[mythtv-users] HDTV/CPU survey

[Jason Schloer]

As for the OSD problems. One of the ideas I've been toying with, but haven't
had the time to start messing with is using openGL for display. I found some
source for mpeg2play which uses XvMC's XvMCCopySurfaceToGLXPbuffer to copy
the frames to an openGL surface. I think if we can get this going we should
be able to have all the overlays done as openGL surfaces. So there shouldn't
be much of a slowdown with the OSD I would think. The sample I have seems to
work fairly well with standard mpeg PS streams, but I haven't been able to
get it to play an HDTV stream yet. It plays at full frame rate and allows me
to move the image in 3D with no stuttering. Also I made the texture size
1440x1080 and it had no problem with that, I just need to pump in a source
that's that big to see how well it handles that. Does anyone else think this
might be worth looking into further, or am I wasting my time even thinking
about it? 

-Jason
-----Original Message-----
From: mythtv-users-bounces at mythtv.org
[mailto:mythtv-users-bounces at mythtv.org] On Behalf Of Daniel Thor
Kristjansson
Sent: Thursday, April 29, 2004 12:07 AM
To: Discussion about mythtv
Subject: Re: [mythtv-users] HDTV/CPU survey

]If I read what you are saying correctly...  You do not get any jitter with
]720p or 1080i, even on severe camera pans?  But you do with OSD?

Yep, OSD is still a big problem.


]Do you think your FX card works better than the 4 MX?

I think so, with OSD at least. But I haven't been using this MX long
enough to be very sure.


]Have you tried turning on "use video as timebase"?  Do you get full-speed
]playback?

Well so far I haven't done any work on improving either of the
experimental playback modes, if either of them has improved it's
an incidental improvement.


]Hmmm, I will re-try your playback patch and make sure that -O3 is used.  I
]believe I have already tried that, but I will double check.  Last time I
]tried it, video playback was Okay, but did not seem improved over having
]XvMC disabled.  It also caused studdering with any OSD.

Make sure this applies to the libavdecode library, it sets it's own
flags based on whether the release or debug flag is set. It doesn't use
the general myth optimization settings.


]The extra audio buffering is not typically needed, but without it I
]*occationally* get an audio glitch.  I am currently using the onboard
]C-Media CMI9739A audio controller with the intel8x0 ALSA driver.  I have
]also used an old SB Live! which gave me equivalent performance.  I have all
]the audio going out the S/PDIF jack to my pre-amp.

I'm using an Ensoniq 1371, but I had problems with my built-in intel8x0
card. I think this is due to the immaturity of the MythTV audio support.
Unfortunately, audio isn't something I have any expertise in.


]I highly recommend that everyone using the HD-2000 card, use your
]hdtvrecorder patch.  Getting rid of the bad packets definitely improves the
]playback.

Thanks, I'm working on improving this by assembling PES packets and
using the redundant TS packets if the CRC doesn't check out.


]I appreciate the work you are doing on this project.  Sorry I was unable to
]help you with your WB audio problems.

That's ok, I'm laying some of the groundwork for fixing that by parsing
the PES and PSIP packets. For now I'm recording my WB shows from the
NTSC feed.


-- Daniel