chromium/third_party/libaom/source/libaom/aom_dsp/sad.c

/*
 * Copyright (c) 2016, 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 <stdlib.h>

#include "config/aom_config.h"
#include "config/aom_dsp_rtcd.h"

#include "aom/aom_integer.h"
#include "aom_ports/mem.h"
#include "aom_dsp/blend.h"

/* Sum the difference between every corresponding element of the buffers. */
static inline unsigned int sad(const uint8_t *a, int a_stride, const uint8_t *b,
                               int b_stride, int width, int height) { … }

#define SADMXN(m, n) …

// Calculate sad against 4 reference locations and store each in sad_array
#define SAD_MXNX4D(m, n) …
// Call SIMD version of aom_sad_mxnx4d if the 3d version is unavailable.
#define SAD_MXNX3D(m, n) …

// 128x128
SADMXN(…)
SAD_MXNX4D(…)
SAD_MXNX3D(…)

// 128x64
SADMXN(…)
SAD_MXNX4D(…)
SAD_MXNX3D(…)

// 64x128
SADMXN(…)
SAD_MXNX4D(…)
SAD_MXNX3D(…)

// 64x64
SADMXN(…)
SAD_MXNX4D(…)
SAD_MXNX3D(…)

// 64x32
SADMXN(…)
SAD_MXNX4D(…)
SAD_MXNX3D(…)

// 32x64
SADMXN(…)
SAD_MXNX4D(…)
SAD_MXNX3D(…)

// 32x32
SADMXN(…)
SAD_MXNX4D(…)
SAD_MXNX3D(…)

// 32x16
SADMXN(…)
SAD_MXNX4D(…)
SAD_MXNX3D(…)

// 16x32
SADMXN(…)
SAD_MXNX4D(…)
SAD_MXNX3D(…)

// 16x16
SADMXN(…)
SAD_MXNX4D(…)
SAD_MXNX3D(…)

// 16x8
SADMXN(…)
SAD_MXNX4D(…)
SAD_MXNX3D(…)

// 8x16
SADMXN(…)
SAD_MXNX4D(…)
SAD_MXNX3D(…)

// 8x8
SADMXN(…)
SAD_MXNX4D(…)
SAD_MXNX3D(…)

// 8x4
SADMXN(…)
SAD_MXNX4D(…)
SAD_MXNX3D(…)

// 4x8
SADMXN(…)
SAD_MXNX4D(…)
SAD_MXNX3D(…)

// 4x4
SADMXN(…)
SAD_MXNX4D(…)
SAD_MXNX3D(…)

#if !CONFIG_REALTIME_ONLY
SADMXN(4, 16)
SAD_MXNX4D(4, 16)
SADMXN(16, 4)
SAD_MXNX4D(16, 4)
SADMXN(8, 32)
SAD_MXNX4D(8, 32)
SADMXN(32, 8)
SAD_MXNX4D(32, 8)
SADMXN(16, 64)
SAD_MXNX4D(16, 64)
SADMXN(64, 16)
SAD_MXNX4D(64, 16)
SAD_MXNX3D(4, 16)
SAD_MXNX3D(16, 4)
SAD_MXNX3D(8, 32)
SAD_MXNX3D(32, 8)
SAD_MXNX3D(16, 64)
SAD_MXNX3D(64, 16)
#endif  // !CONFIG_REALTIME_ONLY

#if CONFIG_AV1_HIGHBITDEPTH
static inline unsigned int highbd_sad(const uint8_t *a8, int a_stride,
                                      const uint8_t *b8, int b_stride,
                                      int width, int height) {
  int y, x;
  unsigned int sad = 0;
  const uint16_t *a = CONVERT_TO_SHORTPTR(a8);
  const uint16_t *b = CONVERT_TO_SHORTPTR(b8);
  for (y = 0; y < height; y++) {
    for (x = 0; x < width; x++) {
      sad += abs(a[x] - b[x]);
    }

    a += a_stride;
    b += b_stride;
  }
  return sad;
}

static inline unsigned int highbd_sadb(const uint8_t *a8, int a_stride,
                                       const uint8_t *b8, int b_stride,
                                       int width, int height) {
  int y, x;
  unsigned int sad = 0;
  const uint16_t *a = CONVERT_TO_SHORTPTR(a8);
  const uint16_t *b = CONVERT_TO_SHORTPTR(b8);
  for (y = 0; y < height; y++) {
    for (x = 0; x < width; x++) {
      sad += abs(a[x] - b[x]);
    }

    a += a_stride;
    b += b_stride;
  }
  return sad;
}

#define HIGHBD_SADMXN …

#define HIGHBD_SAD_MXNX4D …
// Call SIMD version of aom_highbd_sad_mxnx4d if the 3d version is unavailable.
#define HIGHBD_SAD_MXNX3D …

// 128x128
HIGHBD_SADMXN(128, 128)
HIGHBD_SAD_MXNX4D(128, 128)
HIGHBD_SAD_MXNX3D(128, 128)

// 128x64
HIGHBD_SADMXN(128, 64)
HIGHBD_SAD_MXNX4D(128, 64)
HIGHBD_SAD_MXNX3D(128, 64)

// 64x128
HIGHBD_SADMXN(64, 128)
HIGHBD_SAD_MXNX4D(64, 128)
HIGHBD_SAD_MXNX3D(64, 128)

// 64x64
HIGHBD_SADMXN(64, 64)
HIGHBD_SAD_MXNX4D(64, 64)
HIGHBD_SAD_MXNX3D(64, 64)

// 64x32
HIGHBD_SADMXN(64, 32)
HIGHBD_SAD_MXNX4D(64, 32)
HIGHBD_SAD_MXNX3D(64, 32)

// 32x64
HIGHBD_SADMXN(32, 64)
HIGHBD_SAD_MXNX4D(32, 64)
HIGHBD_SAD_MXNX3D(32, 64)

// 32x32
HIGHBD_SADMXN(32, 32)
HIGHBD_SAD_MXNX4D(32, 32)
HIGHBD_SAD_MXNX3D(32, 32)

// 32x16
HIGHBD_SADMXN(32, 16)
HIGHBD_SAD_MXNX4D(32, 16)
HIGHBD_SAD_MXNX3D(32, 16)

// 16x32
HIGHBD_SADMXN(16, 32)
HIGHBD_SAD_MXNX4D(16, 32)
HIGHBD_SAD_MXNX3D(16, 32)

// 16x16
HIGHBD_SADMXN(16, 16)
HIGHBD_SAD_MXNX4D(16, 16)
HIGHBD_SAD_MXNX3D(16, 16)

// 16x8
HIGHBD_SADMXN(16, 8)
HIGHBD_SAD_MXNX4D(16, 8)
HIGHBD_SAD_MXNX3D(16, 8)

// 8x16
HIGHBD_SADMXN(8, 16)
HIGHBD_SAD_MXNX4D(8, 16)
HIGHBD_SAD_MXNX3D(8, 16)

// 8x8
HIGHBD_SADMXN(8, 8)
HIGHBD_SAD_MXNX4D(8, 8)
HIGHBD_SAD_MXNX3D(8, 8)

// 8x4
HIGHBD_SADMXN(8, 4)
HIGHBD_SAD_MXNX4D(8, 4)
HIGHBD_SAD_MXNX3D(8, 4)

// 4x8
HIGHBD_SADMXN(4, 8)
HIGHBD_SAD_MXNX4D(4, 8)
HIGHBD_SAD_MXNX3D(4, 8)

// 4x4
HIGHBD_SADMXN(4, 4)
HIGHBD_SAD_MXNX4D(4, 4)
HIGHBD_SAD_MXNX3D(4, 4)

#if !CONFIG_REALTIME_ONLY
HIGHBD_SADMXN(4, 16)
HIGHBD_SAD_MXNX4D(4, 16)
HIGHBD_SADMXN(16, 4)
HIGHBD_SAD_MXNX4D(16, 4)
HIGHBD_SADMXN(8, 32)
HIGHBD_SAD_MXNX4D(8, 32)
HIGHBD_SADMXN(32, 8)
HIGHBD_SAD_MXNX4D(32, 8)
HIGHBD_SADMXN(16, 64)
HIGHBD_SAD_MXNX4D(16, 64)
HIGHBD_SADMXN(64, 16)
HIGHBD_SAD_MXNX4D(64, 16)

HIGHBD_SAD_MXNX3D(4, 16)
HIGHBD_SAD_MXNX3D(16, 4)
HIGHBD_SAD_MXNX3D(8, 32)
HIGHBD_SAD_MXNX3D(32, 8)
HIGHBD_SAD_MXNX3D(16, 64)
HIGHBD_SAD_MXNX3D(64, 16)
#endif  // !CONFIG_REALTIME_ONLY
#endif  // CONFIG_AV1_HIGHBITDEPTH