Commit c1fe2db3 authored by Michael Niedermayer's avatar Michael Niedermayer

swr: add ff_int32_to_float_a_avx

Signed-off-by: 's avatarMichael Niedermayer <michaelni@gmx.at>
parent 29ed1900
...@@ -22,7 +22,7 @@ ...@@ -22,7 +22,7 @@
%include "libavutil/x86/x86util.asm" %include "libavutil/x86/x86util.asm"
SECTION_RODATA SECTION_RODATA
align 32
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 flt2p15 : times 8 dd 32768.0
...@@ -87,8 +87,8 @@ int32_to_float_u_int %+ SUFFIX ...@@ -87,8 +87,8 @@ int32_to_float_u_int %+ SUFFIX
cvtdq2ps m0, m0 cvtdq2ps m0, m0
cvtdq2ps m1, m1 cvtdq2ps m1, m1
%endif %endif
mulps m0, m2 mulps m0, m0, m2
mulps m1, m2 mulps m1, m1, m2
mov%1 [ dstq+lenq], m0 mov%1 [ dstq+lenq], m0
mov%1 [mmsize + dstq+lenq], m1 mov%1 [mmsize + dstq+lenq], m1
add lenq, 2*mmsize add lenq, 2*mmsize
...@@ -243,3 +243,9 @@ FLOAT_TO_INT32 u ...@@ -243,3 +243,9 @@ FLOAT_TO_INT32 u
FLOAT_TO_INT32 a FLOAT_TO_INT32 a
FLOAT_TO_INT16 u FLOAT_TO_INT16 u
FLOAT_TO_INT16 a FLOAT_TO_INT16 a
%if HAVE_AVX
INIT_YMM avx
INT32_TO_FLOAT u
INT32_TO_FLOAT a
%endif
...@@ -33,6 +33,8 @@ void ff_int16_to_float_a_sse2(uint8_t **dst, const uint8_t **src, int len); ...@@ -33,6 +33,8 @@ 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 ff_float_to_int16_a_sse2(uint8_t **dst, const uint8_t **src, int len);
void ff_int32_to_float_a_avx(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,
enum AVSampleFormat in_fmt, enum AVSampleFormat in_fmt,
...@@ -64,4 +66,8 @@ MULTI_CAPS_FUNC(AV_CPU_FLAG_SSE, sse) ...@@ -64,4 +66,8 @@ MULTI_CAPS_FUNC(AV_CPU_FLAG_SSE, sse)
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) 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; ac->simd_f = ff_float_to_int16_a_sse2;
} }
if(HAVE_AVX && mm_flags & AV_CPU_FLAG_AVX) {
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;
}
} }
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment