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_idct16x16_1_add_neon(
17        int16_t *input,
18        uint8_t *dest,
19        int dest_stride) {
20    uint8x8_t d2u8, d3u8, d30u8, d31u8;
21    uint64x1_t d2u64, d3u64, d4u64, d5u64;
22    uint16x8_t q0u16, q9u16, q10u16, q11u16, q12u16;
23    int16x8_t q0s16;
24    uint8_t *d1, *d2;
25    int16_t i, j, 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, 6);
29
30    q0s16 = vdupq_n_s16(a1);
31    q0u16 = vreinterpretq_u16_s16(q0s16);
32
33    for (d1 = d2 = dest, i = 0; i < 4; i++) {
34        for (j = 0; j < 2; j++) {
35            d2u64 = vld1_u64((const uint64_t *)d1);
36            d3u64 = vld1_u64((const uint64_t *)(d1 + 8));
37            d1 += dest_stride;
38            d4u64 = vld1_u64((const uint64_t *)d1);
39            d5u64 = vld1_u64((const uint64_t *)(d1 + 8));
40            d1 += dest_stride;
41
42            q9u16 = vaddw_u8(q0u16, vreinterpret_u8_u64(d2u64));
43            q10u16 = vaddw_u8(q0u16, vreinterpret_u8_u64(d3u64));
44            q11u16 = vaddw_u8(q0u16, vreinterpret_u8_u64(d4u64));
45            q12u16 = vaddw_u8(q0u16, vreinterpret_u8_u64(d5u64));
46
47            d2u8 = vqmovun_s16(vreinterpretq_s16_u16(q9u16));
48            d3u8 = vqmovun_s16(vreinterpretq_s16_u16(q10u16));
49            d30u8 = vqmovun_s16(vreinterpretq_s16_u16(q11u16));
50            d31u8 = vqmovun_s16(vreinterpretq_s16_u16(q12u16));
51
52            vst1_u64((uint64_t *)d2, vreinterpret_u64_u8(d2u8));
53            vst1_u64((uint64_t *)(d2 + 8), vreinterpret_u64_u8(d3u8));
54            d2 += dest_stride;
55            vst1_u64((uint64_t *)d2, vreinterpret_u64_u8(d30u8));
56            vst1_u64((uint64_t *)(d2 + 8), vreinterpret_u64_u8(d31u8));
57            d2 += dest_stride;
58        }
59    }
60    return;
61}
62