diff options
| author | Ganesh Ajjanagadde <gajjanagadde@gmail.com> | 2015-12-21 19:05:00 -0800 |
|---|---|---|
| committer | Ganesh Ajjanagadde <gajjanagadde@gmail.com> | 2015-12-23 09:22:59 -0800 |
| commit | e29db08cf7f68fefe18f043250e2292aaf91ef3b (patch) | |
| tree | d69978eab4ed9dbf310009a7aa64bfba794da728 /libavutil/libm.h | |
| parent | 520a5d33f0ea9f8838dbc7282470db700d248065 (diff) | |
| download | ffmpeg-e29db08cf7f68fefe18f043250e2292aaf91ef3b.tar.gz | |
lavu/libm: add exp10 support
exp10 is a function available in GNU libm. Looks like no other common
libm has it. This adds support for it to FFmpeg.
There are essentially 2 ways of handling the fallback:
1. Using pow(10, x)
2. Using exp2(M_LOG2_10 * x).
First one represents a Pareto improvement, with no speed or accuracy
regression anywhere, but speed improvement limited to GNU libm.
Second one represents a slight accuracy loss (relative error ~ 1e-13)
for non GNU libm. Speedup of > 2x is obtained on non GNU libm platforms,
~30% on GNU libm. These are "average case numbers", another benefit is
the lack of triggering of the well-known terrible worst case paths
through pow.
Based on reviews, second one chosen. Comment added accordingly.
Reviewed-by: Hendrik Leppkes <h.leppkes@gmail.com>
Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Reviewed-by: Ronald S. Bultje <rsbultje@gmail.com>
Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>
Diffstat (limited to 'libavutil/libm.h')
| -rw-r--r-- | libavutil/libm.h | 19 |
1 files changed, 19 insertions, 0 deletions
diff --git a/libavutil/libm.h b/libavutil/libm.h index 146768aac6..6f9ac1b349 100644 --- a/libavutil/libm.h +++ b/libavutil/libm.h @@ -29,6 +29,7 @@ #include "config.h" #include "attributes.h" #include "intfloat.h" +#include "mathematics.h" #if HAVE_MIPSFPU && HAVE_INLINE_ASM #include "libavutil/mips/libm_mips.h" @@ -292,6 +293,24 @@ static inline double erf(double z) #define exp2f(x) ((float)exp2(x)) #endif /* HAVE_EXP2F */ +/* Somewhat inaccurate fallbacks, relative error ~ 1e-13 concentrated on very +small and very large values. For perfection accuracy-wise, should use pow. +Speed benefits (>2x average, with no super slow paths) deemed to be worth the +accuracy tradeoff */ +#if !HAVE_EXP10 +static av_always_inline double exp10(double x) +{ + return exp2(M_LOG2_10 * x); +} +#endif /* HAVE_EXP10 */ + +#if !HAVE_EXP10F +static av_always_inline float exp10f(float x) +{ + return exp2f(M_LOG2_10 * x); +} +#endif /* HAVE_EXP10F */ + #if !HAVE_ISINF #undef isinf /* Note: these do not follow the BSD/Apple/GNU convention of returning -1 for |