hevcpred_init_aarch64.c

Go to the documentation of this file.

/*
 * HEVC Intra Prediction NEON initialization
 *
 * Copyright (c) 2026 Jun Zhao <barryjzhao@tencent.com>
 *
 * This file is part of FFmpeg.
 *
 * FFmpeg 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.
 *
 * FFmpeg 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 Lesser General Public
 * License along with FFmpeg; if not, write to the Free Software
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
 */
 
#include "libavutil/attributes.h"
#include "libavutil/avassert.h"
#include "libavutil/aarch64/cpu.h"
#include "libavcodec/hevc/pred.h"
 
// DC prediction
void ff_hevc_pred_dc_4x4_8_neon(uint8_t *src, const uint8_t *top,
                               const uint8_t *left, ptrdiff_t stride,
                               int c_idx);
void ff_hevc_pred_dc_8x8_8_neon(uint8_t *src, const uint8_t *top,
                               const uint8_t *left, ptrdiff_t stride,
                               int c_idx);
void ff_hevc_pred_dc_16x16_8_neon(uint8_t *src, const uint8_t *top,
                                const uint8_t *left, ptrdiff_t stride,
                                int c_idx);
void ff_hevc_pred_dc_32x32_8_neon(uint8_t *src, const uint8_t *top,
                                const uint8_t *left, ptrdiff_t stride,
                                int c_idx);
 
// Planar prediction
void ff_hevc_pred_planar_4x4_8_neon(uint8_t *src, const uint8_t *top,
                                   const uint8_t *left, ptrdiff_t stride);
void ff_hevc_pred_planar_8x8_8_neon(uint8_t *src, const uint8_t *top,
                                   const uint8_t *left, ptrdiff_t stride);
void ff_hevc_pred_planar_16x16_8_neon(uint8_t *src, const uint8_t *top,
                                    const uint8_t *left, ptrdiff_t stride);
void ff_hevc_pred_planar_32x32_8_neon(uint8_t *src, const uint8_t *top,
                                    const uint8_t *left, ptrdiff_t stride);
 
// 3-tap reference sample filter
void ff_hevc_ref_filter_3tap_8x8_8_neon(uint8_t *filtered_left,
                                        uint8_t *filtered_top,
                                        const uint8_t *left,
                                        const uint8_t *top, int size);
void ff_hevc_ref_filter_3tap_16x16_8_neon(uint8_t *filtered_left,
                                          uint8_t *filtered_top,
                                          const uint8_t *left,
                                          const uint8_t *top, int size);
void ff_hevc_ref_filter_3tap_32x32_8_neon(uint8_t *filtered_left,
                                          uint8_t *filtered_top,
                                          const uint8_t *left,
                                          const uint8_t *top, int size);
 
// Strong intra smoothing
void ff_hevc_ref_filter_strong_8_neon(uint8_t *filtered_top, uint8_t *left,
                                      const uint8_t *top);
 
// Mode 10 and 26
void ff_hevc_pred_angular_mode_10_8_neon(uint8_t *src, const uint8_t *top,
                                        const uint8_t *left, ptrdiff_t stride,
                                        int c_idx, int log2_size);
void ff_hevc_pred_angular_mode_26_8_neon(uint8_t *src, const uint8_t *top,
                                        const uint8_t *left, ptrdiff_t stride,
                                        int c_idx, int log2_size);
 
static void pred_dc_neon(uint8_t *src, const uint8_t *top,
                         const uint8_t *left, ptrdiff_t stride,
                         int log2_size, int c_idx)
{
    switch (log2_size) {
    case 2:
        ff_hevc_pred_dc_4x4_8_neon(src, top, left, stride, c_idx);
        break;
    case 3:
        ff_hevc_pred_dc_8x8_8_neon(src, top, left, stride, c_idx);
        break;
    case 4:
        ff_hevc_pred_dc_16x16_8_neon(src, top, left, stride, c_idx);
        break;
    case 5:
        ff_hevc_pred_dc_32x32_8_neon(src, top, left, stride, c_idx);
        break;
    default:
        av_unreachable("log2_size must be 2, 3, 4 or 5");
    }
}
 
#define PRED_ANGULAR_NEON(IDX, LOG2)                                          \
static void pred_angular_##IDX##_neon(uint8_t *src, const uint8_t *top,       \
                                      const uint8_t *left, ptrdiff_t stride,  \
                                      int c_idx, int mode)                    \
{                                                                             \
    if (mode == 10)                                                           \
        ff_hevc_pred_angular_mode_10_8_neon(src, top, left, stride,           \
                                           c_idx, LOG2);                      \
    else if (mode == 26)                                                      \
        ff_hevc_pred_angular_mode_26_8_neon(src, top, left, stride,           \
                                           c_idx, LOG2);                      \
    else                                                                      \
        ff_hevc_pred_angular_##IDX##_8(src, top, left, stride, c_idx, mode);  \
}
 
PRED_ANGULAR_NEON(0, 2)
PRED_ANGULAR_NEON(1, 3)
PRED_ANGULAR_NEON(2, 4)
PRED_ANGULAR_NEON(3, 5)
 
#undef PRED_ANGULAR_NEON
 
av_cold void ff_hevc_pred_init_aarch64(HEVCPredContext *hpc, int bit_depth)
{
    int cpu_flags = av_get_cpu_flags();
 
    if (!have_neon(cpu_flags))
        return;
 
    if (bit_depth == 8) {
        hpc->pred_dc        = pred_dc_neon;
        hpc->pred_planar[0] = ff_hevc_pred_planar_4x4_8_neon;
        hpc->pred_planar[1] = ff_hevc_pred_planar_8x8_8_neon;
        hpc->pred_planar[2] = ff_hevc_pred_planar_16x16_8_neon;
        hpc->pred_planar[3] = ff_hevc_pred_planar_32x32_8_neon;
 
        hpc->ref_filter_3tap[0] = ff_hevc_ref_filter_3tap_8x8_8_neon;
        hpc->ref_filter_3tap[1] = ff_hevc_ref_filter_3tap_16x16_8_neon;
        hpc->ref_filter_3tap[2] = ff_hevc_ref_filter_3tap_32x32_8_neon;
        hpc->ref_filter_strong  = ff_hevc_ref_filter_strong_8_neon;
 
        hpc->pred_angular[0] = pred_angular_0_neon;
        hpc->pred_angular[1] = pred_angular_1_neon;
        hpc->pred_angular[2] = pred_angular_2_neon;
        hpc->pred_angular[3] = pred_angular_3_neon;
    }
}