On Sat, May 14, 2005 at 03:18:34AM -0700, Joe Barnhart wrote:
> > While it is possible, in theory, to write a re-encoder that takes
> > advantage of compression hints learned from the original encoding,
> > that sounds like a lot of work for a rare task just to save some
> > CPU, and CPU is cheap in such instances anyway.
> 
> Thank you for the thoughtful reply, Brad.  The CPU power in this
> case is not insubstantial.  It took over 7 hours of processing to
> transcode the two hour movie down to a 720x480 DVD.  That was on a
> 3000MHz AMD-64 Windoze machine running Nero Video Express.  If I had
> an adequate solution on my mythbox, i.e. one that could be used
> effectively with only console I/O, I could have used my P4 HT
> processor for a bit quicker transcoding.  But I doubt the difference
> would have even been 2:1, so it would still be horrifically slow.
> (The console I/O requirement comes from the fact that the mythbox
> uses the HDTV as its primary monitor so I have to use ssh to do
> anything outside of Myth.)

On my AMD64 3500+ 90nm (2.2GHz) CPU (combined frontend/backend
system), mythtranscode transcodes my HD recordings to MPEG-4 (a
variety of resolutions from 720x480 up to 960x540) at approximately
real-time (e.g., 1 hour to transcode a 1-hour recording), in the
background, while other stuff is competing for CPU/disk resources
(watching other HD recordings, recording other HD broadcasts).

I've never transcoded to RTjpeg/MPEG-2, so I don't know if it is
"harder" or "easier" to transcode to than to MPEG-4.

This is just a data point for what Linux/mythtranscode can do with a
CPU similar to yours.

--Rob

Attachment: signature.asc
Description: Digital signature

_______________________________________________
mythtv-users mailing list
[email protected]
http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users

Reply via email to