[mythtv-users] Mythtv playback problem with NVIDIA
videoa cceleration - HELP! (still unsolved)
ahagge at wbfa.com
Mon Apr 4 14:31:45 UTC 2005
Jarod Wilson wrote:
>[ stuff about XvMC in ATrpms packages not working w/71xx nvidia driver ]
>On Tuesday 29 March 2005 01:53, Jarod Wilson wrote:
>>>>>So does it work in mythtv now, or not?
>>>>NO it doesnt! :):) Same problem.
>>>>All the rendering stuff was working all along its the first thing I
>>>>checked when Mythtv didnt work!
>>>Any word on this oh atrpm gurus?
>>Was gonna try to see for myself this evening, but for whatever reason, I
>>can't even get X to actually work with 7167 on my main HD playback box at
>>the moment... Will try again tomorrow night with another box. If I get that
>>problem myself, I'll rebuild Myth against the 7167 libs and try again.
>Okay, finally got back to this. Just got 7174 playing nice on my Athlon XP
>1800. It definitely lowers cpu usage for HDTV playback, though still not
>enough that I can watch 1080i programming w/a deint filter enabled on this
>lowly box without stutter (plays well enough w/no deint filter, but lots of
>motion jaggies). Will try again when I swap my GF 6200 back in there,
>currently has a GF4MX...
Interesting - I have an Athlon XP 1700+ cranked up to 2000+, and for the
life of me I can't get HD playback without stutter, even with the 7174
driver (I'm using XvMC, too!). I _think_ it MIGHT be the onboard audio
(i810-like) that's the culpret, but I can't be sure. MythTV doesn't
seem to log much useful audio info, even with "-v all" turned on in the
>Ah yes, as for the XvMC thing, I'm getting no-go also:
>Using XvMC version: 1.0
>XvMC found and using IDCT surface
>2005-04-01 23:57:23.224 Using XV port 105
>Error: API mismatch: the NVIDIA kernel module is version 0.7174.0, but
>this library is version 1.0.6629. Please be sure that your kernel
>module and all NVIDIA driver files have the same driver version.
>Unable to create XvMC Context return status:11 BadAlloc
>2005-04-01 23:57:44.760 Changing from None to WatchingPreRecorded
>2005-04-01 23:57:44.760 Decoder not alive, and trying to play..
>2005-04-01 23:57:44.860 Changing from None to None
>That happens even after yanking all traces of 6629 from the system, so it
>would seem 6629's libXvMC is staticly linked to libmyth, or something like
>that. I'm kicking off a rebuild of myth against 7174 before hitting the sack,
>we'll see what we can see tomorrow...
Yes, MythTV needs to be recompiled. I don't _think_ it's a static vs.
dynamic issue, though. I think the nVidia headers must force this to
spit out if the version of the library isn't the same as the version of
the includes that were used to build the executable. I had to do
likewise with the Xine-HD app. that was supplied by pcHTV for use with
their HD-3000 card.
More information about the mythtv-users