2 * Copyright (c) 2010 Mans Rullgard
3 * Copyright (c) 2014 James Yu <james.yu@linaro.org>
5 * This file is part of FFmpeg.
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.
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.
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
26 #include "libavutil/attributes.h"
27 #include "libavutil/cpu.h"
29 # include "libavutil/aarch64/cpu.h"
31 # include "libavutil/arm/cpu.h"
34 #include "libavcodec/mpegvideo.h"
36 static void inline ff_dct_unquantize_h263_neon(int qscale, int qadd, int nCoeffs,
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;
45 dzs16 = vdup_n_s16(0);
46 qzs16 = vdupq_n_s16(0);
48 q15s16 = vdupq_n_s16(qscale << 1);
49 q14s16 = vdupq_n_s16(qadd);
50 q13s16 = vnegq_s16(q14s16);
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);
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);
86 static void dct_unquantize_h263_inter_neon(MpegEncContext *s, int16_t *block,
89 int nCoeffs = s->inter_scantable.raster_end[s->block_last_index[n]];
90 int qadd = (qscale - 1) | 1;
92 ff_dct_unquantize_h263_neon(qscale, qadd, nCoeffs + 1, block);
95 static void dct_unquantize_h263_intra_neon(MpegEncContext *s, int16_t *block,
103 block[0] *= s->y_dc_scale;
105 block[0] *= s->c_dc_scale;
106 qadd = (qscale - 1) | 1;
114 nCoeffs = s->inter_scantable.raster_end[s->block_last_index[n]];
121 ff_dct_unquantize_h263_neon(qscale, qadd, nCoeffs + 1, block);
127 av_cold void ff_mpv_common_init_neon(MpegEncContext *s)
129 int cpu_flags = av_get_cpu_flags();
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;