diff options
| author | Michael Niedermayer <michaelni@gmx.at> | 2012-06-18 20:05:32 +0200 |
|---|---|---|
| committer | Michael Niedermayer <michaelni@gmx.at> | 2012-06-18 20:07:00 +0200 |
| commit | 82edf6727f0663601351081ca1e4fb20d1752972 (patch) | |
| tree | 12479c3ec8cedfa0ec4dda38a72023224f2b5b73 /libavresample/x86/audio_mix.asm | |
| parent | f87dacb27de93f995cb18f9dcc73581ef8fc157b (diff) | |
| parent | f61ce90caa909d131ea6ec205823568a38115529 (diff) | |
| download | ffmpeg-82edf6727f0663601351081ca1e4fb20d1752972.tar.gz | |
Merge remote-tracking branch 'qatar/master'
* qatar/master:
lavr: add x86-optimized functions for mixing 1-to-2 s16p with flt coeffs
lavr: add x86-optimized functions for mixing 1-to-2 fltp with flt coeffs
Add Dolby/DPLII downmix support to libavresample
vorbisdec: replace div/mod in loop with a counter
fate: vorbis: add 5.1 surround test
rtpenc: Allow requesting H264 RTP packetization mode 0
configure: Sort the library listings in the help text alphabetically
dwt: remove variable-length arrays
RTMPT protocol support
http: Properly handle chunked transfer-encoding for replies to post data
http: Fail reading if the connection has gone away
amr: Mark an array const
amr: More space cleanup
rtpenc: Fix memory leaks in the muxer open function
Conflicts:
Changelog
configure
doc/APIchanges
libavformat/version.h
Merged-by: Michael Niedermayer <michaelni@gmx.at>
Diffstat (limited to 'libavresample/x86/audio_mix.asm')
| -rw-r--r-- | libavresample/x86/audio_mix.asm | 81 |
1 files changed, 81 insertions, 0 deletions
diff --git a/libavresample/x86/audio_mix.asm b/libavresample/x86/audio_mix.asm index 8a4cf061cd..4b0434dd6d 100644 --- a/libavresample/x86/audio_mix.asm +++ b/libavresample/x86/audio_mix.asm @@ -150,3 +150,84 @@ cglobal mix_2_to_1_s16p_q8, 3,4,6, src, matrix, len, src1 sub lend, mmsize/2 jg .loop REP_RET + +;----------------------------------------------------------------------------- +; void ff_mix_1_to_2_fltp_flt(float **src, float **matrix, int len, +; int out_ch, int in_ch); +;----------------------------------------------------------------------------- + +%macro MIX_1_TO_2_FLTP_FLT 0 +cglobal mix_1_to_2_fltp_flt, 3,5,4, src0, matrix0, len, src1, matrix1 + mov src1q, [src0q+gprsize] + mov src0q, [src0q] + sub src1q, src0q + mov matrix1q, [matrix0q+gprsize] + mov matrix0q, [matrix0q] + VBROADCASTSS m2, [matrix0q] + VBROADCASTSS m3, [matrix1q] + ALIGN 16 +.loop: + mova m0, [src0q] + mulps m1, m0, m3 + mulps m0, m0, m2 + mova [src0q ], m0 + mova [src0q+src1q], m1 + add src0q, mmsize + sub lend, mmsize/4 + jg .loop + REP_RET +%endmacro + +INIT_XMM sse +MIX_1_TO_2_FLTP_FLT +%if HAVE_AVX +INIT_YMM avx +MIX_1_TO_2_FLTP_FLT +%endif + +;----------------------------------------------------------------------------- +; void ff_mix_1_to_2_s16p_flt(int16_t **src, float **matrix, int len, +; int out_ch, int in_ch); +;----------------------------------------------------------------------------- + +%macro MIX_1_TO_2_S16P_FLT 0 +cglobal mix_1_to_2_s16p_flt, 3,5,6, src0, matrix0, len, src1, matrix1 + mov src1q, [src0q+gprsize] + mov src0q, [src0q] + sub src1q, src0q + mov matrix1q, [matrix0q+gprsize] + mov matrix0q, [matrix0q] + VBROADCASTSS m4, [matrix0q] + VBROADCASTSS m5, [matrix1q] + ALIGN 16 +.loop: + mova m0, [src0q] + S16_TO_S32_SX 0, 2 + cvtdq2ps m0, m0 + cvtdq2ps m2, m2 + mulps m1, m0, m5 + mulps m0, m0, m4 + mulps m3, m2, m5 + mulps m2, m2, m4 + cvtps2dq m0, m0 + cvtps2dq m1, m1 + cvtps2dq m2, m2 + cvtps2dq m3, m3 + packssdw m0, m2 + packssdw m1, m3 + mova [src0q ], m0 + mova [src0q+src1q], m1 + add src0q, mmsize + sub lend, mmsize/2 + jg .loop + REP_RET +%endmacro + +INIT_XMM sse2 +MIX_1_TO_2_S16P_FLT +INIT_XMM sse4 +MIX_1_TO_2_S16P_FLT +%if HAVE_AVX +INIT_XMM avx +MIX_1_TO_2_S16P_FLT +%endif |