[mythtv-users] Just how great can VDPAU be?

Scott D. Davilla davilla at 4pi.com
Sat Nov 29 16:45:03 UTC 2008


>Scott D. Davilla wrote:
>>>   >Has anyone been able to test on a P4 based system with a good old
>>>>  fashion PCI card?  My quick search of Newegg didn't find any AGP 8xxx
>>>>  or newer cards, but I did find PCI cards.
>>>>
>>>> 
>>>>http://www.newegg.com/Product/ProductList.aspx?Submit=ENE&N=2000380048%201305520548%201069609642%20106791921&name=GeForce%208%20series
>>>>
>>>>  Here's hoping that my MythTV system built from old parts may be
>>>>  able to do HD.
>>>>  -Mark
>  >> ooohhh, been looking for a PCI 8400 to test VDPAU on the Intel dual
>>>  core atom 330 ITX motherboard. Thanks. If this works, $80 for the
>>>  atom 330 mb, $60 for the video card, +case and ps == real cheap 1080p
>>>  capable media center.
>>
>>  Just an update to testing VDPAU under the atom 330 and a PCI nvidia
>>  8400. All I can say is WOW. H.264 video content that took 45 and 160
>>  percent CPU under the atom 330 without VDPAU takes less than 5
>>  percent cpu with VDPAU. Not a type, that's really less than 5
>>  percent. VDPAU == very cool.
>>
>
>That's very interesting - thanks for the update.
>
>What's the "watchability" like with that PCI Nvidia 8400?  Any
>stuttering?  What resolution and bitrate of H.264 content were you
>testing with?

testing under mplayer so I can tell what's doing what. With VDPAU, 
absolutly no stuttering, the GPU is handling everything :) 720p mid 
to high 4.1 profiles. Does not seem to matter, any 4.1 or below at 
any bit rate seems fine. 5.1 profiles will have to wait for nvidia to 
fix these.


>This could open up some options for people with older hardware that only
>has AGP for video, not to mention lower-power (wattage) hardware, like
>you seem to pursuing with the Intel Atom.  I could use a PCI 8xxx video
>card in place of an AGP card, and still get watchable performance on
>Hi-def H.264 content, in theory.

That would be atom 330 (dual-core) not the previous single-core atoms 
but from these tests a single core atom should also be fine.

>
>  From the notes on the patch that Isaac submitted, it looks like MPEG2
>video is still not-quite visuallyoptimal, but that's probably just a
>matter of time before that gets smoothed out.

Not sure under mythtv, like I've mentioned, this is under mplayer to 
reduce the dependencies. A would assume that if mplayer can do it, a 
myth frontend should also be able to do it.

Remember that VDPAU has only been out for a couple of weeks so as 
nvidia gets feedback and makes changes, I'm sure it will improve. And 
a side note, under the dual-core atom 330, all mpeg2 is easily 
handled in software decode alone. So even if there are issue with 
VDPAU and mpeg2, who cares :)






More information about the mythtv-users mailing list