From 04fbb48a09fdff77c64386c699d213f04a22bdb3 Mon Sep 17 00:00:00 2001 From: Ari Lemmetti Date: Wed, 24 Feb 2016 17:02:50 +0200 Subject: [PATCH] Add strategy for avx2. Copy generic functions there. --- build/kvazaar_lib/kvazaar_lib.vcxproj | 7 + build/kvazaar_lib/kvazaar_lib.vcxproj.filters | 6 + src/Makefile.am | 5 +- src/strategies/avx2/sao-avx2.c | 187 ++++++++++++++++++ src/strategies/avx2/sao-avx2.h | 33 ++++ src/strategies/strategies-sao.c | 5 + 6 files changed, 242 insertions(+), 1 deletion(-) create mode 100644 src/strategies/avx2/sao-avx2.c create mode 100644 src/strategies/avx2/sao-avx2.h diff --git a/build/kvazaar_lib/kvazaar_lib.vcxproj b/build/kvazaar_lib/kvazaar_lib.vcxproj index 9495386d..384ee727 100644 --- a/build/kvazaar_lib/kvazaar_lib.vcxproj +++ b/build/kvazaar_lib/kvazaar_lib.vcxproj @@ -163,6 +163,12 @@ AdvancedVectorExtensions2 AdvancedVectorExtensions2 + + AdvancedVectorExtensions2 + AdvancedVectorExtensions2 + AdvancedVectorExtensions2 + AdvancedVectorExtensions2 + @@ -217,6 +223,7 @@ + diff --git a/build/kvazaar_lib/kvazaar_lib.vcxproj.filters b/build/kvazaar_lib/kvazaar_lib.vcxproj.filters index d03890dc..f53b9391 100644 --- a/build/kvazaar_lib/kvazaar_lib.vcxproj.filters +++ b/build/kvazaar_lib/kvazaar_lib.vcxproj.filters @@ -213,6 +213,9 @@ Optimization\strategies\generic + + Optimization\strategies\avx2 + @@ -395,6 +398,9 @@ Optimization\strategies\generic + + Optimization\strategies\avx2 + diff --git a/src/Makefile.am b/src/Makefile.am index de00bf89..3c45a482 100644 --- a/src/Makefile.am +++ b/src/Makefile.am @@ -154,7 +154,10 @@ libavx2_la_SOURCES = \ strategies/avx2/picture-avx2.c \ strategies/avx2/picture-avx2.h \ strategies/avx2/quant-avx2.c \ - strategies/avx2/quant-avx2.h + strategies/avx2/quant-avx2.h \ + strategies/avx2/sao-avx2.c \ + strategies/avx2/sao-avx2.h + libsse2_la_SOURCES = \ strategies/sse2/picture-sse2.c \ diff --git a/src/strategies/avx2/sao-avx2.c b/src/strategies/avx2/sao-avx2.c new file mode 100644 index 00000000..3adfd2e8 --- /dev/null +++ b/src/strategies/avx2/sao-avx2.c @@ -0,0 +1,187 @@ +/***************************************************************************** + * This file is part of Kvazaar HEVC encoder. + * + * Copyright (C) 2013-2015 Tampere University of Technology and others (see + * COPYING file). + * + * Kvazaar is free software: you can redistribute it and/or modify it under + * the terms of the GNU Lesser General Public License as published by the + * Free Software Foundation; either version 2.1 of the License, or (at your + * option) any later version. + * + * Kvazaar is distributed in the hope that it will be useful, but WITHOUT ANY + * WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS + * FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public License for + * more details. + * + * You should have received a copy of the GNU General Public License along + * with Kvazaar. If not, see . + ****************************************************************************/ + +#include + +#include "sao-avx2.h" +#include "sao.h" +#include "strategyselector.h" +#include "rdo.h" + +#if COMPILE_INTEL_AVX2 +#include + +// Mapping of edge_idx values to eo-classes. +static int sao_calc_eo_cat(kvz_pixel a, kvz_pixel b, kvz_pixel c) +{ + // Mapping relationships between a, b and c to eo_idx. + static const int sao_eo_idx_to_eo_category[] = { 