highbd_idct32x32_add_sse2.c | Explore in Territory

/*
 *  Copyright (c) 2015 The WebM project authors. All Rights Reserved.
 *
 *  Use of this source code is governed by a BSD-style license
 *  that can be found in the LICENSE file in the root of the source
 *  tree. An additional intellectual property rights grant can be found
 *  in the file PATENTS.  All contributing project authors may
 *  be found in the AUTHORS file in the root of the source tree.
 */

#include "./vpx_dsp_rtcd.h"
#include "vpx_dsp/x86/highbd_inv_txfm_sse2.h"
#include "vpx_dsp/x86/inv_txfm_sse2.h"
#include "vpx_dsp/x86/transpose_sse2.h"
#include "vpx_dsp/x86/txfm_common_sse2.h"

static INLINE void highbd_idct32_4x32_quarter_2_stage_4_to_6(
    __m128i *const step1 /*step1[16]*/, __m128i *const out /*out[16]*/) { … }

static INLINE void highbd_idct32_4x32_quarter_3_4_stage_4_to_7(
    __m128i *const step1 /*step1[32]*/, __m128i *const out /*out[32]*/) { … }

// Group the coefficient calculation into smaller functions to prevent stack
// spillover in 32x32 idct optimizations:
// quarter_1: 0-7
// quarter_2: 8-15
// quarter_3_4: 16-23, 24-31

// For each 4x32 block __m128i in[32],
// Input with index, 0, 4, 8, 12, 16, 20, 24, 28
// output pixels: 0-7 in __m128i out[32]
static INLINE void highbd_idct32_1024_4x32_quarter_1(
    const __m128i *const in /*in[32]*/, __m128i *const out /*out[8]*/) { … }

// For each 4x32 block __m128i in[32],
// Input with index, 2, 6, 10, 14, 18, 22, 26, 30
// output pixels: 8-15 in __m128i out[32]
static INLINE void highbd_idct32_1024_4x32_quarter_2(
    const __m128i *in /*in[32]*/, __m128i *out /*out[16]*/) { … }

static INLINE void highbd_idct32_1024_4x32_quarter_1_2(
    const __m128i *const in /*in[32]*/, __m128i *const out /*out[32]*/) { … }

// For each 4x32 block __m128i in[32],
// Input with odd index,
// 1, 3, 5, 7, 9, 11, 13, 15, 17, 19, 21, 23, 25, 27, 29, 31
// output pixels: 16-23, 24-31 in __m128i out[32]
static INLINE void highbd_idct32_1024_4x32_quarter_3_4(
    const __m128i *const in /*in[32]*/, __m128i *const out /*out[32]*/) { … }

static void highbd_idct32_1024_4x32(__m128i *const io /*io[32]*/) { … }

void vpx_highbd_idct32x32_1024_add_sse2(const tran_low_t *input, uint16_t *dest,
                                        int stride, int bd) { … }

// -----------------------------------------------------------------------------

// For each 4x32 block __m128i in[32],
// Input with index, 0, 4, 8, 12
// output pixels: 0-7 in __m128i out[32]
static INLINE void highbd_idct32_135_4x32_quarter_1(
    const __m128i *const in /*in[32]*/, __m128i *const out /*out[8]*/) { … }

// For each 4x32 block __m128i in[32],
// Input with index, 2, 6, 10, 14
// output pixels: 8-15 in __m128i out[32]
static INLINE void highbd_idct32_135_4x32_quarter_2(
    const __m128i *in /*in[32]*/, __m128i *out /*out[16]*/) { … }

static INLINE void highbd_idct32_135_4x32_quarter_1_2(
    const __m128i *const in /*in[32]*/, __m128i *const out /*out[32]*/) { … }

// For each 4x32 block __m128i in[32],
// Input with odd index,
// 1, 3, 5, 7, 9, 11, 13, 15
// output pixels: 16-23, 24-31 in __m128i out[32]
static INLINE void highbd_idct32_135_4x32_quarter_3_4(
    const __m128i *const in /*in[32]*/, __m128i *const out /*out[32]*/) { … }

static void highbd_idct32_135_4x32(__m128i *const io /*io[32]*/) { … }

void vpx_highbd_idct32x32_135_add_sse2(const tran_low_t *input, uint16_t *dest,
                                       int stride, int bd) { … }

// -----------------------------------------------------------------------------

// For each 4x32 block __m128i in[32],
// Input with index, 0, 4
// output pixels: 0-7 in __m128i out[32]
static INLINE void highbd_idct32_34_4x32_quarter_1(
    const __m128i *const in /*in[32]*/, __m128i *const out /*out[8]*/) { … }

// For each 4x32 block __m128i in[32],
// Input with index, 2, 6
// output pixels: 8-15 in __m128i out[32]
static INLINE void highbd_idct32_34_4x32_quarter_2(const __m128i *in /*in[32]*/,
                                                   __m128i *out /*out[16]*/) { … }

static INLINE void highbd_idct32_34_4x32_quarter_1_2(
    const __m128i *const in /*in[32]*/, __m128i *const out /*out[32]*/) { … }

// For each 4x32 block __m128i in[32],
// Input with odd index,
// 1, 3, 5, 7
// output pixels: 16-23, 24-31 in __m128i out[32]
static INLINE void highbd_idct32_34_4x32_quarter_3_4(
    const __m128i *const in /*in[32]*/, __m128i *const out /*out[32]*/) { … }

static void highbd_idct32_34_4x32(__m128i *const io /*io[32]*/) { … }

void vpx_highbd_idct32x32_34_add_sse2(const tran_low_t *input, uint16_t *dest,
                                      int stride, int bd) { … }

void vpx_highbd_idct32x32_1_add_sse2(const tran_low_t *input, uint16_t *dest,
                                     int stride, int bd) { … }
chromium/third_party/libvpx/source/libvpx/vpx_dsp/x86/highbd_idct32x32_add_sse2.c
