virtualenv/FFmpeg
1
0
Fork 0
mirror of https://github.com/FFmpeg/FFmpeg.git synced 2024-12-23 12:43:46 +02:00
Code Issues Releases Activity
d03b327787
FFmpeg/libswresample/x86/audio_convert_init.c

183 lines
10 KiB
C
Raw Normal View History

swr: add int16_to_int32_mmx/sse Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2012-04-28 13:23:42 +03:00
/*
* Copyright (C) 2012 Michael Niedermayer (michaelni@gmx.at)
*
* This file is part of libswresample
*
* libswresample is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
* version 2.1 of the License, or (at your option) any later version.
*
* libswresample is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with libswresample; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
Include attributes.h directly Some files currently rely on libavutil/cpu.h to include it for them; yet said file won't use include it any more after the currently deprecated functions are removed, so include attributes.h directly. Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
2021-04-14 15:11:43 +02:00
#include "libavutil/attributes.h"
x86/swr: use lavu helper macros to check CPU extensions Signed-off-by: James Almer <jamrial@gmail.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2014-07-03 04:16:50 +03:00
#include "libavutil/x86/cpu.h"
swr: add int16_to_int32_mmx/sse Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2012-04-28 13:23:42 +03:00
#include "libswresample/swresample_internal.h"
#include "libswresample/audioconvert.h"
Avoid using empty macro arguments. These are not supported by all compilers (gcc 2.95 but also older SPARC compilers, see gcc bug #33304 for example), and there is no real need for them. One use of this feature remains in libavdevice/v4l2.c which can't be replaced quite as easily. Signed-off-by: Reimar Döffinger <Reimar.Doeffinger@gmx.de>
2013-12-30 14:10:35 +03:00
#define PROTO(pre, in, out, cap) void ff ## pre ## in## _to_ ##out## _a_ ##cap(uint8_t **dst, const uint8_t **src, int len);
swr-x86-simd: create prototypes with macros, this is simpler. Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2012-05-06 16:22:42 +03:00
#define PROTO2(pre, out, cap) PROTO(pre, int16, out, cap) PROTO(pre, int32, out, cap) PROTO(pre, float, out, cap)
#define PROTO3(pre, cap) PROTO2(pre, int16, cap) PROTO2(pre, int32, cap) PROTO2(pre, float, cap)
x86/swr: add ff_float_to_int32_a_avx2 13797 decicycles in ff_float_to_int32_a_sse2, 32768 runs, 0 skips 8603 decicycles in ff_float_to_int32_a_avx2, 32766 runs, 2 skips Reviewed-by: Christophe Gisquet <christophe.gisquet@gmail.com> Signed-off-by: James Almer <jamrial@gmail.com>
2014-11-06 22:38:37 +02:00
#define PROTO4(pre) PROTO3(pre, mmx) PROTO3(pre, sse) PROTO3(pre, sse2) PROTO3(pre, ssse3) PROTO3(pre, sse4) PROTO3(pre, avx) PROTO3(pre, avx2)
Avoid using empty macro arguments. These are not supported by all compilers (gcc 2.95 but also older SPARC compilers, see gcc bug #33304 for example), and there is no real need for them. One use of this feature remains in libavdevice/v4l2.c which can't be replaced quite as easily. Signed-off-by: Reimar Döffinger <Reimar.Doeffinger@gmx.de>
2013-12-30 14:10:35 +03:00
PROTO4(_)
PROTO4(_pack_2ch_)
PROTO4(_pack_6ch_)
x86/swr: add SSE2/AVX pack_8ch functions Reviewed-by: Michael Niedermayer <michaelni@gmx.at> Reviewed-by: Ronald S. Bultje <rsbultje@gmail.com> Signed-off-by: James Almer <jamrial@gmail.com>
2014-12-30 23:16:37 +02:00
PROTO4(_pack_8ch_)
Avoid using empty macro arguments. These are not supported by all compilers (gcc 2.95 but also older SPARC compilers, see gcc bug #33304 for example), and there is no real need for them. One use of this feature remains in libavdevice/v4l2.c which can't be replaced quite as easily. Signed-off-by: Reimar Döffinger <Reimar.Doeffinger@gmx.de>
2013-12-30 14:10:35 +03:00
PROTO4(_unpack_2ch_)
