idct4x4_1_add_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/inv_txfm.h"
14#include "vpx_ports/mem.h"
15
16void vpx_idct4x4_1_add_neon(
17        int16_t *input,
18        uint8_t *dest,
19        int dest_stride) {
20    uint8x8_t d6u8;
21    uint32x2_t d2u32 = vdup_n_u32(0);
22    uint16x8_t q8u16;
23    int16x8_t q0s16;
24    uint8_t *d1, *d2;
25    int16_t i, a1, cospi_16_64 = 11585;
26    int16_t out = dct_const_round_shift(input[0] * cospi_16_64);
27    out = dct_const_round_shift(out * cospi_16_64);
28    a1 = ROUND_POWER_OF_TWO(out, 4);
29
30    q0s16 = vdupq_n_s16(a1);
31
32    // dc_only_idct_add
33    d1 = d2 = dest;
34    for (i = 0; i < 2; i++) {
35        d2u32 = vld1_lane_u32((const uint32_t *)d1, d2u32, 0);
36        d1 += dest_stride;
37        d2u32 = vld1_lane_u32((const uint32_t *)d1, d2u32, 1);
38        d1 += dest_stride;
39
40        q8u16 = vaddw_u8(vreinterpretq_u16_s16(q0s16),
41                         vreinterpret_u8_u32(d2u32));
42        d6u8 = vqmovun_s16(vreinterpretq_s16_u16(q8u16));
43
44        vst1_lane_u32((uint32_t *)d2, vreinterpret_u32_u8(d6u8), 0);
45        d2 += dest_stride;
46        vst1_lane_u32((uint32_t *)d2, vreinterpret_u32_u8(d6u8), 1);
47        d2 += dest_stride;
48    }
49    return;
50}
51