---
libavresample/x86/audio_convert.asm | 52 ++++++++++++++++++++++++++++++++
libavresample/x86/audio_convert_init.c | 9 +++++
libavutil/x86/x86util.asm | 6 ++++
3 files changed, 67 insertions(+), 0 deletions(-)
diff --git a/libavresample/x86/audio_convert.asm
b/libavresample/x86/audio_convert.asm
index 17c9f22..34449c9 100644
--- a/libavresample/x86/audio_convert.asm
+++ b/libavresample/x86/audio_convert.asm
@@ -589,6 +589,58 @@ CONV_FLTP_TO_S16_6CH
INIT_XMM sse2
CONV_FLTP_TO_S16_6CH
+;------------------------------------------------------------------------------
+; void ff_conv_fltp_to_flt_2ch(float *dst, float *const *src, int len,
+; int channels);
+;------------------------------------------------------------------------------
+
+%macro CONV_FLTP_TO_FLT_2CH 0
+cglobal conv_fltp_to_flt_2ch, 3,4,5, dst, src, len, src1
+ mov src1q, [srcq+gprsize]
+ mov srcq, [srcq ]
+ sub src1q, srcq
+ ALIGN 16
+.loop
+ mova m0, [srcq ]
+ mova m1, [srcq+src1q ]
+ mova m2, [srcq +mmsize]
+ mova m3, [srcq+src1q+mmsize]
+ SBUTTERFLYPS2 0, 1, 4
+ SBUTTERFLYPS2 2, 3, 4
+%if cpuflag(avx)
+ vextractf128 [dstq ], m0, 0
+ vextractf128 [dstq+ 16], m1, 0
+ vextractf128 [dstq+ 32], m0, 1
+ vextractf128 [dstq+ 48], m1, 1
+ vextractf128 [dstq+ 64], m2, 0
+ vextractf128 [dstq+ 80], m3, 0
+ vextractf128 [dstq+ 96], m2, 1
+ vextractf128 [dstq+112], m3, 1
+%else
+ mova [dstq ], m0
+ mova [dstq+1*mmsize], m1
+ mova [dstq+2*mmsize], m2
+ mova [dstq+3*mmsize], m3
+%endif
+ add srcq, mmsize*2
+ add dstq, mmsize*4
+ sub lend, mmsize/2
+ jg .loop
+%if mmsize == 32
+ vzeroupper
+ RET
+%else
+ REP_RET
+%endif
+%endmacro
+
+INIT_XMM sse
+CONV_FLTP_TO_FLT_2CH
+%if HAVE_AVX
+INIT_YMM avx
+CONV_FLTP_TO_FLT_2CH
+%endif
+
;-----------------------------------------------------------------------------
; void ff_conv_fltp_to_flt_6ch(float *dst, float *const *src, int len,
; int channels);
diff --git a/libavresample/x86/audio_convert_init.c
b/libavresample/x86/audio_convert_init.c
index 72c78bd..6544678 100644
--- a/libavresample/x86/audio_convert_init.c
+++ b/libavresample/x86/audio_convert_init.c
@@ -62,6 +62,11 @@ extern void ff_conv_fltp_to_s16_6ch_sse (int16_t *dst, float
*const *src,
extern void ff_conv_fltp_to_s16_6ch_sse2(int16_t *dst, float *const *src,
int len, int channels);
+extern void ff_conv_fltp_to_flt_2ch_sse(float *dst, float *const *src, int len,
+ int channels);
+extern void ff_conv_fltp_to_flt_2ch_avx(float *dst, float *const *src, int len,
+ int channels);
+
extern void ff_conv_fltp_to_flt_6ch_mmx (float *dst, float *const *src, int
len,
int channels);
extern void ff_conv_fltp_to_flt_6ch_sse4(float *dst, float *const *src, int
len,
@@ -83,6 +88,8 @@ av_cold void ff_audio_convert_init_x86(AudioConvert *ac)
if (mm_flags & AV_CPU_FLAG_SSE && HAVE_SSE) {
ff_audio_convert_set_func(ac, AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_FLTP,
6, 1, 2, "SSE", ff_conv_fltp_to_s16_6ch_sse);
+ ff_audio_convert_set_func(ac, AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_FLTP,
+ 2, 16, 8, "SSE",
ff_conv_fltp_to_flt_2ch_sse);
}
if (mm_flags & AV_CPU_FLAG_SSE2 && HAVE_SSE) {
if (!(mm_flags & AV_CPU_FLAG_SSE2SLOW)) {
@@ -124,6 +131,8 @@ av_cold void ff_audio_convert_init_x86(AudioConvert *ac)
ff_audio_convert_set_func(ac, AV_SAMPLE_FMT_S32, AV_SAMPLE_FMT_FLT,
0, 32, 32, "AVX", ff_conv_flt_to_s32_avx);
ff_audio_convert_set_func(ac, AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_FLTP,
+ 2, 32, 16, "AVX",
ff_conv_fltp_to_flt_2ch_avx);
+ ff_audio_convert_set_func(ac, AV_SAMPLE_FMT_FLT, AV_SAMPLE_FMT_FLTP,
6, 16, 4, "AVX",
ff_conv_fltp_to_flt_6ch_avx);
}
#endif
diff --git a/libavutil/x86/x86util.asm b/libavutil/x86/x86util.asm
index 066384b..26a92bc 100644
--- a/libavutil/x86/x86util.asm
+++ b/libavutil/x86/x86util.asm
@@ -47,6 +47,12 @@
SWAP %1, %3, %2
%endmacro
+%macro SBUTTERFLYPS2 3
+ unpcklps m%3, m%1, m%2
+ unpckhps m%1, m%1, m%2
+ SWAP %1, %3, %2
+%endmacro
+
%macro TRANSPOSE4x4B 5
SBUTTERFLY bw, %1, %2, %5
SBUTTERFLY bw, %3, %4, %5
--
1.7.1
_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel
