OLD | NEW |
1 /* | 1 /* |
2 * Copyright (c) 2014 The WebM project authors. All Rights Reserved. | 2 * Copyright (c) 2014 The WebM project authors. All Rights Reserved. |
3 * | 3 * |
4 * Use of this source code is governed by a BSD-style license | 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 | 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 | 6 * tree. An additional intellectual property rights grant can be found |
7 * in the file PATENTS. All contributing project authors may | 7 * in the file PATENTS. All contributing project authors may |
8 * be found in the AUTHORS file in the root of the source tree. | 8 * be found in the AUTHORS file in the root of the source tree. |
9 */ | 9 */ |
10 | 10 |
11 #include <arm_neon.h> | 11 #include <arm_neon.h> |
12 | 12 |
13 #include "vpx_dsp/inv_txfm.h" | 13 #include "vpx_dsp/inv_txfm.h" |
14 #include "vpx_ports/mem.h" | 14 #include "vpx_ports/mem.h" |
15 | 15 |
16 void vp9_idct4x4_1_add_neon( | 16 void vpx_idct4x4_1_add_neon( |
17 int16_t *input, | 17 int16_t *input, |
18 uint8_t *dest, | 18 uint8_t *dest, |
19 int dest_stride) { | 19 int dest_stride) { |
20 uint8x8_t d6u8; | 20 uint8x8_t d6u8; |
21 uint32x2_t d2u32 = vdup_n_u32(0); | 21 uint32x2_t d2u32 = vdup_n_u32(0); |
22 uint16x8_t q8u16; | 22 uint16x8_t q8u16; |
23 int16x8_t q0s16; | 23 int16x8_t q0s16; |
24 uint8_t *d1, *d2; | 24 uint8_t *d1, *d2; |
25 int16_t i, a1, cospi_16_64 = 11585; | 25 int16_t i, a1, cospi_16_64 = 11585; |
26 int16_t out = dct_const_round_shift(input[0] * cospi_16_64); | 26 int16_t out = dct_const_round_shift(input[0] * cospi_16_64); |
(...skipping 14 matching lines...) Expand all Loading... |
41 vreinterpret_u8_u32(d2u32)); | 41 vreinterpret_u8_u32(d2u32)); |
42 d6u8 = vqmovun_s16(vreinterpretq_s16_u16(q8u16)); | 42 d6u8 = vqmovun_s16(vreinterpretq_s16_u16(q8u16)); |
43 | 43 |
44 vst1_lane_u32((uint32_t *)d2, vreinterpret_u32_u8(d6u8), 0); | 44 vst1_lane_u32((uint32_t *)d2, vreinterpret_u32_u8(d6u8), 0); |
45 d2 += dest_stride; | 45 d2 += dest_stride; |
46 vst1_lane_u32((uint32_t *)d2, vreinterpret_u32_u8(d6u8), 1); | 46 vst1_lane_u32((uint32_t *)d2, vreinterpret_u32_u8(d6u8), 1); |
47 d2 += dest_stride; | 47 d2 += dest_stride; |
48 } | 48 } |
49 return; | 49 return; |
50 } | 50 } |
OLD | NEW |