[FFmpeg-devel] [PATCH] ra144enc: fix use of scalarprod_int16

Christophe Gisquet christophe.gisquet at gmail.com
Wed Mar 5 14:43:34 CET 2014


Hi,

2014-03-05 14:31 GMT+01:00 Michael Niedermayer <michaelni at gmx.at>:
> also made the emms more conditional, though iam not totally happy
> about my solution its kind of ugly ...

Seeing your 2nd change, I'm wondering if this check is resolved during
compilation. If not, then it's highly predictable, but indeed.

I thought emms was only expensive if the FPU had actually been
"dirtied" by MMX, and rather fast on new CPUs, hence the unconditional
call. But I haven't benchmarked either.

-- 
Christophe


More information about the ffmpeg-devel mailing list