[MEncoder-users] Performance on multi core machines - a specific enquiry re avisynth mt input

[Michael Rozdoba]

RC wrote:

> mencoder -nosound -ovc x264enc -x264encopts
> pass=(1/2):turbo=2:threads=auto:bitrate=1000 (Input) -o (Output.avi)

Using the build James linked to http://oss.netfarm.it/mplayer-win32.php 
and a 20s mpeg2 clip, I get 96% cpu for the first pass & 100% for the 
second.

Repeating the original:
mencoder.exe "E:\Work\test\fm-deblock.avs" -o "hfyu_fm-deblock.avi" -of 
avi -forceidx -ovc lavc -lavcopts 
codec=ffvhuff:vstrict=-2:pred=2:context=1:threads=4

with this new MEncoder build, I still get a 25% cpu threshold & 3fps.

Similarly, if I run the MEncoder x264 encode above using the 
fm-deblock.avs input, I am stuck at the 25% cpu limit.

So, it isn't specific to the Megui build of MEncoder, but is specific to 
having an input to MEncoder which takes a lot of power to decode.

Whoever's responsible, it does seem that the decoding of the input to 
MEncoder is tied to a single thread.

It would be good to confirm this isn't specific to an avisynth input, 
but to do that I'd need an alternative video source which is very costly 
to decode & I can't think of one.

Can anyone comment further?

-- 
Michael Rozdoba