[FFmpeg-user] Performance issue/question (maybe my misunderstanding)

[Frank Häfemeier]

Am Montag, den 20.08.2012, 10:17 -0600 schrieb Roger Pack:
> > I have two computer. An older one with an 'Intel(R) Core(TM)2 Duo CPU
> > E8400 @ 3.00GHz' and a brand new 'Intel(R) Core(TM) i7-3930K CPU @
> > 3.20GHz'. The first has two cores (2 threads, 6M cache), the second 6
> > cores (12 threads, 12M cache).
> >
> > My thoughts are the second should be faster to transcode videos as the
> > first one. But it isn't! My new computer was <10% faster as the other.
> > It seems to me there are some parameter which are influent the process.
> > But I have no idea which?
> 
> real	355m8.107s
> 
> whoa! That's some big file.  I assume your computers are both
> desktops? What if you try it with -threads 1? Are you possibly limited
> by disk speed?
The source file has a size of 2.5 GB and is an one hour video captured
from our local cable network. This is not too big I think to get a big
influence from disk speed. But in my opinion it does not explain why
aren't the difference in speed between both machine are so small.
The absolute time is not the important part here. It was transcode with
the slowest preset I could found for libx264.
If you compare the real time of both computer you will see the new
computer is 14% faster as the old one. And I think this is too small.
But I have no idea at the moment if there is a logic explanation about
it or only I have a complete wrong understanding of it.
That is the reason to start the discussion here. I will also start in
the next days a similar discussion in a x264 mailing list to collect
more information.
Because I think it is not a specific ffmpeg or libx264 issue. Or
possible it is... I am confused:-|

I will start a new test with -thread 1 and the same source file soon.
I'll keep you informed.
> _______________________________________________
> ffmpeg-user mailing list
> ffmpeg-user at ffmpeg.org
> http://ffmpeg.org/mailman/listinfo/ffmpeg-user