1, 2, 0, 3, 4 }; + + int eo_idx = 2 + SIGN3((int)c - (int)a) + SIGN3((int)c - (int)b); + + return sao_eo_idx_to_eo_category[eo_idx]; +} + + +int kvz_sao_edge_ddistortion_avx2(const kvz_pixel *orig_data, const kvz_pixel *rec_data, + int block_width, int block_height, + int eo_class, int offsets[NUM_SAO_EDGE_CATEGORIES]) +{ + int y, x; + int sum = 0; + vector2d_t a_ofs = g_sao_edge_offsets[eo_class][0]; + vector2d_t b_ofs = g_sao_edge_offsets[eo_class][1]; + + for (y = 1; y < block_height - 1; ++y) { + for (x = 1; x < block_width - 1; ++x) { + const kvz_pixel *c_data = &rec_data[y * block_width + x]; + kvz_pixel a = c_data[a_ofs.y * block_width + a_ofs.x]; + kvz_pixel c = c_data[0]; + kvz_pixel b = c_data[b_ofs.y * block_width + b_ofs.x]; + + int offset = offsets[sao_calc_eo_cat(a, b, c)]; + + if (offset != 0) { + int diff = orig_data[y * block_width + x] - c; + // Offset is applied to reconstruction, so it is subtracted from diff. + sum += (diff - offset) * (diff - offset) - diff * diff; + } + } + } + + return sum; +} + + +/** + * \param orig_data Original pixel data. 64x64 for luma, 32x32 for chroma. + * \param rec_data Reconstructed pixel data. 64x64 for luma, 32x32 for chroma. + * \param dir_offsets + * \param is_chroma 0 for luma, 1 for chroma. Indicates + */ +void kvz_calc_sao_edge_dir_avx2(const kvz_pixel *orig_data, const kvz_pixel *rec_data, + int eo_class, int block_width, int block_height, + int cat_sum_cnt[2][NUM_SAO_EDGE_CATEGORIES]) +{ + int y, x; + vector2d_t a_ofs = g_sao_edge_offsets[eo_class][0]; + vector2d_t b_ofs = g_sao_edge_offsets[eo_class][1]; + // Arrays orig_data and rec_data are quarter size for chroma. + + // Don't sample the edge pixels because this function doesn't have access to + // their neighbours. + for (y = 1; y < block_height - 1; ++y) { + for (x = 1; x < block_width - 1; ++x) { + const kvz_pixel *c_data = &rec_data[y * block_width + x]; + kvz_pixel a = c_data[a_ofs.y * block_width + a_ofs.x]; + kvz_pixel c = c_data[0]; + kvz_pixel b = c_data[b_ofs.y * block_width + b_ofs.x]; + + int eo_cat = sao_calc_eo_cat(a, b, c); + + cat_sum_cnt[0][eo_cat] += orig_data[y * block_width + x] - c; + cat_sum_cnt[1][eo_cat] += 1; + } + } +} + + +void kvz_sao_reconstruct_color_avx2(const encoder_control_t * const encoder, + const kvz_pixel *rec_data, kvz_pixel *new_rec_data, + const sao_info_t *sao, + int stride, int new_stride, + int block_width, int block_height, + color_t color_i) +{ + int y, x; + // Arrays orig_data and rec_data are quarter size for chroma. + int offset_v = color_i == COLOR_V ? 5 : 0; + + if(sao->type == SAO_TYPE_BAND) { + int offsets[1<eo_class][0]; + vector2d_t b_ofs = g_sao_edge_offsets[sao->eo_class][1]; + const kvz_pixel *c_data = &rec_data[y * stride + x]; + kvz_pixel *new_data = &new_rec_data[y * new_stride + x]; + kvz_pixel a = c_data[a_ofs.y * stride + a_ofs.x]; + kvz_pixel c = c_data[0]; + kvz_pixel b = c_data[b_ofs.y * stride + b_ofs.x]; + + int eo_cat = sao_calc_eo_cat(a, b, c); + + new_data[0] = (kvz_pixel)CLIP(0, (1 << KVZ_BIT_DEPTH) - 1, c_data[0] + sao->offsets[eo_cat + offset_v]); + } + } + } +} + + +int kvz_sao_band_ddistortion_avx2(const encoder_state_t * const state, const kvz_pixel *orig_data, const kvz_pixel *rec_data, + int block_width, int block_height, + int band_pos, int sao_bands[4]) +{ + int y, x; + int shift = state->encoder_control->bitdepth-5; + int sum = 0; + + for (y = 0; y < block_height; ++y) { + for (x = 0; x < block_width; ++x) { + int band = (rec_data[y * block_width + x] >> shift) - band_pos; + int offset = 0; + if (band >= 0 && band < 4) { + offset = sao_bands[band]; + } + if (offset != 0) { + int diff = orig_data[y * block_width + x] - rec_data[y * block_width + x]; + // Offset is applied to reconstruction, so it is subtracted from diff. + sum += (diff - offset) * (diff - offset) - diff * diff; + } + } + } + + return sum; +} + +#endif //COMPILE_INTEL_AVX2 + +int kvz_strategy_register_sao_avx2(void* opaque, uint8_t bitdepth) +{ + bool success = true; +#if COMPILE_INTEL_AVX2 + if (bitdepth == 8) { + success &= kvz_strategyselector_register(opaque, "sao_edge_ddistortion", "avx2", 40, &kvz_sao_edge_ddistortion_avx2); + success &= kvz_strategyselector_register(opaque, "calc_sao_edge_dir", "avx2", 40, &kvz_calc_sao_edge_dir_avx2); + success &= kvz_strategyselector_register(opaque, "sao_reconstruct_color", "avx2", 40, &kvz_sao_reconstruct_color_avx2); + success &= kvz_strategyselector_register(opaque, "sao_band_ddistortion", "avx2", 40, &kvz_sao_band_ddistortion_avx2); + } +#endif //COMPILE_INTEL_AVX2 + return success; +} diff --git a/src/strategies/avx2/sao-avx2.h b/src/strategies/avx2/sao-avx2.h new file mode 100644 index 00000000..b9ecca35 --- /dev/null +++ b/src/strategies/avx2/sao-avx2.h @@ -0,0 +1,33 @@ +#ifndef STRATEGIES_SAO_AVX2_H_ +#define STRATEGIES_SAO_AVX2_H_ +/***************************************************************************** + * This file is part of Kvazaar HEVC encoder. + * + * Copyright (C) 2013-2015 Tampere University of Technology and others (see + * COPYING file). + * + * Kvazaar is free software: you can redistribute it and/or modify it under + * the terms of the GNU Lesser General Public License as published by the + * Free Software Foundation; either version 2.1 of the License, or (at your + * option) any later version. + * + * Kvazaar is distributed in the hope that it will be useful, but WITHOUT ANY + * WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS + * FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public License for + * more details. + * + * You should have received a copy of the GNU General Public License along + * with Kvazaar. If not, see . + ****************************************************************************/ + +/** + * \ingroup Optimization + * \file + * AVX2 implementations of optimized functions. + */ + +#include "global.h" + +int kvz_strategy_register_sao_avx2(void* opaque, uint8_t bitdepth); + +#endif //STRATEGIES_SAO_AVX2_H_ diff --git a/src/strategies/strategies-sao.c b/src/strategies/strategies-sao.c index c45e7ff7..1660d711 100644 --- a/src/strategies/strategies-sao.c +++ b/src/strategies/strategies-sao.c @@ -29,6 +29,7 @@ sao_band_ddistortion_func * kvz_sao_band_ddistortion; // Headers for platform optimizations. #include "generic/sao-generic.h" +#include "avx2/sao-avx2.h" int kvz_strategy_register_sao(void* opaque, uint8_t bitdepth) { @@ -36,5 +37,9 @@ int kvz_strategy_register_sao(void* opaque, uint8_t bitdepth) { success &= kvz_strategy_register_sao_generic(opaque, bitdepth); + if (kvz_g_hardware_flags.intel_flags.avx2) { + success &= kvz_strategy_register_sao_avx2(opaque, bitdepth); + } + return success; } \ No newline at end of file