_epu8(uv0_0_16x8b, uv1_0_16x8b); uv0_1_16x8b = _mm_avg_epu8(uv0_1_16x8b, uv1_1_16x8b); uv0_2_16x8b = _mm_avg_epu8(uv0_2_16x8b, uv1_2_16x8b); uv0_3_16x8b = _mm_avg_epu8(uv0_3_16x8b, uv1_3_16x8b); _mm_storeu_si128((__m128i *)pu1_dst, uv0_0_16x8b); _mm_storeu_si128((__m128i *)(pu1_dst + dst_strd), uv0_1_16x8b); _mm_storeu_si128( (__m128i *)(pu1_dst + (dst_strd << 1)), uv0_2_16x8b); _mm_storeu_si128((__m128i *)(pu1_dst + dst_strd * 3), uv0_3_16x8b); ht -= 4; pu1_src1 += src_strd1 << 2; pu1_src2 += src_strd2 << 2; pu1_dst += dst_strd << 2; } while(ht > 0); } } /*****************************************************************************/ /* */ /* Function Name : ih264_weighted_pred_luma_sse42 */ /* */ /* Description : This function performs the weighted prediction as */ /* described in sec 8.4.2.3.2 titled "Weighted sample */ /* prediction process" for luma. The function gets one */ /* ht x wd block, weights it, rounds it off, offsets it, */ /* saturates it to unsigned 8-bit and stores it in the */ /* destination block. (ht,wd) can be (4,4), (8,4), (4,8), */ /* (8,8), (16,8), (8,16) or (16,16). */ /* */ /* Inputs : pu1_src - Pointer to source */ /* pu1_dst - Pointer to destination */ /* src_strd - stride for source */ /* dst_strd - stride for destination */ /* log_wd - number of bits to be rounded off */ /* wt - weight value */ /* ofst - offset value */ /* ht - height of the block */ /* wd - width of the block */ /* */ /* Issues : None */ /* */ /* Revision History: */ /* */ /* DD MM YYYY Author(s) Changes */ /* 04 02 2015 Kaushik Initial Version */ /* Senthoor */ /* */ /*****************************************************************************/ void ih264_weighted_pred_luma_sse42(UWORD8 *pu1_src, UWORD8 *pu1_dst, WORD32 src_strd, WORD32 dst_strd, WORD32 log_wd, WORD32 wt, WORD32 ofst, WORD32 ht, WORD32 wd) { __m128i y_0_16x8b, y_1_16x8b, y_2_16x8b, y_3_16x8b; __m128i wt_8x16b, round_8x16b, ofst_8x16b; WORD32 round_val; wt = (WORD16)(wt & 0xffff); round_val = 1 << (log_wd - 1); ofst = (WORD8)(ofst & 0xff); wt_8x16b = _mm_set1_epi16(wt); round_8x16b = _mm_set1_epi16(round_val); ofst_8x16b = _mm_set1_epi16(ofst); if(wd == 4) { __m128i y_0_8x16b, y_2_8x16b; do { y_0_16x8b = _mm_loadl_epi64((__m128i *)pu1_src); y_1_16x8b = _mm_loadl_epi64((__m128i *)(pu1_src + src_strd)); y_2_16x8b = _mm_loadl_epi64((__m128i *)(pu1_src + (src_strd << 1))); y_3_16x8b = _mm_loadl_epi64((__m128i *)(pu1_src + src_strd * 3)); y_0_16x8b = _mm_unpacklo_epi32(y_0_16x8b, y_1_16x8b); y_2_16x8b = _mm_unpacklo_epi32(y_2_16x8b, y_3_16x8b); y_0_8x16b = _mm_cvtepu8_epi16(y_0_16x8b); y_2_8x16b = _mm_cvtepu8_epi16(y_2_16x8b); y_0_8x16b = _mm_mullo_epi16(y_0_8x16b, wt_8x16b); y_2_8x16b = _mm_mullo_epi16(y_2_8x16b, wt_8x16b); y_0_8x16b = _mm_adds_epi16(round_8x16b, y_0_8x16b); y_2_8x16b = _mm_adds_epi16(round_8x16b, y_2_8x16b); y_0_8x16b = _mm_srai_epi16(y_0_8x16b, log_wd); y_2_8x16b = _mm_srai_epi16(y_2_8x16b, log_wd); y_0_8x16b = _mm_adds_epi16(ofst_8x16b, y_0_8x16b); y_2_8x16b = _mm_adds_epi16(ofst_8x16b, y_2_8x16b); y_0_16x8b = _mm_packus_epi16(y_0_8x16b, y_2_8x16b); y_1_16x8b = _mm_srli_si128(y_0_16x8b, 4); y_2_16x8b = _mm_srli_si128(y_0_16x8b, 8); y_3_16x8b = _mm_srli_si128(y_0_16x8b, 12); *((WORD32 *)(pu1_dst)) = _mm_cvtsi128_si32(y_0_16x8b); *((WORD32 *)(pu1_dst + dst_strd)) = _mm_cvtsi128_si32(y_1_16x8b); *((WORD32 *)(pu1_dst + (dst_strd << 1))) = _mm_cvtsi128_si32(y_2_16x8b); *((WORD32 *)(pu1_dst + dst_strd * 3)) = _mm_cvtsi128_si32(y_3_16x8b); ht -= 4; pu1_src += src_strd << 2; pu1_dst += dst_strd << 2; } while(ht > 0); } else if(wd == 8) { __m128i y_0_8x16b, y_1_8x16b, y_2_8x16b, y_3_8x16b; do { y_0_16x8b = _mm_loadl_epi64((__m128i *)pu1_src); y_1_16x8b = _mm_loadl_epi64((__m128i *)(pu1_src + src_strd)); y_2_16x8b = _mm_loadl_epi64((__m128i *)(pu1_src + (src_strd << 1))); y_3_16x8b = _mm_loadl_epi64((__m128i *)(pu1_src + src_strd * 3)); y_0_8x16b = _mm_cvtepu8_epi16(y_0_16x8b); y_1_8x16b = _mm_cvtepu8_epi16(y_1_16x8b); y_2_8x16b = _mm_cvtepu8_epi16(y_2_16x8b); y_3_8x16b = _mm_cvtepu8_epi16(y_3_16x8b); y_0_8x16b = _mm_mullo_epi16(y_0_8x16b, wt_8x16b); y_1_8x16b = _mm_mullo_epi16(y_1_8x16b, wt_8x16b); y_2_8x16b = _mm_mullo_epi16(y_2_8x16b, wt_8x16b); y_3_8x16b = _mm_mullo_epi16(y_3_8x16b, wt_8x16b); y_0_8x16b = _mm_adds_epi16(round_8x16b, y_0_8x16b); y_1_8x16b = _mm_adds_epi16(round_8x16b, y_1_8x16b); y_2_8x16b = _mm_adds_epi16(round_8x16b, y_2_8x16b); y_3_8x16b = _mm_adds_epi16(round_8x16b, y_3_8x16b); y_0_8x16b = _mm_srai_epi16(y_0_8x16b, log_wd); y_1_8x16b = _mm_srai_epi16(y_1_8x16b, log_wd); y_2_8x16b = _mm_srai_epi16(y_2_8x16b, log_wd); y_3_8x16b = _mm_srai_epi16(y_3_8x16b, log_wd); y_0_8x16b = _mm_adds_epi16(ofst_8x16b, y_0_8x16b); y_1_8x16b = _mm_adds_epi16(ofst_8x16b, y_1_8x16b); y_2_8x16b = _mm_adds_epi16(ofst_8x16b, y_2_8x16b); y_3_8x16b = _mm_adds_epi16(ofst_8x16b, y_3_8x16b); y_0_16x8b = _mm_packus_epi16(y_0_8x16b, y_1_8x16b); y_2_16x8b = _mm_packus_epi16(y_2_8x16b, y_3_8x16b); y_1_16x8b = _mm_srli_si128(y_0_16x8b, 8); y_3_16x8b = _mm_srli_si128(y_2_16x8b, 8); _mm_storel_epi64((__m128i *)pu1_dst, y_0_16x8b); _mm_storel_epi64((__m128i *)(pu1_dst + dst_strd), y_1_16x8b); _mm_storel_epi64((__m128i *)(pu1_dst + (dst_strd << 1)), y_2_16x8b); _mm_storel_epi64((__m128i *)(pu1_dst + dst_strd * 3), y_3_16x8b); ht -= 4; pu1_src += src_strd << 2; pu1_dst += dst_strd << 2; } while(ht > 0); } else // wd == 16 { __m128i y_0L_8x16b, y_1L_8x16b, y_2L_8x16b, y_3L_8x16b; __m128i y_0H_8x16b, y_1H_8x16b, y_2H_8x16b, y_3H_8x16b; __m128i zero_16x8b; zero_16x8b = _mm_set1_epi8(0); do { y_0_16x8b = _mm_loadu_si128((__m128i *)pu1_src); y_1_16x8b = _mm_loadu_si128((__m128i *)(pu1_src + src_strd)); y_2_16x8b = _mm_loadu_si128((__m128i *)(pu1_src + (src_strd << 1))); y_3_16x8b = _mm_loadu_si128((__m128i *)(pu1_src + src_strd * 3)); y_0L_8x16b = _mm_cvtepu8_epi16(y_0_16x8b); y_0H_8x16b = _mm_unpackhi_epi8(y_0_16x8b, zero_16x8b); y_1L_8x16b = _mm_cvtepu8_epi16(y_1_16x8b); y_1H_8x16b = _mm_unpackhi_epi8(y_1_16x8b, zero_16x8b); y_2L_8x16b = _mm_cvtepu8_epi16(y_2_16x8b); y_2H_8x16b = _mm_unpackhi_epi8(y_2_16x8b, zero_16x8b); y_3L_8x16b = _mm_cvtepu8_epi16(y_3_16x8b); y_3H_8x16b = _mm_unpackhi_epi8(y_3_16x8b, zero_16x8b); y_0L_8x16b = _mm_mullo_epi16(y_0L_8x16b, wt_8x16b); y_0H_8x16b = _mm_mullo_epi16(y_0H_8x16b, wt_8x16b); y_1L_8x16b = _mm_mullo_epi16(y_1L_8x16b, wt_8x16b); y_1H_8x16b = _mm_mullo_epi16(y_1H_8x16b, wt_8x16b); y_2L_8x16b = _mm_mullo_epi16(y_2L_8x16b, wt_8x16b); y_2H_8x16b = _mm_mullo_epi16(y_2H_8x16b, wt_8x16b); y_3L_8x16b = _mm_mullo_epi16(y_3L_8x16b, wt_8x16b); y_3H_8x16b = _mm_mullo_epi16(y_3H_8x16b, wt_8x16b); y_0L_8x16b = _mm_adds_epi16(round_8x16b, y_0L_8x16b); y_0H_8x16b = _mm_adds_epi16(round_8x16b, y_0H_8x16b); y_1L_8x16b = _mm_adds_epi16(round_8x16b, y_1L_8x16b); y_1H_8x16b = _mm_adds_epi16(round_8x16b, y_1H_8x16b); y_2L_8x16b = _mm_adds_epi16(round_8x16b, y_2L_8x16b); y_2H_8x16b = _mm_adds_epi16(round_8x16b, y_2H_8x16b); y_3L_8x16b = _mm_adds_epi16(round_8x16b, y_3L_8x16b); y_3H_8x16b = _mm_adds_epi16(round_8x16b, y_3H_8x16b); y_0L_8x16b = _mm_srai_epi16(y_0L_8x16b, log_wd); y_0H_8x16b = _mm_srai_epi16(y_0H_8x16b, log_wd); y_1L_8x16b = _mm_srai_epi16(y_1L_8x16b, log_wd); y_1H_8x16b = _mm_srai_epi16(y_1H_8x16b, log_wd); y_2L_8x16b = _mm_srai_epi16(y_2L_8x16b, log_wd); y_2H_8x16b = _mm_srai_epi16(y_2H_8x16b, log_wd); y_3L_8x16b = _mm_srai_epi16(y_3L_8x16b, log_wd); y_3H_8x16b = _mm_srai_epi16(y_3H_8x16b, log_wd); y_0L_8x16b = _mm_adds_epi16(ofst_8x16b, y_0L_8x16b); y_0H_8x16b = _mm_adds_epi16(ofst_8x16b, y_0H_8x16b); y_1L_8x16b = _mm_adds_epi16(ofst_8x16b, y_1L_8x16b); y_1H_8x16b = _mm_adds_epi16(ofst_8x16b, y_1H_8x16b); y_2L_8x16b = _mm_adds_epi16(ofst_8x16b, y_2L_8x16b); y_2H_8x16b = _mm_adds_epi16(ofst_8x16b, y_2H_8x16b); y_3L_8x16b = _mm_adds_epi16(ofst_8x16b, y_3L_8x16b); y_3H_8x16b = _mm_adds_epi16(ofst_8x16b, y_3H_8x16b); y_0_16x8b = _mm_packus_epi16(y_0L_8x16b, y_0H_8x16b); y_1_16x8b = _mm_packus_epi16(y_1L_8x16b, y_1H_8x16b); y_2_16x8b = _mm_packus_epi16(y_2L_8x16b, y_2H_8x16b); y_3_16x8b = _mm_packus_epi16(y_3L_8x16b, y_3H_8x16b); _mm_storeu_si128((__m128i *)pu1_dst, y_0_16x8b); _mm_storeu_si128((__m128i *)(pu1_dst + dst_strd), y_1_16x8b); _mm_storeu_si128((__m128i *)(pu1_dst + (dst_strd << 1)), y_2_16x8b); _mm_storeu_si128((__m128i *)(pu1_dst + dst_strd * 3), y_3_16x8b); ht -= 4; pu1_src += src_strd << 2; pu1_dst += dst_strd << 2; } while(ht > 0); } } /*****************************************************************************/ /* */ /* Function Name : ih264_weighted_pred_chroma_sse42 */ /* */ /* Description : This function performs the weighted prediction as */ /* described in sec 8.4.2.3.2 titled "Weighted sample */ /* prediction process" for chroma. The function gets one */ /* ht x wd block, weights it, rounds it off, offsets it, */ /* saturates it to unsigned 8-bit and stores it in the */ /* destination block. (ht,wd) can be (2,2), (4,2), (2,4), */ /* (4,4), (8,4), (4,8) or (8,8). */ /* */ /* Inputs : pu1_src - Pointer to source */ /* pu1_dst - Pointer to destination */ /* src_strd - stride for source */ /* dst_strd - stride for destination */ /* log_wd - number of bits to be rounded off */ /* wt - weight values for u and v */ /* ofst - offset values for u and v */ /* ht - height of the block */ /* wd - width of the block */ /* */ /* Issues : None */ /* */ /* Revision History: */ /* */ /* DD MM YYYY Author(s) Changes */ /* 04 02 2015 Kaushik Initial Version */ /* Senthoor */ /* */ /*****************************************************************************/ void ih264_weighted_pred_chroma_sse42(UWORD8 *pu1_src, UWORD8 *pu1_dst, WORD32 src_strd, WORD32 dst_strd, WORD32 log_wd, WORD32 wt, WORD32 ofst, WORD32 ht, WORD32 wd) { __m128i y_0_16x8b, y_1_16x8b; __m128i wt_8x16b, round_8x16b, ofst_8x16b; WORD32 ofst_u, ofst_v; WORD32 round_val; ofst_u = (WORD8)(ofst & 0xff); ofst_v = (WORD8)(ofst >> 8); round_val = 1 << (log_wd - 1); ofst = (ofst_u & 0xffff) | (ofst_v << 16); wt_8x16b = _mm_set1_epi32(wt); round_8x16b = _mm_set1_epi16(round_val); ofst_8x16b = _mm_set1_epi32(ofst); if(wd == 2) { __m128i y_0_8x16b; do { y_0_16x8b = _mm_loadl_epi64((__m128i *)pu1_src); y_1_16x8b = _mm_loadl_epi64((__m128i *)(pu1_src + src_strd)); y_0_16x8b = _mm_unpacklo_epi32(y_0_16x8b, y_1_16x8b); y_0_8x16b = _mm_cvtepu8_epi16(y_0_16x8b); y_0_8x16b = _mm_mullo_epi16(y_0_8x16b, wt_8x16b); y_0_8x16b = _mm_adds_epi16(round_8x16b, y_0_8x16b); y_0_8x16b = _mm_srai_epi16(y_0_8x16b, log_wd); y_0_8x16b = _mm_adds_epi16(ofst_8x16b, y_0_8x16b); y_0_16x8b = _mm_packus_epi16(y_0_8x16b, y_0_8x16b); y_1_16x8b = _mm_srli_si128(y_0_16x8b, 4); *((WORD32 *)(pu1_dst)) = _mm_cvtsi128_si32(y_0_16x8b); *((WORD32 *)(pu1_dst + dst_strd)) = _mm_cvtsi128_si32(y_1_16x8b); ht -= 2; pu1_src += src_strd << 1; pu1_dst += dst_strd << 1; } while(ht > 0); } else if(wd == 4) { __m128i y_0_8x16b, y_1_8x16b; do { y_0_16x8b = _mm_loadl_epi64((__m128i *)pu1_src); y_1_16x8b = _mm_loadl_epi64((__m128i *)(pu1_src + src_strd)); y_0_8x16b = _mm_cvtepu8_epi16(y_0_16x8b); y_1_8x16b = _mm_cvtepu8_epi16(y_1_16x8b); y_0_8x16b = _mm_mullo_epi16(y_0_8x16b, wt_8x16b); y_1_8x16b = _mm_mullo_epi16(y_1_8x16b, wt_8x16b); y_0_8x16b = _mm_adds_epi16(round_8x16b, y_0_8x16b); y_1_8x16b = _mm_adds_epi16(round_8x16b, y_1_8x16b); y_0_8x16b = _mm_srai_epi16(y_0_8x16b, log_wd); y_1_8x16b = _mm_srai_epi16(y_1_8x16b, log_wd); y_0_8x16b = _mm_adds_epi16(ofst_8x16b, y_0_8x16b); y_1_8x16b = _mm_adds_epi16(ofst_8x16b, y_1_8x16b); y_0_16x8b = _mm_packus_epi16(y_0_8x16b, y_1_8x16b); y_1_16x8b = _mm_srli_si128(y_0_16x8b, 8); _mm_storel_epi64((__m128i *)pu1_dst, y_0_16x8b); _mm_storel_epi64((__m128i *)(pu1_dst + dst_strd), y_1_16x8b); ht -= 2; pu1_src += src_strd << 1; pu1_dst += dst_strd << 1; } while(ht > 0); } else // wd == 16 { __m128i y_2_16x8b, y_3_16x8b; __m128i y_0L_8x16b, y_1L_8x16b, y_2L_8x16b, y_3L_8x16b; __m128i y_0H_8x16b, y_1H_8x16b, y_2H_8x16b, y_3H_8x16b; __m128i zero_16x8b; zero_16x8b = _mm_set1_epi8(0); do { y_0_16x8b = _mm_loadu_si128((__m128i *)pu1_src); y_1_16x8b = _mm_loadu_si128((__m128i *)(pu1_src + src_strd)); y_2_16x8b = _mm_loadu_si128((__m128i *)(pu1_src + (src_strd << 1))); y_3_16x8b = _mm_loadu_si128((__m128i *)(pu1_src + src_strd * 3)); y_0L_8x16b = _mm_cvtepu8_epi16(y_0_16x8b); y_0H_8x16b = _mm_unpackhi_epi8(y_0_16x8b, zero_16x8b); y_1L_8x16b = _mm_cvtepu8_epi16(y_1_16x8b); y_1H_8x16b = _mm_unpackhi_epi8(y_1_16x8b, zero_16x8b); y_2L_8x16b = _mm_cvtepu8_epi16(y_2_16x8b); y_2H_8x16b = _mm_unpackhi_epi8(y_2_16x8b, zero_16x8b); y_3L_8x16b = _mm_cvtepu8_epi16(y_3_16x8b); y_3H_8x16b = _mm_unpackhi_epi8(y_3_16x8b, zero_16x8b); y_0L_8x16b = _mm_mullo_epi16(y_0L_8x16b, wt_8x16b); y_0H_8x16b = _mm_mullo_epi16(y_0H_8x16b, wt_8x16b); y_1L_8x16b = _mm_mullo_epi16(y_1L_8x16b, wt_8x16b); y_1H_8x16b = _mm_mullo_epi16(y_1H_8x16b, wt_8x16b); y_2L_8x16b = _mm_mullo_epi16(y_2L_8x16b, wt_8x16b); y_2H_8x16b = _mm_mullo_epi16(y_2H_8x16b, wt_8x16b); y_3L_8x16b = _mm_mullo_epi16(y_3L_8x16b, wt_8x16b); y_3H_8x16b = _mm_mullo_epi16(y_3H_8x16b, wt_8x16b); y_0L_8x16b = _mm_adds_epi16(round_8x16b, y_0L_8x16b); y_0H_8x16b = _mm_adds_epi16(round_8x16b, y_0H_8x16b); y_1L_8x16b = _mm_adds_epi16(round_8x16b, y_1L_8x16b); y_1H_8x16b = _mm_adds_epi16(round_8x16b, y_1H_8x16b); y_2L_8x16b = _mm_adds_epi16(round_8x16b, y_2L_8x16b); y_2H_8x16b = _mm_adds_epi16(round_8x16b, y_2H_8x16b); y_3L_8x16b = _mm_adds_epi16(round_8x16b, y_3L_8x16b); y_3H_8x16b = _mm_adds_epi16(round_8x16b, y_3H_8x16b); y_0L_8x16b = _mm_srai_epi16(y_0L_8x16b, log_wd); y_0H_8x16b = _mm_srai_epi16(y_0H_8x16b, log_wd); y_1L_8x16b = _mm_srai_epi16(y_1L_8x16b, log_wd); y_1H_8x16b = _mm_srai_epi16(y_1H_8x16b, log_wd); y_2L_8x16b = _mm_srai_epi16(y_2L_8x16b, log_wd); y_2H_8x16b = _mm_srai_epi16(y_2H_8x16b, log_wd); y_3L_8x16b = _mm_srai_epi16(y_3L_8x16b, log_wd); y_3H_8x16b = _mm_srai_epi16(y_3H_8x16b, log_wd); y_0L_8x16b = _mm_adds_epi16(ofst_8x16b, y_0L_8x16b); y_0H_8x16b = _mm_adds_epi16(ofst_8x16b, y_0H_8x16b); y_1L_8x16b = _mm_adds_epi16(ofst_8x16b, y_1L_8x16b); y_1H_8x16b = _mm_adds_epi16(ofst_8x16b, y_1H_8x16b); y_2L_8x16b = _mm_adds_epi16(ofst_8x16b, y_2L_8x16b); y_2H_8x16b = _mm_adds_epi16(ofst_8x16b, y_2H_8x16b); y_3L_8x16b = _mm_adds_epi16(ofst_8x16b, y_3L_8x16b); y_3H_8x16b = _mm_adds_epi16(ofst_8x16b, y_3H_8x16b); y_0_16x8b = _mm_packus_epi16(y_0L_8x16b, y_0H_8x16b); y_1_16x8b = _mm_packus_epi16(y_1L_8x16b, y_1H_8x16b); y_2_16x8b = _mm_packus_epi16(y_2L_8x16b, y_2H_8x16b); y_3_16x8b = _mm_packus_epi16(y_3L_8x16b, y_3H_8x16b); _mm_storeu_si128((__m128i *)pu1_dst, y_0_16x8b); _mm_storeu_si128((__m128i *)(pu1_dst + dst_strd), y_1_16x8b); _mm_storeu_si128((__m128i *)(pu1_dst + (dst_strd << 1)), y_2_16x8b); _mm_storeu_si128((__m128i *)(pu1_dst + dst_strd * 3), y_3_16x8b); ht -= 4; pu1_src += src_strd << 2; pu1_dst += dst_strd << 2; } while(ht > 0); } } /*****************************************************************************/ /* */ /* Function Name : ih264_weighted_bi_pred_luma_sse42 */ /* */ /* Description : This function performs the weighted biprediction as */ /* described in sec 8.4.2.3.2 titled "Weighted sample */ /* prediction process" for luma. The function gets two */ /* ht x wd blocks, weights them, adds them, rounds off the */ /* sum, offsets it, saturates it to unsigned 8-bit and */ /* stores it in the destination block. (ht,wd) can be */ /* (4,4), (8,4), (4,8), (8,8), (16,8), (8,16) or (16,16). */ /* */ /* Inputs : pu1_src1 - Pointer to source 1 */ /* pu1_src2 - Pointer to source 2 */ /* pu1_dst - Pointer to destination */ /* src_strd1 - stride for source 1 */ /* src_strd2 - stride for source 2 */ /* dst_strd2 - stride for destination */ /* log_wd - number of bits to be rounded off */ /* wt1 - weight value for source 1 */ /* wt2 - weight value for source 2 */ /* ofst1 - offset value for source 1 */ /* ofst2 - offset value for source 2 */ /* ht - height of the block */ /* wd - width of the block */ /* */ /* Issues : None */ /* */ /* Revision History: */ /* */ /* DD MM YYYY Author(s) Changes */ /* 04 02 2015 Kaushik Initial Version */ /* Senthoor */ /* */ /*****************************************************************************/ void ih264_weighted_bi_pred_luma_sse42(UWORD8 *pu1_src1, UWORD8 *pu1_src2, UWORD8 *pu1_dst, WORD32 src_strd1, WORD32 src_strd2, WORD32 dst_strd, WORD32 log_wd, WORD32 wt1, WORD32 wt2, WORD32 ofst1, WORD32 ofst2, WORD32 ht, WORD32 wd) { __m128i y1_0_16x8b, y1_1_16x8b; __m128i y2_0_16x8b, y2_1_16x8b; __m128i wt1_8x16b, wt2_8x16b; __m128i ofst_8x16b, round_8x16b; WORD32 ofst; WORD32 round_val, shft; wt1 = (WORD16)(wt1 & 0xffff); wt2 = (WORD16)(wt2 & 0xffff); round_val = 1 << log_wd; shft = log_wd + 1; ofst1 = (WORD8)(ofst1 & 0xff); ofst2 = (WORD8)(ofst2 & 0xff); ofst = (ofst1 + ofst2 + 1) >> 1; wt1_8x16b = _mm_set1_epi16(wt1); wt2_8x16b = _mm_set1_epi16(wt2); round_8x16b = _mm_set1_epi16(round_val); ofst_8x16b = _mm_set1_epi16(ofst); if(wd == 4) { __m128i y1_2_16x8b, y1_3_16x8b; __m128i y2_2_16x8b, y2_3_16x8b; __m128i y1_0_8x16b, y1_2_8x16b; __m128i y2_0_8x16b, y2_2_8x16b; do { y1_0_16x8b = _mm_loadl_epi64((__m128i *)pu1_src1); y1_1_16x8b = _mm_loadl_epi64((__m128i *)(pu1_src1 + src_strd1)); y1_2_16x8b = _mm_loadl_epi64( (__m128i *)(pu1_src1 + (src_strd1 << 1))); y1_3_16x8b = _mm_loadl_epi64((__m128i *)(pu1_src1 + src_strd1 * 3)); y2_0_16x8b = _mm_loadl_epi64((__m128i *)pu1_src2); y2_1_16x8b = _mm_loadl_epi64((__m128i *)(pu1_src2 + src_strd2)); y2_2_16x8b = _mm_loadl_epi64( (__m128i *)(pu1_src2 + (src_strd2 << 1))); y2_3_16x8b = _mm_loadl_epi64((__m128i *)(pu1_src2 + src_strd2 * 3)); y1_0_16x8b = _mm_unpacklo_epi32(y1_0_16x8b, y1_1_16x8b); y1_2_16x8b = _mm_unpacklo_epi32(y1_2_16x8b, y1_3_16x8b); y2_0_16x8b = _mm_unpacklo_epi32(y2_0_16x8b, y2_1_16x8b); y2_2_16x8b = _mm_unpacklo_epi32(y2_2_16x8b, y2_3_16x8b); y1_0_8x16b = _mm_cvtepu8_epi16(y1_0_16x8b); y1_2_8x16b = _mm_cvtepu8_epi16(y1_2_16x8b); y2_0_8x16b = _mm_cvtepu8_epi16(y2_0_16x8b); y2_2_8x16b = _mm_cvtepu8_epi16(y2_2_16x8b); y1_0_8x16b = _mm_mullo_epi16(y1_0_8x16b, wt1_8x16b); y2_0_8x16b = _mm_mullo_epi16(y2_0_8x16b, wt2_8x16b); y1_2_8x16b = _mm_mullo_epi16(y1_2_8x16b, wt1_8x16b); y2_2_8x16b = _mm_mullo_epi16(y2_2_8x16b, wt2_8x16b); y1_0_8x16b = _mm_adds_epi16(y1_0_8x16b, y2_0_8x16b); y1_2_8x16b = _mm_adds_epi16(y1_2_8x16b, y2_2_8x16b); y1_0_8x16b = _mm_adds_epi16(round_8x16b, y1_0_8x16b); y1_2_8x16b = _mm_adds_epi16(round_8x16b, y1_2_8x16b); y1_0_8x16b = _mm_srai_epi16(y1_0_8x16b, shft); y1_2_8x16b = _mm_srai_epi16(y1_2_8x16b, shft); y1_0_8x16b = _mm_adds_epi16(ofst_8x16b, y1_0_8x16b); y1_2_8x16b = _mm_adds_epi16(ofst_8x16b, y1_2_8x16b); y1_0_16x8b = _mm_packus_epi16(y1_0_8x16b, y1_2_8x16b); y1_1_16x8b = _mm_srli_si128(y1_0_16x8b, 4); y1_2_16x8b = _mm_srli_si128(y1_0_16x8b, 8); y1_3_16x8b = _mm_srli_si128(y1_0_16x8b, 12); *((WORD32 *)(pu1_dst)) = _mm_cvtsi128_si32(y1_0_16x8b); *((WORD32 *)(pu1_dst + dst_strd)) = _mm_cvtsi128_si32(y1_1_16x8b); *((WORD32 *)(pu1_dst + (dst_strd << 1))) = _mm_cvtsi128_si32(y1_2_16x8b); *((WORD32 *)(pu1_dst + dst_strd * 3)) = _mm_cvtsi128_si32(y1_3_16x8b); ht -= 4; pu1_src1 += src_strd1 << 2; pu1_src2 += src_strd2 << 2; pu1_dst += dst_strd << 2; } while(ht > 0); } else if(wd == 8) { __m128i y1_2_16x8b, y1_3_16x8b; __m128i y2_2_16x8b, y2_3_16x8b; __m128i y1_0_8x16b, y1_1_8x16b, y1_2_8x16b, y1_3_8x16b; __m128i y2_0_8x16b, y2_1_8x16b, y2_2_8x16b, y2_3_8x16b; do { y1_0_16x8b = _mm_loadl_epi64((__m128i *)pu1_src1); y1_1_16x8b = _mm_loadl_epi64((__m128i *)(pu1_src1 + src_strd1)); y1_2_16x8b = _mm_loadl_epi64( (__m128i *)(pu1_src1 + (src_strd1 << 1))); y1_3_16x8b = _mm_loadl_epi64((__m128i *)(pu1_src1 + src_strd1 * 3)); y2_0_16x8b = _mm_loadl_epi64((__m128i *)pu1_src2); y2_1_16x8b = _mm_loadl_epi64((__m128i *)(pu1_src2 + src_strd2)); y2_2_16x8b = _mm_loadl_epi64( (__m128i *)(pu1_src2 + (src_strd2 << 1))); y2_3_16x8b = _mm_loadl_epi64((__m128i *)(pu1_src2 + src_strd2 * 3)); y1_0_8x16b = _mm_cvtepu8_epi16(y1_0_16x8b); y1_1_8x16b = _mm_cvtepu8_epi16(y1_1_16x8b); y1_2_8x16b = _mm_cvtepu8_epi16(y1_2_16x8b); y1_3_8x16b = _mm_cvtepu8_epi16(y1_3_16x8b); y2_0_8x16b = _mm_cvtepu8_epi16(y2_0_16x8b); y2_1_8x16b = _mm_cvtepu8_epi16(y2_1_16x8b); y2_2_8x16b = _mm_cvtepu8_epi16(y2_2_16x8b); y2_3_8x16b = _mm_cvtepu8_epi16(y2_3_16x8b); y1_0_8x16b = _mm_mullo_epi16(y1_0_8x16b, wt1_8x16b); y2_0_8x16b = _mm_mullo_epi16(y2_0_8x16b, wt2_8x16b); y1_1_8x16b = _mm_mullo_epi16(y1_1_8x16b, wt1_8x16b); y2_1_8x16b = _mm_mullo_epi16(y2_1_8x16b, wt2_8x16b); y1_2_8x16b = _mm_mullo_epi16(y1_2_8x16b, wt1_8x16b); y2_2_8x16b = _mm_mullo_epi16(y2_2_8x16b, wt2_8x16b); y1_3_8x16b = _mm_mullo_epi16(y1_3_8x16b, wt1_8x16b); y2_3_8x16b = _mm_mullo_epi16(y2_3_8x16b, wt2_8x16b); y1_0_8x16b = _mm_adds_epi16(y1_0_8x16b, y2_0_8x16b); y1_1_8x16b = _mm_adds_epi16(y1_1_8x16b, y2_1_8x16b); y1_2_8x16b = _mm_adds_epi16(y1_2_8x16b, y2_2_8x16b); y1_3_8x16b = _mm_adds_epi16(y1_3_8x16b, y2_3_8x16b); y1_0_8x16b = _mm_adds_epi16(round_8x16b, y1_0_8x16b); y1_1_8x16b = _mm_adds_epi16(round_8x16b, y1_1_8x16b); y1_2_8x16b = _mm_adds_epi16(round_8x16b, y1_2_8x16b); y1_3_8x16b = _mm_adds_epi16(round_8x16b, y1_3_8x16b); y1_0_8x16b = _mm_srai_epi16(y1_0_8x16b, shft); y1_1_8x16b = _mm_srai_epi16(y1_1_8x16b, shft); y1_2_8x16b = _mm_srai_epi16(y1_2_8x16b, shft); y1_3_8x16b = _mm_srai_epi16(y1_3_8x16b, shft); y1_0_8x16b = _mm_adds_epi16(ofst_8x16b, y1_0_8x16b); y1_1_8x16b = _mm_adds_epi16(ofst_8x16b, y1_1_8x16b); y1_2_8x16b = _mm_adds_epi16(ofst_8x16b, y1_2_8x16b); y1_3_8x16b = _mm_adds_epi16(ofst_8x16b, y1_3_8x16b); y1_0_16x8b = _mm_packus_epi16(y1_0_8x16b, y1_1_8x16b); y1_2_16x8b = _mm_packus_epi16(y1_2_8x16b, y1_3_8x16b); y1_1_16x8b = _mm_srli_si128(y1_0_16x8b, 8); y1_3_16x8b = _mm_srli_si128(y1_2_16x8b, 8); _mm_storel_epi64((__m128i *)pu1_dst, y1_0_16x8b); _mm_storel_epi64((__m128i *)(pu1_dst + dst_strd), y1_1_16x8b); _mm_storel_epi64((__m128i *)(pu1_dst + (dst_strd << 1)), y1_2_16x8b); _mm_storel_epi64((__m128i *)(pu1_dst + dst_strd * 3), y1_3_16x8b); ht -= 4; pu1_src1 += src_strd1 << 2; pu1_src2 += src_strd2 << 2; pu1_dst += dst_strd << 2; } while(ht > 0); } else // wd == 16 { __m128i y1_0L_8x16b, y1_0H_8x16b, y1_1L_8x16b, y1_1H_8x16b; __m128i y2_0L_8x16b, y2_0H_8x16b, y2_1L_8x16b, y2_1H_8x16b; __m128i zero_16x8b; zero_16x8b = _mm_set1_epi8(0); do { y1_0_16x8b = _mm_loadu_si128((__m128i *)pu1_src1); y1_1_16x8b = _mm_loadu_si128((__m128i *)(pu1_src1 + src_strd1)); y2_0_16x8b = _mm_loadu_si128((__m128i *)pu1_src2); y2_1_16x8b = _mm_loadu_si128((__m128i *)(pu1_src2 + src_strd2)); y1_0L_8x16b = _mm_cvtepu8_epi16(y1_0_16x8b); y1_0H_8x16b = _mm_unpackhi_epi8(y1_0_16x8b, zero_16x8b); y1_1L_8x16b = _mm_cvtepu8_epi16(y1_1_16x8b); y1_1H_8x16b = _mm_unpackhi_epi8(y1_1_16x8b, zero_16x8b); y2_0L_8x16b = _mm_cvtepu8_epi16(y2_0_16x8b); y2_0H_8x16b = _mm_unpackhi_epi8(y2_0_16x8b, zero_16x8b); y2_1L_8x16b = _mm_cvtepu8_epi16(y2_1_16x8b); y2_1H_8x16b = _mm_unpackhi_epi8(y2_1_16x8b, zero_16x8b); y1_0L_8x16b = _mm_mullo_epi16(y1_0L_8x16b, wt1_8x16b); y1_0H_8x16b = _mm_mullo_epi16(y1_0H_8x16b, wt1_8x16b); y1_1L_8x16b = _mm_mullo_epi16(y1_1L_8x16b, wt1_8x16b); y1_1H_8x16b = _mm_mullo_epi16(y1_1H_8x16b, wt1_8x16b); y2_0L_8x16b = _mm_mullo_epi16(y2_0L_8x16b, wt2_8x16b); y2_0H_8x16b = _mm_mullo_epi16(y2_0H_8x16b, wt2_8x16b); y2_1L_8x16b = _mm_mullo_epi16(y2_1L_8x16b, wt2_8x16b); y2_1H_8x16b = _mm_mullo_epi16(y2_1H_8x16b, wt2_8x16b); y1_0L_8x16b = _mm_adds_epi16(y1_0L_8x16b, y2_0L_8x16b); y1_0H_8x16b = _mm_adds_epi16(y1_0H_8x16b, y2_0H_8x16b); y1_1L_8x16b = _mm_adds_epi16(y1_1L_8x16b, y2_1L_8x16b); y1_1H_8x16b = _mm_adds_epi16(y1_1H_8x16b, y2_1H_8x16b); y1_0L_8x16b = _mm_adds_epi16(round_8x16b, y1_0L_8x16b); y1_0H_8x16b = _mm_adds_epi16(round_8x16b, y1_0H_8x16b); y1_1L_8x16b = _mm_adds_epi16(round_8x16b, y1_1L_8x16b); y1_1H_8x16b = _mm_adds_epi16(round_8x16b, y1_1H_8x16b); y1_0L_8x16b = _mm_srai_epi16(y1_0L_8x16b, shft); y1_0H_8x16b = _mm_srai_epi16(y1_0H_8x16b, shft); y1_1L_8x16b = _mm_srai_epi16(y1_1L_8x16b, shft); y1_1H_8x16b = _mm_srai_epi16(y1_1H_8x16b, shft); y1_0L_8x16b = _mm_adds_epi16(ofst_8x16b, y1_0L_8x16b); y1_0H_8x16b = _mm_adds_epi16(ofst_8x16b, y1_0H_8x16b); y1_1L_8x16b = _mm_adds_epi16(ofst_8x16b, y1_1L_8x16b); y1_1H_8x16b = _mm_adds_epi16(ofst_8x16b, y1_1H_8x16b); y1_0_16x8b = _mm_packus_epi16(y1_0L_8x16b, y1_0H_8x16b); y1_1_16x8b = _mm_packus_epi16(y1_1L_8x16b, y1_1H_8x16b); _mm_storeu_si128((__m128i *)pu1_dst, y1_0_16x8b); _mm_storeu_si128((__m128i *)(pu1_dst + dst_strd), y1_1_16x8b); ht -= 2; pu1_src1 += src_strd1 << 1; pu1_src2 += src_strd2 << 1; pu1_dst += dst_strd << 1; } while(ht > 0); } } /*****************************************************************************/ /* */ /* Function Name : ih264_weighted_bi_pred_chroma_sse42 */ /* */ /* Description : This function performs the weighted biprediction as */ /* described in sec 8.4.2.3.2 titled "Weighted sample */ /* prediction process" for chroma. The function gets two */ /* ht x wd blocks, weights them, adds them, rounds off the */ /* sum, offsets it, saturates it to unsigned 8-bit and */ /* stores it in the destination block. (ht,wd) can be */ /* (2,2), (4,2), (2,4), (4,4), (8,4), (4,8) or (8,8). */ /* */ /* Inputs : pu1_src1 - Pointer to source 1 */ /* pu1_src2 - Pointer to source 2 */ /* pu1_dst - Pointer to destination */ /* src_strd1 - stride for source 1 */ /* src_strd2 - stride for source 2 */ /* dst_strd2 - stride for destination */ /* log_wd - number of bits to be rounded off */ /* wt1 - weight values for u and v in source 1 */ /* wt2 - weight values for u and v in source 2 */ /* ofst1 - offset value for u and v in source 1 */ /* ofst2 - offset value for u and v in source 2 */ /* ht - height of the block */ /* wd - width of the block */ /* */ /* Issues : None */ /* */ /* Revision History: */ /* */ /* DD MM YYYY Author(s) Changes */ /* 04 02 2015 Kaushik Initial Version */ /* Senthoor */ /* */ /*****************************************************************************/ void ih264_weighted_bi_pred_chroma_sse42(UWORD8 *pu1_src1, UWORD8 *pu1_src2, UWORD8 *pu1_dst, WORD32 src_strd1, WORD32 src_strd2, WORD32 dst_strd, WORD32 log_wd, WORD32 wt1, WORD32 wt2, WORD32 ofst1, WORD32 ofst2, WORD32 ht, WORD32 wd) { __m128i y1_0_16x8b, y1_1_16x8b; __m128i y2_0_16x8b, y2_1_16x8b; __m128i wt1_8x16b, wt2_8x16b; __m128i ofst_8x16b, round_8x16b; WORD32 ofst1_u, ofst2_u, ofst_u; WORD32 ofst1_v, ofst2_v, ofst_v; WORD32 round_val, shft, ofst_val; round_val = 1 << log_wd; shft = log_wd + 1; ofst1_u = (WORD8)(ofst1 & 0xff); ofst1_v = (WORD8)(ofst1 >> 8); ofst2_u = (WORD8)(ofst2 & 0xff); ofst2_v = (WORD8)(ofst2 >> 8); wt1_8x16b = _mm_set1_epi32(wt1); wt2_8x16b = _mm_set1_epi32(wt2); ofst_u = (ofst1_u + ofst2_u + 1) >> 1; ofst_v = (ofst1_v + ofst2_v + 1) >> 1; ofst_val = (ofst_u & 0xffff) | (ofst_v << 16); round_8x16b = _mm_set1_epi16(round_val); ofst_8x16b = _mm_set1_epi32(ofst_val); if(wd == 2) { __m128i y1_0_8x16b, y2_0_8x16b; do { y1_0_16x8b = _mm_loadl_epi64((__m128i *)pu1_src1); y1_1_16x8b = _mm_loadl_epi64((__m128i *)(pu1_src1 + src_strd1)); y2_0_16x8b = _mm_loadl_epi64((__m128i *)pu1_src2); y2_1_16x8b = _mm_loadl_epi64((__m128i *)(pu1_src2 + src_strd2)); y1_0_16x8b = _mm_unpacklo_epi32(y1_0_16x8b, y1_1_16x8b); y2_0_16x8b = _mm_unpacklo_epi32(y2_0_16x8b, y2_1_16x8b); y1_0_8x16b = _mm_cvtepu8_epi16(y1_0_16x8b); y2_0_8x16b = _mm_cvtepu8_epi16(y2_0_16x8b); y1_0_8x16b = _mm_mullo_epi16(y1_0_8x16b, wt1_8x16b); y2_0_8x16b = _mm_mullo_epi16(y2_0_8x16b, wt2_8x16b); y1_0_8x16b = _mm_adds_epi16(y1_0_8x16b, y2_0_8x16b); y1_0_8x16b = _mm_adds_epi16(round_8x16b, y1_0_8x16b); y1_0_8x16b = _mm_srai_epi16(y1_0_8x16b, shft); y1_0_8x16b = _mm_adds_epi16(ofst_8x16b, y1_0_8x16b); y1_0_16x8b = _mm_packus_epi16(y1_0_8x16b, y1_0_8x16b); y1_1_16x8b = _mm_srli_si128(y1_0_16x8b, 4); *((WORD32 *)(pu1_dst)) = _mm_cvtsi128_si32(y1_0_16x8b); *((WORD32 *)(pu1_dst + dst_strd)) = _mm_cvtsi128_si32(y1_1_16x8b); ht -= 2; pu1_src1 += src_strd1 << 1; pu1_src2 += src_strd2 << 1; pu1_dst += dst_strd << 1; } while(ht > 0); } else if(wd == 4) { __m128i y1_0_8x16b, y1_1_8x16b; __m128i y2_0_8x16b, y2_1_8x16b; do { y1_0_16x8b = _mm_loadl_epi64((__m128i *)pu1_src1); y1_1_16x8b = _mm_loadl_epi64((__m128i *)(pu1_src1 + src_strd1)); y2_0_16x8b = _mm_loadl_epi64((__m128i *)pu1_src2); y2_1_16x8b = _mm_loadl_epi64((__m128i *)(pu1_src2 + src_strd2)); y1_0_8x16b = _mm_cvtepu8_epi16(y1_0_16x8b); y1_1_8x16b = _mm_cvtepu8_epi16(y1_1_16x8b); y2_0_8x16b = _mm_cvtepu8_epi16(y2_0_16x8b); y2_1_8x16b = _mm_cvtepu8_epi16(y2_1_16x8b); y1_0_8x16b = _mm_mullo_epi16(y1_0_8x16b, wt1_8x16b); y2_0_8x16b = _mm_mullo_epi16(y2_0_8x16b, wt2_8x16b); y1_1_8x16b = _mm_mullo_epi16(y1_1_8x16b, wt1_8x16b); y2_1_8x16b = _mm_mullo_epi16(y2_1_8x16b, wt2_8x16b); y1_0_8x16b = _mm_adds_epi16(y1_0_8x16b, y2_0_8x16b); y1_1_8x16b = _mm_adds_epi16(y1_1_8x16b, y2_1_8x16b); y1_0_8x16b = _mm_adds_epi16(round_8x16b, y1_0_8x16b); y1_1_8x16b = _mm_adds_epi16(round_8x16b, y1_1_8x16b); y1_0_8x16b = _mm_srai_epi16(y1_0_8x16b, shft); y1_1_8x16b = _mm_srai_epi16(y1_1_8x16b, shft); y1_0_8x16b = _mm_adds_epi16(ofst_8x16b, y1_0_8x16b); y1_1_8x16b = _mm_adds_epi16(ofst_8x16b, y1_1_8x16b); y1_0_16x8b = _mm_packus_epi16(y1_0_8x16b, y1_1_8x16b); y1_1_16x8b = _mm_srli_si128(y1_0_16x8b, 8); _mm_storel_epi64((__m128i *)pu1_dst, y1_0_16x8b); _mm_storel_epi64((__m128i *)(pu1_dst + dst_strd), y1_1_16x8b); ht -= 2; pu1_src1 += src_strd1 << 1; pu1_src2 += src_strd2 << 1; pu1_dst += dst_strd << 1; } while(ht > 0); } else // wd == 8 { __m128i y1_0L_8x16b, y1_0H_8x16b, y1_1L_8x16b, y1_1H_8x16b; __m128i y2_0L_8x16b, y2_0H_8x16b, y2_1L_8x16b, y2_1H_8x16b; __m128i zero_16x8b; zero_16x8b = _mm_set1_epi8(0); do { y1_0_16x8b = _mm_loadu_si128((__m128i *)pu1_src1); y1_1_16x8b = _mm_loadu_si128((__m128i *)(pu1_src1 + src_strd1)); y2_0_16x8b = _mm_loadu_si128((__m128i *)pu1_src2); y2_1_16x8b = _mm_loadu_si128((__m128i *)(pu1_src2 + src_strd2)); y1_0L_8x16b = _mm_cvtepu8_epi16(y1_0_16x8b); y1_0H_8x16b = _mm_unpackhi_epi8(y1_0_16x8b, zero_16x8b); y1_1L_8x16b = _mm_cvtepu8_epi16(y1_1_16x8b); y1_1H_8x16b = _mm_unpackhi_epi8(y1_1_16x8b, zero_16x8b); y2_0L_8x16b = _mm_cvtepu8_epi16(y2_0_16x8b); y2_0H_8x16b = _mm_unpackhi_epi8(y2_0_16x8b, zero_16x8b); y2_1L_8x16b = _mm_cvtepu8_epi16(y2_1_16x8b); y2_1H_8x16b = _mm_unpackhi_epi8(y2_1_16x8b, zero_16x8b); y1_0L_8x16b = _mm_mullo_epi16(y1_0L_8x16b, wt1_8x16b); y1_0H_8x16b = _mm_mullo_epi16(y1_0H_8x16b, wt1_8x16b); y1_1L_8x16b = _mm_mullo_epi16(y1_1L_8x16b, wt1_8x16b); y1_1H_8x16b = _mm_mullo_epi16(y1_1H_8x16b, wt1_8x16b); y2_0L_8x16b = _mm_mullo_epi16(y2_0L_8x16b, wt2_8x16b); y2_0H_8x16b = _mm_mullo_epi16(y2_0H_8x16b, wt2_8x16b); y2_1L_8x16b = _mm_mullo_epi16(y2_1L_8x16b, wt2_8x16b); y2_1H_8x16b = _mm_mullo_epi16(y2_1H_8x16b, wt2_8x16b); y1_0L_8x16b = _mm_adds_epi16(y1_0L_8x16b, y2_0L_8x16b); y1_0H_8x16b = _mm_adds_epi16(y1_0H_8x16b, y2_0H_8x16b); y1_1L_8x16b = _mm_adds_epi16(y1_1L_8x16b, y2_1L_8x16b); y1_1H_8x16b = _mm_adds_epi16(y1_1H_8x16b, y2_1H_8x16b); y1_0L_8x16b = _mm_adds_epi16(round_8x16b, y1_0L_8x16b); y1_0H_8x16b = _mm_adds_epi16(round_8x16b, y1_0H_8x16b); y1_1L_8x16b = _mm_adds_epi16(round_8x16b, y1_1L_8x16b); y1_1H_8x16b = _mm_adds_epi16(round_8x16b, y1_1H_8x16b); y1_0L_8x16b = _mm_srai_epi16(y1_0L_8x16b, shft); y1_0H_8x16b = _mm_srai_epi16(y1_0H_8x16b, shft); y1_1L_8x16b = _mm_srai_epi16(y1_1L_8x16b, shft); y1_1H_8x16b = _mm_srai_epi16(y1_1H_8x16b, shft); y1_0L_8x16b = _mm_adds_epi16(ofst_8x16b, y1_0L_8x16b); y1_0H_8x16b = _mm_adds_epi16(ofst_8x16b, y1_0H_8x16b); y1_1L_8x16b = _mm_adds_epi16(ofst_8x16b, y1_1L_8x16b); y1_1H_8x16b = _mm_adds_epi16(ofst_8x16b, y1_1H_8x16b); y1_0_16x8b = _mm_packus_epi16(y1_0L_8x16b, y1_0H_8x16b); y1_1_16x8b = _mm_packus_epi16(y1_1L_8x16b, y1_1H_8x16b); _mm_storeu_si128((__m128i *)pu1_dst, y1_0_16x8b); _mm_storeu_si128((__m128i *)(pu1_dst + dst_strd), y1_1_16x8b); ht -= 2; pu1_src1 += src_strd1 << 1; pu1_src2 += src_strd2 << 1; pu1_dst += dst_strd << 1; } while(ht > 0); } }