1/*
2 * Copyright (c) 2010 Mans Rullgard
3 * Copyright (c) 2014 James Yu <james.yu@linaro.org>
4 *
5 * This file is part of FFmpeg.
6 *
7 * FFmpeg is free software; you can redistribute it and/or
8 * modify it under the terms of the GNU Lesser General Public
9 * License as published by the Free Software Foundation; either
10 * version 2.1 of the License, or (at your option) any later version.
11 *
12 * FFmpeg is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
15 * Lesser General Public License for more details.
16 *
17 * You should have received a copy of the GNU Lesser General Public
18 * License along with FFmpeg; if not, write to the Free Software
19 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
20 */
21
22#include <arm_neon.h>
23
24#include "config.h"
25
26#include "libavutil/attributes.h"
27#include "libavutil/cpu.h"
28#if   ARCH_AARCH64
29#   include "libavutil/aarch64/cpu.h"
30#elif ARCH_ARM
31#   include "libavutil/arm/cpu.h"
32#endif
33
34#include "libavcodec/mpegvideo.h"
35
36static void inline ff_dct_unquantize_h263_neon(int qscale, int qadd, int nCoeffs,
37                                               int16_t *block)
38{
39    int16x8_t q0s16, q2s16, q3s16, q8s16, q10s16, q11s16, q13s16;
40    int16x8_t q14s16, q15s16, qzs16;
41    int16x4_t d0s16, d2s16, d3s16, dzs16;
42    uint16x8_t q1u16, q9u16;
43    uint16x4_t d1u16;
44
45    dzs16 = vdup_n_s16(0);
46    qzs16 = vdupq_n_s16(0);
47
48    q15s16 = vdupq_n_s16(qscale << 1);
49    q14s16 = vdupq_n_s16(qadd);
50    q13s16 = vnegq_s16(q14s16);
51
52    if (nCoeffs > 4) {
53        for (; nCoeffs > 8; nCoeffs -= 16, block += 16) {
54            q0s16 = vld1q_s16(block);
55            q3s16 = vreinterpretq_s16_u16(vcltq_s16(q0s16, qzs16));
56            q8s16 = vld1q_s16(block + 8);
57            q1u16 = vceqq_s16(q0s16, qzs16);
58            q2s16 = vmulq_s16(q0s16, q15s16);
59            q11s16 = vreinterpretq_s16_u16(vcltq_s16(q8s16, qzs16));
60            q10s16 = vmulq_s16(q8s16, q15s16);
61            q3s16 = vbslq_s16(vreinterpretq_u16_s16(q3s16), q13s16, q14s16);
62            q11s16 = vbslq_s16(vreinterpretq_u16_s16(q11s16), q13s16, q14s16);
63            q2s16 = vaddq_s16(q2s16, q3s16);
64            q9u16 = vceqq_s16(q8s16, qzs16);
65            q10s16 = vaddq_s16(q10s16, q11s16);
66            q0s16 = vbslq_s16(q1u16, q0s16, q2s16);
67            q8s16 = vbslq_s16(q9u16, q8s16, q10s16);
68            vst1q_s16(block, q0s16);
69            vst1q_s16(block + 8, q8s16);
70        }
71    }
72    if (nCoeffs <= 0)
73        return;
74
75    d0s16 = vld1_s16(block);
76    d3s16 = vreinterpret_s16_u16(vclt_s16(d0s16, dzs16));
77    d1u16 = vceq_s16(d0s16, dzs16);
78    d2s16 = vmul_s16(d0s16, vget_high_s16(q15s16));
79    d3s16 = vbsl_s16(vreinterpret_u16_s16(d3s16),
80                     vget_high_s16(q13s16), vget_high_s16(q14s16));
81    d2s16 = vadd_s16(d2s16, d3s16);
82    d0s16 = vbsl_s16(d1u16, d0s16, d2s16);
83    vst1_s16(block, d0s16);
84}
85
86static void dct_unquantize_h263_inter_neon(MpegEncContext *s, int16_t *block,
87                                           int n, int qscale)
88{
89    int nCoeffs = s->inter_scantable.raster_end[s->block_last_index[n]];
90    int qadd    = (qscale - 1) | 1;
91
92    ff_dct_unquantize_h263_neon(qscale, qadd, nCoeffs + 1, block);
93}
94
95static void dct_unquantize_h263_intra_neon(MpegEncContext *s, int16_t *block,
96                                           int n, int qscale)
97{
98    int qadd;
99    int nCoeffs, blk0;
100
101    if (!s->h263_aic) {
102        if (n < 4)
103            block[0] *= s->y_dc_scale;
104        else
105            block[0] *= s->c_dc_scale;
106        qadd = (qscale - 1) | 1;
107    } else {
108        qadd = 0;
109    }
110
111    if (s->ac_pred) {
112        nCoeffs = 63;
113    } else {
114        nCoeffs = s->inter_scantable.raster_end[s->block_last_index[n]];
115        if (nCoeffs <= 0)
116            return;
117    }
118
119    blk0 = block[0];
120
121    ff_dct_unquantize_h263_neon(qscale, qadd, nCoeffs + 1, block);
122
123    block[0] = blk0;
124}
125
126
127av_cold void ff_mpv_common_init_neon(MpegEncContext *s)
128{
129    int cpu_flags = av_get_cpu_flags();
130
131    if (have_neon(cpu_flags)) {
132        s->dct_unquantize_h263_intra = dct_unquantize_h263_intra_neon;
133        s->dct_unquantize_h263_inter = dct_unquantize_h263_inter_neon;
134    }
135}
136