[MEncoder-users] x264 file playback is too slow!
Corey Hickey
bugfood-ml at fatooh.org
Fri Mar 24 07:57:33 CET 2006
The Wanderer wrote:
>> 2) It runs very slowly and is generally rather odd. Here is the full
>> mplayer -v output attached. See how it complains that I don't have
>> -vo null etc.
>
> No, I don't see where it does that.
>
> The only messages I see involving the word 'null' are standard,
> unremarkable (if overly verbose) status information. The only
> complaining messages I see in the initialization section are the ones
> involving trying to detect various container formats and failing on the
> first several (which makes sense, since the file isn't in any of those
> formats). The only message I see in the initialization section which you
> might - or might not - want to pay any attention to is the informational
> one about the "waveheader" suboption.
>
>> and then says my system is too slow to play the file!
>
> This is quite strange, since with '-vo null' and '-vc null' you have
> eliminated two of the biggest CPU-hogging parts of playback. The only
> remaining plausible cause I can think of offhand is slow media - in
> other words, that the drive you are reading the file from is not
> providing the data to MPlayer quickly enough, and MPlayer's internal
> buffers are emptying out. To work around this, the only thing I know of
> which you can do which is not too complicated for me to go over on a
> mailing list is play around with the '-cache' option.
I just tested it, and, now that you mention it, it does seem rather
slow. MPlayer isn't using much CPU time or stressing the disk much.
Since oggenc doesn't need to seek, though, you can use a fifo and
generally be much more efficient about it.
$ mkfifo audio.fifo
$ oggenc audio.fifo -o audio.ogg
At this point, oggenc will hang until audio.fifo gets written to. Switch
to another terminal and run
$ mplayer -ao pcm:file=audio.fifo ....
That'll save you some file-writing overhead, lots of disk space, and
some time.
-Corey
More information about the MEncoder-users
mailing list