diff options
author | Claudio Freire <klaussfreire@gmail.com> | 2016-01-14 00:38:22 -0300 |
---|---|---|
committer | Claudio Freire <klaussfreire@gmail.com> | 2016-01-14 00:38:22 -0300 |
commit | 62dfe1d40d87f8f67cd77d4b769b7c6163083c5e (patch) | |
tree | 1b127ce3b9f6e58d2ef943d4070d8642b7029b9f /libavcodec/aacenc_utils.h | |
parent | a0a47a09b0e204216072c1f77643de3f6f869732 (diff) | |
download | ffmpeg-streaming-62dfe1d40d87f8f67cd77d4b769b7c6163083c5e.zip ffmpeg-streaming-62dfe1d40d87f8f67cd77d4b769b7c6163083c5e.tar.gz |
avcodec/aacenc_is: replace pow(x, 0.75) by x/sqrtf(sqrtf(x))
This is quite an accurate approximation; testing shows ~ 2ulp error in
the floating point result. Tested with FATE.
Alternatively, if one wants "full accuracy", one can use powf, or sqrt
instead of sqrtf. With powf, one gets 1 ulp error (theoretically should be 0, as
0.75 is exactly representable) on GNU libm, with sqrt, 0 ulp error.
Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>
Abstracted into pos_pow34 utility function
Signed-off-by: Claudio Freire <klaussfreire@gmail.com>
Diffstat (limited to 'libavcodec/aacenc_utils.h')
-rw-r--r-- | libavcodec/aacenc_utils.h | 5 |
1 files changed, 5 insertions, 0 deletions
diff --git a/libavcodec/aacenc_utils.h b/libavcodec/aacenc_utils.h index 736e4a0..cb5bc8d 100644 --- a/libavcodec/aacenc_utils.h +++ b/libavcodec/aacenc_utils.h @@ -45,6 +45,11 @@ static inline void abs_pow34_v(float *out, const float *in, const int size) } } +static inline float pos_pow34(float a) +{ + return sqrtf(a * sqrtf(a)); +} + /** * Quantize one coefficient. * @return absolute value of the quantized coefficient |