x86/swr: add SSE/AVX unpack_6ch functions int32/float only Reviewed-by: Michael Niedermayer <michaelni@gmx.at> Signed-off-by: James Almer <jamrial@gmail.com>
2015-01-12 20:32:24 +02:00
PROTO4(_unpack_6ch_)
swr: implement stereo S16/S32/FLT->S16/S32/FLT planar->packed in SSE/SSE2 Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2012-05-05 16:31:06 +03:00
swr: add av_cold to init/free functions Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2012-09-09 03:26:20 +03:00
av_cold void swri_audio_convert_init_x86(struct AudioConvert *ac,
swr: add int16_to_int32_mmx/sse Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2012-04-28 13:23:42 +03:00
enum AVSampleFormat out_fmt,
enum AVSampleFormat in_fmt,
int channels){
int mm_flags = av_get_cpu_flags();
ac->simd_f= NULL;
//FIXME add memcpy case
#define MULTI_CAPS_FUNC(flag, cap) \
x86/swr: use lavu helper macros to check CPU extensions Signed-off-by: James Almer <jamrial@gmail.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2014-07-03 04:16:50 +03:00
if (EXTERNAL_##flag(mm_flags)) {\
swr: add int16_to_int32_mmx/sse Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2012-04-28 13:23:42 +03:00
if( out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_S16 || out_fmt == AV_SAMPLE_FMT_S32P && in_fmt == AV_SAMPLE_FMT_S16P)\
swr: seperate functions for aligned & unaligned If someone has an idea on how to do this cleaner, its welcome Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2012-04-28 14:01:50 +03:00
ac->simd_f = ff_int16_to_int32_a_ ## cap;\
swr: int32_to_int16_mmx/sse Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2012-04-29 15:10:34 +03:00
if( out_fmt == AV_SAMPLE_FMT_S16 && in_fmt == AV_SAMPLE_FMT_S32 || out_fmt == AV_SAMPLE_FMT_S16P && in_fmt == AV_SAMPLE_FMT_S32P)\
ac->simd_f = ff_int32_to_int16_a_ ## cap;\
swr: add int16_to_int32_mmx/sse Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2012-04-28 13:23:42 +03:00
}
x86/swr: use lavu helper macros to check CPU extensions Signed-off-by: James Almer <jamrial@gmail.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2014-07-03 04:16:50 +03:00
MULTI_CAPS_FUNC(MMX, mmx)
MULTI_CAPS_FUNC(SSE2, sse2)
swr: add int32_to_float_sse2 could be done for sse/3dnow too if someone wants Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2012-04-28 18:04:42 +03:00
x86/swr: use lavu helper macros to check CPU extensions Signed-off-by: James Almer <jamrial@gmail.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2014-07-03 04:16:50 +03:00
if(EXTERNAL_MMX(mm_flags)) {
libswresample: Implement MMX, SSE4 and AVX 6ch float and int32 packing function. Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2012-05-13 20:09:01 +03:00
if(channels == 6) {
if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_FLTP || out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_S32P)
ac->simd_f = ff_pack_6ch_float_to_float_a_mmx;
}
}
x86/swr: replace sse4 instructions in pack_6ch with sse ones There's no benefit from using blendps here except on CPUs with AVX, where it's faster than shufps according to Intel's documentation. As such, rename the sse4 functions to sse/sse2 and use shufps instead. Reviewed-by: Michael Niedermayer <michaelni@gmx.at> Signed-off-by: James Almer <jamrial@gmail.com>
2014-11-07 01:43:06 +02:00
if(EXTERNAL_SSE(mm_flags)) {
if(channels == 6) {
if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_FLTP || out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_S32P)
ac->simd_f = ff_pack_6ch_float_to_float_a_sse;
x86/swr: add SSE/AVX unpack_6ch functions int32/float only Reviewed-by: Michael Niedermayer <michaelni@gmx.at> Signed-off-by: James Almer <jamrial@gmail.com>
2015-01-12 20:32:24 +02:00
if( out_fmt == AV_SAMPLE_FMT_FLTP && in_fmt == AV_SAMPLE_FMT_FLT || out_fmt == AV_SAMPLE_FMT_S32P && in_fmt == AV_SAMPLE_FMT_S32)
ac->simd_f = ff_unpack_6ch_float_to_float_a_sse;
