diff options
| author | Michael Niedermayer <michaelni@gmx.at> | 2012-12-06 14:33:38 +0100 |
|---|---|---|
| committer | Michael Niedermayer <michaelni@gmx.at> | 2012-12-06 14:33:38 +0100 |
| commit | 15784c2bab56508565771cd04b5dda64d6717953 (patch) | |
| tree | a0eade0816e7ea46a08a4cf7ffd2b25e6666c59b /libavutil/x86/float_dsp.asm | |
| parent | 32aedebdc59d5b34ab7a9137855dcc602267e00f (diff) | |
| parent | 9d5c62ba5b586c80af508b5914934b1c439f6652 (diff) | |
| download | ffmpeg-15784c2bab56508565771cd04b5dda64d6717953.tar.gz | |
Merge commit '9d5c62ba5b586c80af508b5914934b1c439f6652'
* commit '9d5c62ba5b586c80af508b5914934b1c439f6652':
lavu/opt: do not filter out the initial sign character except for flags
eval: treat dB as decibels instead of decibytes
float_dsp: add vector_dmul_scalar() to multiply a vector of doubles
Conflicts:
libavutil/eval.c
tests/ref/fate/eval
Merged-by: Michael Niedermayer <michaelni@gmx.at>
Diffstat (limited to 'libavutil/x86/float_dsp.asm')
| -rw-r--r-- | libavutil/x86/float_dsp.asm | 45 |
1 files changed, 45 insertions, 0 deletions
diff --git a/libavutil/x86/float_dsp.asm b/libavutil/x86/float_dsp.asm index a3200c64b6..365bafea00 100644 --- a/libavutil/x86/float_dsp.asm +++ b/libavutil/x86/float_dsp.asm @@ -120,3 +120,48 @@ cglobal vector_fmul_scalar, 4,4,3, dst, src, mul, len INIT_XMM sse VECTOR_FMUL_SCALAR + +;------------------------------------------------------------------------------ +; void ff_vector_dmul_scalar(double *dst, const double *src, double mul, +; int len) +;------------------------------------------------------------------------------ + +%macro VECTOR_DMUL_SCALAR 0 +%if UNIX64 +cglobal vector_dmul_scalar, 3,3,3, dst, src, len +%else +cglobal vector_dmul_scalar, 4,4,3, dst, src, mul, len +%endif +%if ARCH_X86_32 + VBROADCASTSD m0, mulm +%else +%if WIN64 + movlhps xmm2, xmm2 +%if cpuflag(avx) + vinsertf128 ymm2, ymm2, xmm2, 1 +%endif + SWAP 0, 2 +%else + movlhps xmm0, xmm0 +%if cpuflag(avx) + vinsertf128 ymm0, ymm0, xmm0, 1 +%endif +%endif +%endif + lea lenq, [lend*8-2*mmsize] +.loop: + mulpd m1, m0, [srcq+lenq ] + mulpd m2, m0, [srcq+lenq+mmsize] + mova [dstq+lenq ], m1 + mova [dstq+lenq+mmsize], m2 + sub lenq, 2*mmsize + jge .loop + REP_RET +%endmacro + +INIT_XMM sse2 +VECTOR_DMUL_SCALAR +%if HAVE_AVX_EXTERNAL +INIT_YMM avx +VECTOR_DMUL_SCALAR +%endif |