Re: [oc] new project...MP3 (audio) core?

[Rudolf Usselmann <rudi@asics.ws>]

on 7/19/01 0:44, lior.shtram@flextronicssemi.com at
lior.shtram@flextronicssemi.com wrote:

> Hi All,
> 
> It is true about the patents that only encoders are charged with royalties and
> decoders you can make for free. Thinking about this block for a long time, we
> got to think that a hardware approach should not give such a good benefit in
> any respect (size, power, speed??, flexibility) but if you think otherwise I
> would like to hear it. Also it would be nice to support the other standards
> such as Ogg (which is the free one).
> 
> I also wanted to add that there is a nice academic work that was available
> over the net with VHDL code on implementing a decoder for MP3 half CPU half
> DSP based - take a look - don't remember where it is.
> 
> Lior

I think what we really should do is a data flow engine that
is micro-programmable and can do MP3 and OGG (?) and perhaps
other specific audio encoding algorithms.

A data flow engine is a part where you examine the operations
you will have to perform and concentrate on implementing them
as efficient as possible. The control logic for the data path
(typically a hard coded state machine) is replaced with a micro-
programmable state machine. Another way of looking at is to think
of a heavily stripped down DSP ...

Implementing a data flow engine could reduce the power
enormously and should be much smaller than truly general
purpose hardware such as a CPU. And of course it should
be able to run much faster as well.

Cheers !
-- 
rudi


--
To unsubscribe from cores mailing list please visit http://www.opencores.org/mailinglists.shtml