vpx_convolve_neon.c 2.71 KB
Newer Older
Johann's avatar
Johann committed
1 2 3 4 5 6 7 8 9 10
/*
 *  Copyright (c) 2013 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.
 */

11 12
#include <assert.h>

Zoe Liu's avatar
Zoe Liu committed
13
#include "./vpx_dsp_rtcd.h"
Yaowu Xu's avatar
Yaowu Xu committed
14 15
#include "aom_dsp/vpx_dsp_common.h"
#include "aom_ports/mem.h"
Johann's avatar
Johann committed
16

clang-format's avatar
clang-format committed
17 18 19
void vpx_convolve8_neon(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst,
                        ptrdiff_t dst_stride, const int16_t *filter_x,
                        int x_step_q4, const int16_t *filter_y, int y_step_q4,
Johann's avatar
Johann committed
20 21 22 23
                        int w, int h) {
  /* Given our constraints: w <= 64, h <= 64, taps == 8 we can reduce the
   * maximum buffer size to 64 * 64 + 7 (+ 1 to make it divisible by 4).
   */
24
  DECLARE_ALIGNED(8, uint8_t, temp[64 * 72]);
Johann's avatar
Johann committed
25 26 27 28

  // Account for the vertical phase needing 3 lines prior and 4 lines post
  int intermediate_height = h + 7;

29 30
  assert(y_step_q4 == 16);
  assert(x_step_q4 == 16);
Johann's avatar
Johann committed
31 32 33 34 35 36

  /* Filter starting 3 lines back. The neon implementation will ignore the
   * given height and filter a multiple of 4 lines. Since this goes in to
   * the temp buffer which has lots of extra room and is subsequently discarded
   * this is safe if somewhat less than ideal.
   */
clang-format's avatar
clang-format committed
37 38 39
  vpx_convolve8_horiz_neon(src - src_stride * 3, src_stride, temp, 64, filter_x,
                           x_step_q4, filter_y, y_step_q4, w,
                           intermediate_height);
Johann's avatar
Johann committed
40 41

  /* Step into the temp buffer 3 lines to get the actual frame data */
clang-format's avatar
clang-format committed
42 43
  vpx_convolve8_vert_neon(temp + 64 * 3, 64, dst, dst_stride, filter_x,
                          x_step_q4, filter_y, y_step_q4, w, h);
Johann's avatar
Johann committed
44 45
}

Zoe Liu's avatar
Zoe Liu committed
46
void vpx_convolve8_avg_neon(const uint8_t *src, ptrdiff_t src_stride,
Johann's avatar
Johann committed
47 48
                            uint8_t *dst, ptrdiff_t dst_stride,
                            const int16_t *filter_x, int x_step_q4,
clang-format's avatar
clang-format committed
49 50
                            const int16_t *filter_y, int y_step_q4, int w,
                            int h) {
51
  DECLARE_ALIGNED(8, uint8_t, temp[64 * 72]);
Johann's avatar
Johann committed
52 53
  int intermediate_height = h + 7;

54 55
  assert(y_step_q4 == 16);
  assert(x_step_q4 == 16);
Johann's avatar
Johann committed
56 57 58 59

  /* This implementation has the same issues as above. In addition, we only want
   * to average the values after both passes.
   */
clang-format's avatar
clang-format committed
60 61 62 63 64
  vpx_convolve8_horiz_neon(src - src_stride * 3, src_stride, temp, 64, filter_x,
                           x_step_q4, filter_y, y_step_q4, w,
                           intermediate_height);
  vpx_convolve8_avg_vert_neon(temp + 64 * 3, 64, dst, dst_stride, filter_x,
                              x_step_q4, filter_y, y_step_q4, w, h);
Johann's avatar
Johann committed
65
}