RepoMirrors
/
ffmpeg
mirror of https://git.ffmpeg.org/ffmpeg.git
1
0
Fork 0
Code Issues Projects Releases Wiki Activity

swr: float_to_int16_sse2() 

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
This commit is contained in:
Michael Niedermayer 2012-04-29 12:18:14 +02:00
parent 5932938c9a
commit 73edb58c3c
2 changed files with 36 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

33
libswresample/x86/audio_convert.asm
View File

@ -25,6 +25,7 @@ SECTION_RODATA
flt2pm31: times 8 dd 4.6566129e-10 flt2pm31: times 8 dd 4.6566129e-10
flt2p31 : times 8 dd 2147483648.0 flt2p31 : times 8 dd 2147483648.0
flt2p15 : times 8 dd 32768.0
SECTION .text SECTION .text
@ -163,6 +164,36 @@ float_to_int32_u_int %+ SUFFIX
REP_RET REP_RET
%endmacro %endmacro
%macro FLOAT_TO_INT16 1
cglobal float_to_int16_%1, 3, 3, 3, dst, src, len
mov srcq, [srcq]
mov dstq, [dstq]
%ifidn %1, a
test dstq, mmsize-1
jne float_to_int16_u_int %+ SUFFIX
test srcq, mmsize-1
jne float_to_int16_u_int %+ SUFFIX
%else
float_to_int16_u_int %+ SUFFIX
%endif
lea srcq, [srcq + 2*lenq]
add dstq, lenq
neg lenq
mova m2, [flt2p15]
.next:
mov%1 m0, [ srcq+2*lenq]
mov%1 m1, [mmsize + srcq+2*lenq]
mulps m0, m2
mulps m1, m2
cvtps2dq m0, m0
cvtps2dq m1, m1
packssdw m0, m1
mov%1 [ dstq+lenq], m0
add lenq, mmsize
jl .next
REP_RET
%endmacro
INIT_MMX mmx INIT_MMX mmx
INT16_TO_INT32 u INT16_TO_INT32 u
@ -179,3 +210,5 @@ INT16_TO_FLOAT u
INT16_TO_FLOAT a INT16_TO_FLOAT a
FLOAT_TO_INT32 u FLOAT_TO_INT32 u
FLOAT_TO_INT32 a FLOAT_TO_INT32 a
FLOAT_TO_INT16 u
FLOAT_TO_INT16 a

3
libswresample/x86/swresample_x86.c
View File

@ -29,6 +29,7 @@ MULTI_CAPS_FUNC_DECL(sse)
void ff_int32_to_float_a_sse2(uint8_t **dst, const uint8_t **src, int len); void ff_int32_to_float_a_sse2(uint8_t **dst, const uint8_t **src, int len);
void ff_int16_to_float_a_sse2(uint8_t **dst, const uint8_t **src, int len); void ff_int16_to_float_a_sse2(uint8_t **dst, const uint8_t **src, int len);
void ff_float_to_int32_a_sse2(uint8_t **dst, const uint8_t **src, int len); void ff_float_to_int32_a_sse2(uint8_t **dst, const uint8_t **src, int len);
void ff_float_to_int16_a_sse2(uint8_t **dst, const uint8_t **src, int len);
void swri_audio_convert_init_x86(struct AudioConvert *ac, void swri_audio_convert_init_x86(struct AudioConvert *ac,
enum AVSampleFormat out_fmt, enum AVSampleFormat out_fmt,
@ -56,5 +57,7 @@ MULTI_CAPS_FUNC(AV_CPU_FLAG_SSE, sse)
ac->simd_f = ff_int16_to_float_a_sse2; ac->simd_f = ff_int16_to_float_a_sse2;
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) 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; ac->simd_f = ff_float_to_int32_a_sse2;
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;
} }
} }