[mythtv] Re: mythtv-dev digest, Vol 1 #287 - 28 msgs

Dr. J. S. Pezaris mythtv-dev@snowman.net
Fri Jan 10 15:15:01 EST 2003

    From: Bruce Markey <bjm@lvcm.com>
    To: mythtv-dev@snowman.net
    Subject: Re: [mythtv] WinTV PVR-250 (hardware mpeg encoder)
    I'm told that there are up to 933MHz CPUs that don't need
    fans. This should be fast enough to encode mpeg4 and
    playback at 640x480 but is certainly fast enough for the
    MythTV default of 480x480.
I've tried running MythTV on one of the EPIA 800 MHz C3 CPUs.  These chips
have the effective equivalent of a 500 MHz P-III in terms of integer
computational power, but have quite awful floating-point performance.  They
certainly are NOT fast enough for the MythTV default of 480x480 with no
dropped frames.  At least that's my experience; maybe things have changed
since the last time I tried.

    I personally don't believe the CPU fan is a make or break
    issue. I have a system with a 1.2GHz AMD (currently $31)
    and a WinTV model 401 ($29) that can record up to 720x480
    mpeg4 or two tuners at 352x480 with no jitter. I'd rather
    have this today than pay $140 for a card that has no open
    source Linux drivers.

I am not advocating supporting the WinTV board in particular, but any board
with harware encode/decode.  I understand your price/performance point, but
I was asserting that this is not where the largest group of users would be.
    TiVo and ReplayTV already exist as stand alone boxes. A
    MythTV machine could be built as quiet as these. 

This is exactly my assertion: this is currently not possible because of the
lack of support for hardware encode/decode, and, therefore the requirements
of having a high-speed CPU which (a) requires a larger, louder power supply
(which may now require a fan where previously it did not), and (b) requires
a cooling fan.  No matter how hard you try, a box with no fan at all is
going to be quieter than one with even the quietest fans on the market.

    However, the network capability is where myth will outshine the
    commercial products. 

Agreed.  And the integration of other capabilities (eg, MythMusic, MythWeb,
automatic recompression, automatic spooling to writable CD or DVD, etc.),
running under Linux, and the all-important open source are reasons I
continue to follow Myth's development despite the current status not being
sufficiently advanced for my purposes.

    With future versions, I plan to have a myth box on my bedroom TV, home
    entertainment center, and two in the computer room. All four tuners
    coordinated by one scheduler and any recording can be watched at any

I contend, therefore, you are one of the higher-powered users which means
MythTV is going to work very well for you.  It's been nicely designed to
support just this kind of distributed model, *but* I assert that a much
larger group of users would prefer a single-box implementation.

    Have you tried MythTV or are you just assuming a CPU fan is critical?
    It can run on a system without a fan and having a CPU fan is not a
    'critical' issue.
Please give the details of your fanless system.  I was not able to get
MythTV to work beyond low-resolution proof-of-concept level on an Via EPIA
800 MHz C3 system.  Given the descriptions of systems and loads that Isaac
and other people have posted, it still seems that using a cpu which
requires a fan (THAT is the key, not the fan itself, but that one needs a
power-hungry and high-heat-dissipation CPU ... this leads to requiring a
fan, a larger power supply which now requirs a fan, etc.) is a prerequisite
for a reasonable MythTV system.

That said, Isaac has responded to the query for hardware encode/decode
support with two arguments, first that there is no current support for such
under v4l, and second that he won't support hardware that he doesn't own.
I, personally, cannot verify the first assertion, and hope that someone
else can.  I understand the second, but it seems ingenuine since there is
good MythTV support for things like non-US cable systems.

More information about the mythtv-dev mailing list