[FFmpeg-cvslog] lavu/internal: add ff_exp10

[Ganesh Ajjanagadde]

ffmpeg | branch: master | Ganesh Ajjanagadde <gajjanagadde at gmail.com> | Fri Dec 25 08:07:08 2015 -0800| [11388b5a8f187d5b7c8a28437e95762485f51ed8] | committer: Ganesh Ajjanagadde

lavu/internal: add ff_exp10

Fast, reasonably accurate 10^x. Alternative of detection of libm exp10 at configure
time is not worth the trouble, since it is anyway not POSIX or ISO C,
and currently only the GNU libm has it. Furthermore, GNU libm's variant
is ~ 2x slower, and is ironically not correctly rounded (2 ulp off) to justify all
that slowdown.

Reviewed-by: James Almer <jamrial at gmail.com>
Reviewed-by: Hendrik Leppkes <h.leppkes at gmail.com>
Reviewed-by: Michael Niedermayer <michael at niedermayer.cc>
Signed-off-by: Ganesh Ajjanagadde <gajjanagadde at gmail.com>

> http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=11388b5a8f187d5b7c8a28437e95762485f51ed8
---

 libavutil/internal.h |   19 +++++++++++++++++++
 1 file changed, 19 insertions(+)

diff --git a/libavutil/internal.h b/libavutil/internal.h
index f86b7fb..076aa47 100644
--- a/libavutil/internal.h
+++ b/libavutil/internal.h
@@ -292,6 +292,25 @@ static av_always_inline av_const int64_t ff_rint64_clip(double a, int64_t amin,
     return res;
 }
 
+/**
+ * Compute 10^x for floating point values. Note: this function is by no means
+ * "correctly rounded", and is meant as a fast, reasonably accurate approximation.
+ * For instance, maximum relative error for the double precision variant is
+ * ~ 1e-13 for very small and very large values.
+ * This is ~2x faster than GNU libm's approach, which is still off by 2ulp on
+ * some inputs.
+ * @param x exponent
+ * @return 10^x
+ */
+static av_always_inline double ff_exp10(double x)
+{
+    return exp2(M_LOG2_10 * x);
+}
+
+static av_always_inline float ff_exp10f(float x)
+{
+    return exp2f(M_LOG2_10 * x);
+}
 
 /**
  * A wrapper for open() setting O_CLOEXEC.