x86/swr: replace sse4 instructions in pack_6ch with sse ones There's no benefit from using blendps here except on CPUs with AVX, where it's faster than shufps according to Intel's documentation. As such, rename the sse4 functions to sse/sse2 and use shufps instead. Reviewed-by: Michael Niedermayer <michaelni@gmx.at> Signed-off-by: James Almer <jamrial@gmail.com>
2014-11-07 01:43:06 +02:00
}
}
x86/swr: use lavu helper macros to check CPU extensions Signed-off-by: James Almer <jamrial@gmail.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2014-07-03 04:16:50 +03:00
if(EXTERNAL_SSE2(mm_flags)) {
swr: add int32_to_float_sse2 could be done for sse/3dnow too if someone wants Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2012-04-28 18:04:42 +03:00
if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_S32 || out_fmt == AV_SAMPLE_FMT_FLTP && in_fmt == AV_SAMPLE_FMT_S32P)
ac->simd_f = ff_int32_to_float_a_sse2;
swr: add int16_to_float_sse2() Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2012-04-28 19:52:48 +03:00
if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_S16 || out_fmt == AV_SAMPLE_FMT_FLTP && in_fmt == AV_SAMPLE_FMT_S16P)
ac->simd_f = ff_int16_to_float_a_sse2;
swr: float_to_int32_sse2() Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2012-04-29 12:27:22 +03:00
if( out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_FLT || out_fmt == AV_SAMPLE_FMT_S32P && in_fmt == AV_SAMPLE_FMT_FLTP)
ac->simd_f = ff_float_to_int32_a_sse2;
swr: float_to_int16_sse2() Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2012-04-29 13:18:14 +03:00
if( out_fmt == AV_SAMPLE_FMT_S16 && in_fmt == AV_SAMPLE_FMT_FLT || out_fmt == AV_SAMPLE_FMT_S16P && in_fmt == AV_SAMPLE_FMT_FLTP)
ac->simd_f = ff_float_to_int16_a_sse2;
swr: implement stereo S16/S32/FLT->S16/S32/FLT planar->packed in SSE/SSE2 Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2012-05-05 16:31:06 +03:00
if(channels == 2) {
swr/x86: 10l, missed some SSE2 instructions in code marked as SSE. Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2012-07-05 16:17:39 +03:00
if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_FLTP || out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_S32P)
ac->simd_f = ff_pack_2ch_int32_to_int32_a_sse2;
if( out_fmt == AV_SAMPLE_FMT_S16 && in_fmt == AV_SAMPLE_FMT_S16P)
ac->simd_f = ff_pack_2ch_int16_to_int16_a_sse2;
if( out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_S16P)
ac->simd_f = ff_pack_2ch_int16_to_int32_a_sse2;
if( out_fmt == AV_SAMPLE_FMT_S16 && in_fmt == AV_SAMPLE_FMT_S32P)
ac->simd_f = ff_pack_2ch_int32_to_int16_a_sse2;
if( out_fmt == AV_SAMPLE_FMT_FLTP && in_fmt == AV_SAMPLE_FMT_FLT || out_fmt == AV_SAMPLE_FMT_S32P && in_fmt == AV_SAMPLE_FMT_S32)
ac->simd_f = ff_unpack_2ch_int32_to_int32_a_sse2;
if( out_fmt == AV_SAMPLE_FMT_S16P && in_fmt == AV_SAMPLE_FMT_S16)
ac->simd_f = ff_unpack_2ch_int16_to_int16_a_sse2;
if( out_fmt == AV_SAMPLE_FMT_S32P && in_fmt == AV_SAMPLE_FMT_S16)
ac->simd_f = ff_unpack_2ch_int16_to_int32_a_sse2;
if( out_fmt == AV_SAMPLE_FMT_S16P && in_fmt == AV_SAMPLE_FMT_S32)
ac->simd_f = ff_unpack_2ch_int32_to_int16_a_sse2;
swr: implement stereo S16/S32/FLT->S16/S32/FLT planar->packed in SSE/SSE2 Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2012-05-05 16:31:06 +03:00
if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_S32P)
ac->simd_f = ff_pack_2ch_int32_to_float_a_sse2;
if( out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_FLTP)
ac->simd_f = ff_pack_2ch_float_to_int32_a_sse2;
if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_S16P)
ac->simd_f = ff_pack_2ch_int16_to_float_a_sse2;
if( out_fmt == AV_SAMPLE_FMT_S16 && in_fmt == AV_SAMPLE_FMT_FLTP)
ac->simd_f = ff_pack_2ch_float_to_int16_a_sse2;
swr-x86-simd: stereo unpack S16/S32/FLT-> S16/S32/FLT SSE/SSE2 (16 new SIMD functions) Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2012-05-06 18:22:33 +03:00
if( out_fmt == AV_SAMPLE_FMT_FLTP && in_fmt == AV_SAMPLE_FMT_S32)
ac->simd_f = ff_unpack_2ch_int32_to_float_a_sse2;
