[mythtv] A need for advanced player settings

Raymond Wagner raymond at wagnerrp.com
Fri May 27 18:25:47 UTC 2011


On 5/27/2011 12:43, Lennart Sorensen wrote:
> On Fri, May 27, 2011 at 09:33:54AM -0700, Jim Stichnoth wrote:
>> My observation, though not backed up by any rigorous performance
>> testing, is that the Atom is roughly equivalent to a Pentium 4 in
>> terms of performance on video display (decode/scale/deinterlace) and
>> compiling the myth source code.  I don't think the computational
>> requirements of mythbackend have changed much since the days when the
>> Pentium 4 was king, so it doesn't seem unreasonable to use an Atom for
>> the backend.  (Note that I'm well aware of anti-Atom positions like
>> http://www.gossamer-threads.com/lists/mythtv/users/448321#448321 .)
> The P4 was simply a bad design.  The atom isn't a bad design, but one
> has to be aware of the tradeoffs made in the design in some cases.

The Atom is a terrible design.  They threw away the branch predictor, 
out of order execution, and all the bits that make superscaler 
architectures actually work well, in the name of cutting transistor 
count.  Do note that it was _not_ done in the name of power efficiency, 
since they cut out all the processor gating and clock scaling (in the 
desktop versions) that would otherwise make it draw less power when 
idle.  It was done to be cheap, plain and simple.


More information about the mythtv-dev mailing list