x86: check for AV_CPU_FLAG_AVXSLOW where useful
[libav.git] / libavfilter / x86 / af_volume_init.c
CommitLineData
f96f1e06
JR
1/*
2 * This file is part of Libav.
3 *
4 * Libav is free software; you can redistribute it and/or
5 * modify it under the terms of the GNU Lesser General Public
6 * License as published by the Free Software Foundation; either
7 * version 2.1 of the License, or (at your option) any later version.
8 *
9 * Libav is distributed in the hope that it will be useful,
10 * but WITHOUT ANY WARRANTY; without even the implied warranty of
11 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
12 * Lesser General Public License for more details.
13 *
14 * You should have received a copy of the GNU Lesser General Public
15 * License along with Libav; if not, write to the Free Software
16 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
17 */
18
19#include "config.h"
093804a9 20#include "libavutil/attributes.h"
f96f1e06
JR
21#include "libavutil/cpu.h"
22#include "libavutil/samplefmt.h"
23#include "libavutil/x86/cpu.h"
24#include "libavfilter/af_volume.h"
25
26void ff_scale_samples_s16_sse2(uint8_t *dst, const uint8_t *src, int len,
27 int volume);
28
b30a3633
JR
29void ff_scale_samples_s32_sse2(uint8_t *dst, const uint8_t *src, int len,
30 int volume);
31void ff_scale_samples_s32_ssse3_atom(uint8_t *dst, const uint8_t *src, int len,
32 int volume);
33void ff_scale_samples_s32_avx(uint8_t *dst, const uint8_t *src, int len,
34 int volume);
35
093804a9 36av_cold void ff_volume_init_x86(VolumeContext *vol)
f96f1e06 37{
3ac7fa81 38 int cpu_flags = av_get_cpu_flags();
f96f1e06
JR
39 enum AVSampleFormat sample_fmt = av_get_packed_sample_fmt(vol->sample_fmt);
40
41 if (sample_fmt == AV_SAMPLE_FMT_S16) {
3ac7fa81 42 if (EXTERNAL_SSE2(cpu_flags) && vol->volume_i < 32768) {
f96f1e06
JR
43 vol->scale_samples = ff_scale_samples_s16_sse2;
44 vol->samples_align = 8;
45 }
b30a3633 46 } else if (sample_fmt == AV_SAMPLE_FMT_S32) {
3ac7fa81 47 if (EXTERNAL_SSE2(cpu_flags)) {
b30a3633
JR
48 vol->scale_samples = ff_scale_samples_s32_sse2;
49 vol->samples_align = 4;
50 }
3ac7fa81 51 if (EXTERNAL_SSSE3(cpu_flags) && cpu_flags & AV_CPU_FLAG_ATOM) {
b30a3633
JR
52 vol->scale_samples = ff_scale_samples_s32_ssse3_atom;
53 vol->samples_align = 4;
54 }
d68c0538 55 if (EXTERNAL_AVX_FAST(cpu_flags)) {
b30a3633
JR
56 vol->scale_samples = ff_scale_samples_s32_avx;
57 vol->samples_align = 8;
58 }
f96f1e06
JR
59 }
60}