if( out_fmt == AV_SAMPLE_FMT_S32P && in_fmt == AV_SAMPLE_FMT_FLT)
ac->simd_f = ff_unpack_2ch_float_to_int32_a_sse2;
if( out_fmt == AV_SAMPLE_FMT_FLTP && in_fmt == AV_SAMPLE_FMT_S16)
ac->simd_f = ff_unpack_2ch_int16_to_float_a_sse2;
if( out_fmt == AV_SAMPLE_FMT_S16P && in_fmt == AV_SAMPLE_FMT_FLT)
ac->simd_f = ff_unpack_2ch_float_to_int16_a_sse2;
swr: implement stereo S16/S32/FLT->S16/S32/FLT planar->packed in SSE/SSE2 Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2012-05-05 16:31:06 +03:00
}
x86/swr: replace sse4 instructions in pack_6ch with sse ones There's no benefit from using blendps here except on CPUs with AVX, where it's faster than shufps according to Intel's documentation. As such, rename the sse4 functions to sse/sse2 and use shufps instead. Reviewed-by: Michael Niedermayer <michaelni@gmx.at> Signed-off-by: James Almer <jamrial@gmail.com>
2014-11-07 01:43:06 +02:00
if(channels == 6) {
if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_S32P)
ac->simd_f = ff_pack_6ch_int32_to_float_a_sse2;
if( out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_FLTP)
ac->simd_f = ff_pack_6ch_float_to_int32_a_sse2;
x86/swr: add SSE/AVX unpack_6ch functions int32/float only Reviewed-by: Michael Niedermayer <michaelni@gmx.at> Signed-off-by: James Almer <jamrial@gmail.com>
2015-01-12 20:32:24 +02:00
if( out_fmt == AV_SAMPLE_FMT_FLTP && in_fmt == AV_SAMPLE_FMT_S32)
ac->simd_f = ff_unpack_6ch_int32_to_float_a_sse2;
if( out_fmt == AV_SAMPLE_FMT_S32P && in_fmt == AV_SAMPLE_FMT_FLT)
ac->simd_f = ff_unpack_6ch_float_to_int32_a_sse2;
x86/swr: replace sse4 instructions in pack_6ch with sse ones There's no benefit from using blendps here except on CPUs with AVX, where it's faster than shufps according to Intel's documentation. As such, rename the sse4 functions to sse/sse2 and use shufps instead. Reviewed-by: Michael Niedermayer <michaelni@gmx.at> Signed-off-by: James Almer <jamrial@gmail.com>
2014-11-07 01:43:06 +02:00
}
x86/swr: make pack_8ch functions work with compilers without aligned stack Signed-off-by: James Almer <jamrial@gmail.com>
2015-02-15 18:57:02 +02:00
if(channels == 8) {
x86/swr: add SSE2/AVX pack_8ch functions Reviewed-by: Michael Niedermayer <michaelni@gmx.at> Reviewed-by: Ronald S. Bultje <rsbultje@gmail.com> Signed-off-by: James Almer <jamrial@gmail.com>
2014-12-30 23:16:37 +02:00
if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_FLTP || out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_S32P)
ac->simd_f = ff_pack_8ch_float_to_float_a_sse2;
if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_S32P)
ac->simd_f = ff_pack_8ch_int32_to_float_a_sse2;
if( out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_FLTP)
ac->simd_f = ff_pack_8ch_float_to_int32_a_sse2;
}
swr: add int32_to_float_sse2 could be done for sse/3dnow too if someone wants Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2012-04-28 18:04:42 +03:00
}
x86/swr: use lavu helper macros to check CPU extensions Signed-off-by: James Almer <jamrial@gmail.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2014-07-03 04:16:50 +03:00
if(EXTERNAL_SSSE3(mm_flags)) {
swr-x86-simd: add ff_unpack_2ch_int16_to_int16/int32/float_a_ssse3 more than 10% faster (tested on sandybridge) Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2012-05-06 20:39:52 +03:00
if(channels == 2) {
if( out_fmt == AV_SAMPLE_FMT_S16P && in_fmt == AV_SAMPLE_FMT_S16)
ac->simd_f = ff_unpack_2ch_int16_to_int16_a_ssse3;
if( out_fmt == AV_SAMPLE_FMT_S32P && in_fmt == AV_SAMPLE_FMT_S16)
ac->simd_f = ff_unpack_2ch_int16_to_int32_a_ssse3;
if( out_fmt == AV_SAMPLE_FMT_FLTP && in_fmt == AV_SAMPLE_FMT_S16)
ac->simd_f = ff_unpack_2ch_int16_to_float_a_ssse3;
}
}
x86: check for AV_CPU_FLAG_AVXSLOW where useful Signed-off-by: James Almer <jamrial@gmail.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2015-05-31 19:20:29 +02:00
if(EXTERNAL_AVX_FAST(mm_flags)) {
