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
vpx_convolve_copy_neon(const uint8_t * src,ptrdiff_t src_stride,uint8_t * dst,ptrdiff_t dst_stride,const InterpKernel * filter,int x0_q4,int x_step_q4,int y0_q4,int y_step_q4,int w,int h)16 void vpx_convolve_copy_neon(const uint8_t *src, ptrdiff_t src_stride,
17 uint8_t *dst, ptrdiff_t dst_stride,
18 const InterpKernel *filter, int x0_q4,
19 int x_step_q4, int y0_q4, int y_step_q4, int w,
20 int h) {
21 (void)filter;
22 (void)x0_q4;
23 (void)x_step_q4;
24 (void)y0_q4;
25 (void)y_step_q4;
26
27 if (w < 8) { // copy4
28 do {
29 *(uint32_t *)dst = *(const uint32_t *)src;
30 src += src_stride;
31 dst += dst_stride;
32 *(uint32_t *)dst = *(const uint32_t *)src;
33 src += src_stride;
34 dst += dst_stride;
35 h -= 2;
36 } while (h > 0);
37 } else if (w == 8) { // copy8
38 uint8x8_t s0, s1;
39 do {
40 s0 = vld1_u8(src);
41 src += src_stride;
42 s1 = vld1_u8(src);
43 src += src_stride;
44
45 vst1_u8(dst, s0);
46 dst += dst_stride;
47 vst1_u8(dst, s1);
48 dst += dst_stride;
49 h -= 2;
50 } while (h > 0);
51 } else if (w < 32) { // copy16
52 uint8x16_t s0, s1;
53 do {
54 s0 = vld1q_u8(src);
55 src += src_stride;
56 s1 = vld1q_u8(src);
57 src += src_stride;
58
59 vst1q_u8(dst, s0);
60 dst += dst_stride;
61 vst1q_u8(dst, s1);
62 dst += dst_stride;
63 h -= 2;
64 } while (h > 0);
65 } else if (w == 32) { // copy32
66 uint8x16_t s0, s1, s2, s3;
67 do {
68 s0 = vld1q_u8(src);
69 s1 = vld1q_u8(src + 16);
70 src += src_stride;
71 s2 = vld1q_u8(src);
72 s3 = vld1q_u8(src + 16);
73 src += src_stride;
74
75 vst1q_u8(dst, s0);
76 vst1q_u8(dst + 16, s1);
77 dst += dst_stride;
78 vst1q_u8(dst, s2);
79 vst1q_u8(dst + 16, s3);
80 dst += dst_stride;
81 h -= 2;
82 } while (h > 0);
83 } else { // copy64
84 uint8x16_t s0, s1, s2, s3;
85 do {
86 s0 = vld1q_u8(src);
87 s1 = vld1q_u8(src + 16);
88 s2 = vld1q_u8(src + 32);
89 s3 = vld1q_u8(src + 48);
90 src += src_stride;
91
92 vst1q_u8(dst, s0);
93 vst1q_u8(dst + 16, s1);
94 vst1q_u8(dst + 32, s2);
95 vst1q_u8(dst + 48, s3);
96 dst += dst_stride;
97 } while (--h);
98 }
99 }
100