/* * Copyright (c) 2018, Alliance for Open Media. All rights reserved. * * This source code is subject to the terms of the BSD 2 Clause License and * the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License * was not distributed with this source code in the LICENSE file, you can * obtain it at www.aomedia.org/license/software. If the Alliance for Open * Media Patent License 1.0 was not distributed with this source code in the * PATENTS file, you can obtain it at www.aomedia.org/license/patent. */ #include <assert.h> #include <immintrin.h> /*AVX2*/ #include "config/aom_config.h" #include "config/av1_rtcd.h" #include "av1/common/av1_txfm.h" #include "av1/encoder/av1_fwd_txfm1d_cfg.h" #include "aom_dsp/txfm_common.h" #include "aom_ports/mem.h" #include "aom_dsp/x86/txfm_common_sse2.h" #include "aom_dsp/x86/txfm_common_avx2.h" static inline void load_buffer_8x8_avx2(const int16_t *input, __m256i *out, int stride, int flipud, int fliplr, int shift) { … } static inline void col_txfm_8x8_rounding(__m256i *in, int shift) { … } static inline void load_buffer_8x16_avx2(const int16_t *input, __m256i *out, int stride, int flipud, int fliplr, int shift) { … } static inline void load_buffer_16xn_avx2(const int16_t *input, __m256i *out, int stride, int height, int outstride, int flipud, int fliplr) { … } static void fwd_txfm_transpose_8x8_avx2(const __m256i *in, __m256i *out, const int instride, const int outstride) { … } static inline void round_shift_32_8xn_avx2(__m256i *in, int size, int bit, int stride) { … } static inline void store_buffer_avx2(const __m256i *const in, int32_t *out, const int stride, const int out_size) { … } static inline void fwd_txfm_transpose_16x16_avx2(const __m256i *in, __m256i *out) { … } static inline __m256i av1_half_btf_avx2(const __m256i *w0, const __m256i *n0, const __m256i *w1, const __m256i *n1, const __m256i *rounding, int bit) { … } #define btf_32_avx2_type0(w0, w1, in0, in1, out0, out1, bit) … #define btf_32_type0_avx2_new(ww0, ww1, in0, in1, out0, out1, r, bit) … transform_1d_avx2; static void fdct8_avx2(__m256i *in, __m256i *out, const int8_t bit, const int col_num, const int outstride) { … } static void fadst8_avx2(__m256i *in, __m256i *out, const int8_t bit, const int col_num, const int outstirde) { … } static void idtx8_avx2(__m256i *in, __m256i *out, const int8_t bit, int col_num, int outstride) { … } void av1_fwd_txfm2d_8x8_avx2(const int16_t *input, int32_t *coeff, int stride, TX_TYPE tx_type, int bd) { … } static void fdct16_avx2(__m256i *in, __m256i *out, const int8_t bit, const int col_num, const int outstride) { … } static void fadst16_avx2(__m256i *in, __m256i *out, const int8_t bit, const int num_cols, const int outstride) { … } static void idtx16_avx2(__m256i *in, __m256i *out, const int8_t bit, int col_num, const int outstride) { … } static const transform_1d_avx2 col_highbd_txfm8x16_arr[TX_TYPES] = …; static const transform_1d_avx2 row_highbd_txfm8x8_arr[TX_TYPES] = …; void av1_fwd_txfm2d_8x16_avx2(const int16_t *input, int32_t *coeff, int stride, TX_TYPE tx_type, int bd) { … } static const transform_1d_avx2 col_highbd_txfm8x8_arr[TX_TYPES] = …; static const transform_1d_avx2 row_highbd_txfm8x16_arr[TX_TYPES] = …; void av1_fwd_txfm2d_16x8_avx2(const int16_t *input, int32_t *coeff, int stride, TX_TYPE tx_type, int bd) { … } void av1_fwd_txfm2d_16x16_avx2(const int16_t *input, int32_t *coeff, int stride, TX_TYPE tx_type, int bd) { … } static inline void fdct32_avx2(__m256i *input, __m256i *output, const int8_t cos_bit, const int instride, const int outstride) { … } static inline void idtx32x32_avx2(__m256i *input, __m256i *output, const int8_t cos_bit, int instride, int outstride) { … } static const transform_1d_avx2 col_txfm8x32_arr[TX_TYPES] = …; static const transform_1d_avx2 row_txfm8x32_arr[TX_TYPES] = …; void av1_fwd_txfm2d_32x32_avx2(const int16_t *input, int32_t *output, int stride, TX_TYPE tx_type, int bd) { … } static inline void fdct64_stage2_avx2(__m256i *x1, __m256i *x2, __m256i *cospi_m32, __m256i *cospi_p32, const __m256i *__rounding, int8_t cos_bit) { … } static inline void fdct64_stage3_avx2(__m256i *x2, __m256i *x3, __m256i *cospi_m32, __m256i *cospi_p32, const __m256i *__rounding, int8_t cos_bit) { … } static inline void fdct64_stage4_avx2(__m256i *x3, __m256i *x4, __m256i *cospi_m32, __m256i *cospi_p32, __m256i *cospi_m16, __m256i *cospi_p48, __m256i *cospi_m48, const __m256i *__rounding, int8_t cos_bit) { … } static inline void fdct64_stage5_avx2(__m256i *x4, __m256i *x5, __m256i *cospi_m32, __m256i *cospi_p32, __m256i *cospi_m16, __m256i *cospi_p48, __m256i *cospi_m48, const __m256i *__rounding, int8_t cos_bit) { … } static inline void fdct64_stage6_avx2( __m256i *x5, __m256i *x6, __m256i *cospi_p16, __m256i *cospi_p32, __m256i *cospi_m16, __m256i *cospi_p48, __m256i *cospi_m48, __m256i *cospi_m08, __m256i *cospi_p56, __m256i *cospi_m56, __m256i *cospi_m40, __m256i *cospi_p24, __m256i *cospi_m24, const __m256i *__rounding, int8_t cos_bit) { … } static inline void fdct64_stage7_avx2(__m256i *x6, __m256i *x7, __m256i *cospi_p08, __m256i *cospi_p56, __m256i *cospi_p40, __m256i *cospi_p24, __m256i *cospi_m08, __m256i *cospi_m56, __m256i *cospi_m40, __m256i *cospi_m24, const __m256i *__rounding, int8_t cos_bit) { … } static inline void fdct64_stage8_avx2(__m256i *x7, __m256i *x8, const int32_t *cospi, const __m256i *__rounding, int8_t cos_bit) { … } static inline void fdct64_stage9_avx2(__m256i *x8, __m256i *x9, const int32_t *cospi, const __m256i *__rounding, int8_t cos_bit) { … } static inline void fdct64_stage10_avx2(__m256i *x9, __m256i *x10, const int32_t *cospi, const __m256i *__rounding, int8_t cos_bit) { … } static void fdct64_avx2(__m256i *input, __m256i *output, int8_t cos_bit, const int instride, const int outstride) { … } void av1_fwd_txfm2d_64x64_avx2(const int16_t *input, int32_t *output, int stride, TX_TYPE tx_type, int bd) { … }
Codebase Browser
chromium