swr: add ff_int32_to_float_a_avx Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2012-05-03 16:52:36 +03:00
if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_S32 || out_fmt == AV_SAMPLE_FMT_FLTP && in_fmt == AV_SAMPLE_FMT_S32P)
ac->simd_f = ff_int32_to_float_a_avx;
x86: check for AV_CPU_FLAG_AVXSLOW where useful Signed-off-by: James Almer <jamrial@gmail.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2015-05-31 19:20:29 +02:00
}
if(EXTERNAL_AVX(mm_flags)) {
libswresample: Implement MMX, SSE4 and AVX 6ch float and int32 packing function. Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2012-05-13 20:09:01 +03:00
if(channels == 6) {
if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_FLTP || out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_S32P)
ac->simd_f = ff_pack_6ch_float_to_float_a_avx;
libswresample-simd: add ff_pack_6ch_int32_to_float_a_avx and ff_pack_6ch_int32_to_float_a_sse4 Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2012-05-13 21:53:30 +03:00
if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_S32P)
ac->simd_f = ff_pack_6ch_int32_to_float_a_avx;
libswresample-simd: Add ff_pack_6ch_float_to_int32_a_avx and ff_pack_6ch_float_to_int32_a_sse4 Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2012-05-13 21:56:18 +03:00
if( out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_FLTP)
ac->simd_f = ff_pack_6ch_float_to_int32_a_avx;
x86/swr: add SSE/AVX unpack_6ch functions int32/float only Reviewed-by: Michael Niedermayer <michaelni@gmx.at> Signed-off-by: James Almer <jamrial@gmail.com>
2015-01-12 20:32:24 +02:00
if( out_fmt == AV_SAMPLE_FMT_FLTP && in_fmt == AV_SAMPLE_FMT_FLT || out_fmt == AV_SAMPLE_FMT_S32P && in_fmt == AV_SAMPLE_FMT_S32)
ac->simd_f = ff_unpack_6ch_float_to_float_a_avx;
if( out_fmt == AV_SAMPLE_FMT_FLTP && in_fmt == AV_SAMPLE_FMT_S32)
ac->simd_f = ff_unpack_6ch_int32_to_float_a_avx;
if( out_fmt == AV_SAMPLE_FMT_S32P && in_fmt == AV_SAMPLE_FMT_FLT)
ac->simd_f = ff_unpack_6ch_float_to_int32_a_avx;
libswresample: Implement MMX, SSE4 and AVX 6ch float and int32 packing function. Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2012-05-13 20:09:01 +03:00
}
x86/swr: make pack_8ch functions work with compilers without aligned stack Signed-off-by: James Almer <jamrial@gmail.com>
2015-02-15 18:57:02 +02:00
if(channels == 8) {
x86/swr: add SSE2/AVX pack_8ch functions Reviewed-by: Michael Niedermayer <michaelni@gmx.at> Reviewed-by: Ronald S. Bultje <rsbultje@gmail.com> Signed-off-by: James Almer <jamrial@gmail.com>
2014-12-30 23:16:37 +02:00
if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_FLTP || out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_S32P)
ac->simd_f = ff_pack_8ch_float_to_float_a_avx;
if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_S32P)
ac->simd_f = ff_pack_8ch_int32_to_float_a_avx;
if( out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_FLTP)
ac->simd_f = ff_pack_8ch_float_to_int32_a_avx;
}
swr: add ff_int32_to_float_a_avx Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2012-05-03 16:52:36 +03:00
}
x86: use the new helper macros where useful Reviewed-by: Michael Niedermayer <michael@niedermayer.cc> Signed-off-by: James Almer <jamrial@gmail.com>
2016-02-07 05:52:06 +02:00
if(EXTERNAL_AVX2_FAST(mm_flags)) {
x86/swr: add ff_float_to_int32_a_avx2 13797 decicycles in ff_float_to_int32_a_sse2, 32768 runs, 0 skips 8603 decicycles in ff_float_to_int32_a_avx2, 32766 runs, 2 skips Reviewed-by: Christophe Gisquet <christophe.gisquet@gmail.com> Signed-off-by: James Almer <jamrial@gmail.com>
2014-11-06 22:38:37 +02:00
if( out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_FLT || out_fmt == AV_SAMPLE_FMT_S32P && in_fmt == AV_SAMPLE_FMT_FLTP)
ac->simd_f = ff_float_to_int32_a_avx2;
}
swr: add int16_to_int32_mmx/sse Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2012-04-28 13:23:42 +03:00
}
Reference in New Issue Copy Permalink