[mythtv-users] Ceton patch available for trunk

[David Engel]

On Wed, Nov 30, 2011 at 12:46:17PM -0500, Ronald Frazier wrote:
> On Tue, Nov 29, 2011 at 5:40 PM, David Engel <david at istwok.net> wrote:
> >  I hadn't checked the code, but assumed the card
> > number was only to open the right device file.  No device file then no
> > need for a card number.
> 
> Correct. The device number is used to distinguish multiple card when
> opening the /dev/ceton/ctn91xx_mpeg<device>_<tuner> entry. When using
> RTP, you just need to use the IP for access. In the deviceid, I just
> use the string "RTP" in place of the device number to keep things
> consistent.
> 
> 
> > Ron, you've probably tested both device and
> > RTP access quite a bit.  Do you know of any reasons to prefer one over
> > the other?
> 
> I've done plenty of testing with both, with RTP probably being used
> more since my dev machine was using the Ceton in my production system
> (though I did physically steal the card from my production system for
> a couple of late night and overnight test runs). Clearly the RTP tuner
> is more capable. Using RTP did seem to use slightly more CPU than the
> file device, but now that I think of it, with the file device you
> probably have the ceton driver doing much of the same thing behind the
> scenes. This is one area that could perhaps use a little more testing.

I had a minor concern about CPU utilization, so I ran some simple
tests.  With a debug compile last night, RTP used 1% more of total CPU
(6% vs 5%).  That was with the CPUs underclocked from 2.5 to 1.2 GHz,
so it was really more like 0.5% of available CPU.  With a profile
build this morning, both access methods measured the same (2%).  I
think RTV vs. file access is a wash and not worth worrying about.

David

> One disadvantage of the file device I discovered is that it gets
> locked out when something else is using the RTP interface. Normally
> that's not such a big deal because as soon as the other client
> finishes, the device becomes available again. However, in my testing,
> where my production system is using 0.24 with the file device access,
> and my remote dev system is using RTP, I found that if my dev system
> didn't exit cleanly and shut down the RTP interface (by issuing the
> RTSP "TEARDOWN" command), the Ceton was just as happy to keep
> streaming forever. A few times I didn't notice this and thus my
> production system got locked out and missed a few recordings. If you
> use the RTP interface, you can always steal the RTP stream from the
> other client just by issuing a new SETUP command to the RTSP
> interface. For most users in most cases, this probably wouldn't be an
> issue, but it is another advantage of RTP.

-- 
David Engel
david at istwok.net