OLD | NEW |
| (Empty) |
1 /* | |
2 * Copyright (c) 2015 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 #include "./vp9_rtcd.h" | |
13 #include "./vpx_config.h" | |
14 | |
15 #include "vpx/vpx_integer.h" | |
16 | |
17 static INLINE unsigned int horizontal_add_u16x8(const uint16x8_t v_16x8) { | |
18 const uint32x4_t a = vpaddlq_u16(v_16x8); | |
19 const uint64x2_t b = vpaddlq_u32(a); | |
20 const uint32x2_t c = vadd_u32(vreinterpret_u32_u64(vget_low_u64(b)), | |
21 vreinterpret_u32_u64(vget_high_u64(b))); | |
22 return vget_lane_u32(c, 0); | |
23 } | |
24 | |
25 unsigned int vp9_avg_8x8_neon(const uint8_t *s, int p) { | |
26 uint8x8_t v_s0 = vld1_u8(s); | |
27 const uint8x8_t v_s1 = vld1_u8(s + p); | |
28 uint16x8_t v_sum = vaddl_u8(v_s0, v_s1); | |
29 | |
30 v_s0 = vld1_u8(s + 2 * p); | |
31 v_sum = vaddw_u8(v_sum, v_s0); | |
32 | |
33 v_s0 = vld1_u8(s + 3 * p); | |
34 v_sum = vaddw_u8(v_sum, v_s0); | |
35 | |
36 v_s0 = vld1_u8(s + 4 * p); | |
37 v_sum = vaddw_u8(v_sum, v_s0); | |
38 | |
39 v_s0 = vld1_u8(s + 5 * p); | |
40 v_sum = vaddw_u8(v_sum, v_s0); | |
41 | |
42 v_s0 = vld1_u8(s + 6 * p); | |
43 v_sum = vaddw_u8(v_sum, v_s0); | |
44 | |
45 v_s0 = vld1_u8(s + 7 * p); | |
46 v_sum = vaddw_u8(v_sum, v_s0); | |
47 | |
48 return (horizontal_add_u16x8(v_sum) + 32) >> 6; | |
49 } | |
OLD | NEW |