[FFmpeg-user] Reduce CPU usage to match VLC

Loadlinx limiteddi at gmail.com
Mon Feb 10 17:34:29 CET 2014


Tried with 5 processes and it breaks the picture.


On Mon, Feb 10, 2014 at 10:58 AM, Reindl Harald <h.reindl at thelounge.net>wrote:

> do you *think* that or have you tried it?
>
> you can't calculate "booh CPU usage multiplied with cores gives streams"
> the one and only question is if the overall load can handle
> the streams without drop frames and how the OS will schedule
> the threads and processes
>
> the final goal is that any of your cores are 100% loaded and
> nothing breaks down
>
> Am 10.02.2014 16:51, schrieb Loadlinx:
> > At 120% i am still limited to only ~4 transcoding sessions.
> > Maybe the quality is not the same and maybe ffmpeg is faster.
> >
> > I am willing to sacrifice quality for lower CPU usage. Is this possible?
> I
> > tried -preset:ultrafast but no difference in CPU usage observed.
> >
> >
> > On Mon, Feb 10, 2014 at 10:46 AM, Carl Eugen Hoyos <cehoyos at ag.or.at>
> wrote:
> >
> >> Loadlinx <limiteddi <at> gmail.com> writes:
> >>
> >>> /vlc -vvv http://192.168.1.50:6002 --sout
> >>> '#transcode{venc=x264{preset=ultrafast}
> >>                                ^^^^^^^^^
> >>
> >>> ./ffmpeg -i http://192.168.1.50:6002 <http://162.219.0.246:12709>
> >>> -vcodec libx264 -vb 1500k -s 480x320 -acodec libfdk_aac -ab 32k
> >>> -f flv rtmp://192.168.1.50:1935/live/709
> >>
> >> (Complete, uncut console output missing.)
> >>
> >> Aren't these two completely different things?
> >> Or does libx264 nowadays default to ultrafast?
> >>
> >>> Quality seems to me somewhat the same
> >>
> >> somewhat: maybe
> >> the same: most likely not
>
>
> _______________________________________________
> ffmpeg-user mailing list
> ffmpeg-user at ffmpeg.org
> http://ffmpeg.org/mailman/listinfo/ffmpeg-user
>
>


More information about the ffmpeg-user mailing list