Re: What ever happened with hardware decoding?

From: RCMe (rcme_1NOSPAM_at_nospam.hotmail.com)
Date: 02/23/04 

Date: Sun, 22 Feb 2004 18:58:31 -0600

Hi Jeff,

Thanks for your reply. Yes, I have the available h/w assist enabled on the
MX440 card (with NVDVD).

Are there any video cards that support full h/w assist decoding?

-- rcme

"Jeff Griffin" <jeff@mvps.org> wrote in message
news:%23i$jijZ%23DHA.3880@TK2MSFTNGP11.phx.gbl...
> Actually if your video card supports it and you have it enabled in your
> MPEG2 decoder part of the decoding will be offloaded to the video card...
> Not all of it, but some.
>
> Jeff Griffin
> Windows XP Media Center Edition MVP
> www.nacontap.com
> www.thegreenbutton.com
>
>
> "RCMe" <rcme_1NOSPAM@nospam.hotmail.com> wrote in message
> news:OYf1G6Y%23DHA.2636@TK2MSFTNGP09.phx.gbl...
> > I am curious if anyone knows why there is no hardware decoding with MCE.
> >
> > I know there is hardware encoding, but why don't any of the current
> hardware
> > cards support hardware decoding?
> >
> > Is it that hardware decoding is "too hard" and the hardware vendors
would
> > have to charge too much for the hardware? This would be surprising given
> > that hardware encoders are so inexpensive.
> >
> > Or, is it that in trying to keep the playback component too "generic",
> they
> > are trying to do too much in software, which is pushing MCE system CPU
> > requirements to the "unrealistic" level.
> >
> > Or, is this just more marketing "angle" to force consumers into
needlessly
> > buying high-end computers.
> >
> > After reviewing many of the latest CPU requirements for MCE systems and
> the
> > requirement for P4 3.0Ghz HT level CPUs seems quite surprising. Why is
so
> > much "horsepower" needed on these computers? It seems that all these CPU
> > cycles are going solely to support playback decoding. With hardware
assist
> > for playback decoding, I would expect that the system CPU would not be
so
> > taxed, resulting in "medium level" computer systems being able to
support
> > MCE.
> >
> > I can't imagine that the MCE Extenders are going to have P4 3.0 Ghz HT
> CPUs
> > in them just to support playback decoding. I would expect that the
> Extenders
> > would have to have some form of hardware decoding assist to keep the
> overall
> > costs down.
> >
> > To experiment with this a bit, I setup a couple of MCE systems and here
> are
> > the results.
> >
> > MCE 2004
> > Celeron 2.0 Ghz 512MB RAM
> > Hauppauge PVR 250
> > Nvidia MX440 64MB (NVDVD decoding software)
> > CPU usage Encode only: <5%
> > CPU usage Decode/Encode: ~35%
> > Notes: All aspects of playback work great and I would consider this a
very
> > usable/stable system.
> >
> > MCE 2004
> > Pentium 3 1.0 Ghz 512MB RAM
> > Hauppauge PVR 250
> > Nvidia MX440 64MB (NVDVD decoding software)
> > CPU usage Encode only: <5%
> > CPU usage Decode/Encode: ~70%
> > Notes: Playback was fine but FF/RW were a bit "sluggish" so I would
> consider
> > this a "borderline" usable system.
> >
> > Conclusion: Since the encoding is done in hardware, there was very
little
> > difference between the two systems when "at idle" doing just encoding.
> > However, since the decoding is done in software, there was a big
> difference
> > between the two systems during playback. Both systems functioned ok at
> > "steady state" playback. However, while the Celeron system worked fine
> > during all forms of "dynamic" playback (i.e. FF/RW, etc.), the P3 system
> > typically would "peg" the CPU at 100% causing sluggish performance
during
> > these operations.
> >
> > In addition, I looked at these systems from a "media server"
perspective.
> > While the P3 system is borderline usable as a standalone MCE system, it
> > would be unusable as a file/media server.
> > However, the only problem I noticed with the Celeron system was some
> random
> > "stuttering" during playback when background file transfers occurred.
That
> > is, pulling large files (i.e. dvr-ms) off the Celeron system to another
> > computer using Windows Explorer, during playback (i.e. watching TV).
> > However, the Celeron system worked fine (no stuttering or noticeable
> delays)
> > as a file/media server, where 2 WinXP computers were connected to it
> reading
> > dvr-ms files during playback/viewing in WMP9 on each of the WinXP
> computers.
> > This might be expected as the Windows Explorer file transfer used much
> more
> > CPU on the Celeron system (copying at a max rate) opposed to the dvr-ms
> > "streaming" file transfer to the 2 WinXP WMP9 desktops (which
transferred
> > playback data at a much lower rate).
> >
> > - rcme
> >
> >
>
>