vpx_convolve_copy_neon.c revision 7ce0a1d1337c01056ba24006efab21f00e179e04 
1/*
2 *  Copyright (c) 2014 The WebM project authors. All Rights Reserved.
3 *
4 *  Use of this source code is governed by a BSD-style license
5 *  that can be found in the LICENSE file in the root of the source
6 *  tree. An additional intellectual property rights grant can be found
7 *  in the file PATENTS.  All contributing project authors may
8 *  be found in the AUTHORS file in the root of the source tree.
9 */
10
11#include <arm_neon.h>
12
13#include "./vpx_dsp_rtcd.h"
14#include "vpx/vpx_integer.h"
15
16void vpx_convolve_copy_neon(
17    const uint8_t *src,    // r0
18    ptrdiff_t src_stride,  // r1
19    uint8_t *dst,          // r2
20    ptrdiff_t dst_stride,  // r3
21    const int16_t *filter_x,
22    int filter_x_stride,
23    const int16_t *filter_y,
24    int filter_y_stride,
25    int w,
26    int h) {
27  uint8x8_t d0u8, d2u8;
28  uint8x16_t q0u8, q1u8, q2u8, q3u8;
29  (void)filter_x;  (void)filter_x_stride;
30  (void)filter_y;  (void)filter_y_stride;
31
32  if (w > 32) {  // copy64
33    for (; h > 0; h--) {
34      q0u8 = vld1q_u8(src);
35      q1u8 = vld1q_u8(src + 16);
36      q2u8 = vld1q_u8(src + 32);
37      q3u8 = vld1q_u8(src + 48);
38      src += src_stride;
39
40      vst1q_u8(dst, q0u8);
41      vst1q_u8(dst + 16, q1u8);
42      vst1q_u8(dst + 32, q2u8);
43      vst1q_u8(dst + 48, q3u8);
44      dst += dst_stride;
45    }
46  } else if (w == 32) {  // copy32
47    for (; h > 0; h -= 2) {
48      q0u8 = vld1q_u8(src);
49      q1u8 = vld1q_u8(src + 16);
50      src += src_stride;
51      q2u8 = vld1q_u8(src);
52      q3u8 = vld1q_u8(src + 16);
53      src += src_stride;
54
55      vst1q_u8(dst, q0u8);
56      vst1q_u8(dst + 16, q1u8);
57      dst += dst_stride;
58      vst1q_u8(dst, q2u8);
59      vst1q_u8(dst + 16, q3u8);
60      dst += dst_stride;
61    }
62  } else if (w > 8) {  // copy16
63    for (; h > 0; h -= 2) {
64      q0u8 = vld1q_u8(src);
65      src += src_stride;
66      q1u8 = vld1q_u8(src);
67      src += src_stride;
68
69      vst1q_u8(dst, q0u8);
70      dst += dst_stride;
71      vst1q_u8(dst, q1u8);
72      dst += dst_stride;
73    }
74  } else if (w == 8) {  // copy8
75    for (; h > 0; h -= 2) {
76      d0u8 = vld1_u8(src);
77      src += src_stride;
78      d2u8 = vld1_u8(src);
79      src += src_stride;
80
81      vst1_u8(dst, d0u8);
82      dst += dst_stride;
83      vst1_u8(dst, d2u8);
84      dst += dst_stride;
85    }
86  } else {  // copy4
87    for (; h > 0; h--) {
88      *(uint32_t *)dst = *(const uint32_t *)src;
89      src += src_stride;
90      dst += dst_stride;
91    }
92  }
93  return;
94}
95