[mythtv] an architectural issue for future releases

Isaac Richards mythtv-dev@snowman.net
Sat, 12 Oct 2002 16:11:23 -0400

On Saturday 12 October 2002 03:14 pm, Dr. J. S. Pezaris wrote:
> Hi Issac and Other Readers,

My name is 'Isaac', btw.  It's really not _that_ hard to spell.

> Before raising any issues, I want to first be up-front and say that I *do*
> think Issac (with the assistance of the small handful of other contributors
> and developers) has been doing a bang-up job.
> I'd like to ask a high-level question about the basic architecture of the
> system, though.  One of the problems with software-based codecs is that you
> need heavy-duty processing power, at least in contemporary terms,
> especially to be running both ends simultaneously.  With the advent of
> low-power (both in Watts and in CPU cycles per second), small motherboards
> such as the Via Eden series which have nearly everything included *except*
> a video capture card, I wonder if it doesn't make sense to design support
> in MythTV for more powerful video capture cards which include hardware
> encoding and perhaps decoding.
> Does anyone have experience with these cards, and know how well they are
> supported under Linux?  Can anyone familiar with the internal architecture
> of MythTV (like Issac) suggest how difficult this might be?

There aren't any cheap hardware encoding cards that are supported in linux, 
aside from a few that do hardware mjpeg compression.  The API for using those 
cards is quite similar to how mythtv grabs frames currently, so it really 
wouldn't be difficult to add support for them -- just a different method to 
get the compressed frame, and then a flag to store the frame in the file 

There seems to be a partially complete driver for the WinTV PVR USB, which'll 
do mpeg2 compression, but it's a USB device, and as such is relatively 
uninteresting to me.

If any of the other mpeg2 encoder cards ever get linux support, I'll probably 
pick one up and see 'bout supporting it..