[mythtv-users] Accuracy of HDTV processor requirements statement in HOWTO

[Michael T. Dean]

Steven Adeff wrote:

>On 2/7/06, Kevin Kuphal <kuphal at dls.net> wrote:
>  
>
>>Jerry Rubinow wrote:
>>    
>>
>>>I fooled around for months trying to get HD playback working on a
>>>(original) Pundit with 2.8GHz non-HT P4, conncted to a 720p tv.  I
>>>tried both with the onboard Sis graphics and with an FX5200 PCI card.
>>>But I never could get something satisfactory.
>>>
>>>Through the 5200, playback was always too slow/stuttery without XvMC -
>>>probably maxing out the PCI bus.  With XvMC, I had very low CPU usage
>>>but either it would be crashy or lock up occasionally or also be
>>>stuttery.  I could play the same file with mplayer w/ XvMC without a
>>>problem.
>>>
>>So you did get 720p content working on the FX200 PCI with XvMC?  I'm
>>fairly sure now with the existence of various PCI HDTV capture cards and
>>PCI HDTV decoders (MyHD, etc) that hardware accelerated HDTV playback on
>>a PCI device is certainly in the realm of existence.  I would imagein
>>though, that unaccelerated playback would be difficult.
>>    
>>
>doesn't the PCI bus share bandwidth though? From what remember, one of
>the advantages of AGP and PCIX were that they did not share bandwidth
>amongst cards, each had a dedicated pathway. I think also that PCI was
>only capable of handling 4 cards, and boards with 5 pci slots actually
>had two pci bus's, with the 5th pci slot being on a seperate bus from
>the first 4.
>
>this could also all be fud + bad memory...
>  
>
Not the part about a dedicated pathway for AGP, at least (don't know 
about max PCI slots, but I /thought/ that was 7--but included many 
on-board devices, which may explain why only 4 were left on the MB 
you're thinking of).

See http://en.wikipedia.org/wiki/Accelerated_Graphics_Port , 
specifically the section "AGP vs PCI" for some of the benefits of AGP.  
Most important to the curret discussion are the parts about the 
dedicated pathway and GART.

Mike