[hpsdr] [Flexradio] PlayStation 3

[Philip Covington]

On 12/9/06, Jim Lux <James.P.Lux at jpl.nasa.gov> wrote:
> At 05:43 PM 12/8/2006, Philip Covington wrote:
> >O
> >>Somehow, I suspect that an SDR1000 and PowerSDR isn't going to cut it
> >>for that application. And, for that particular application, there are
> >>a host of off-the-shelf ASICs that do the decoding, although there is
> >>some intellectual satisfaction from rolling your own (or, maybe you
> >>want a modified version that, say, doesn't interpret certain
> >>flags).  So, you're already in a high data rate, high bandwidth
> >>platform, and perhaps a software implementation isn't the best system
> >>solution (i.e. why not put it into an FPGA?)
> >
> >You must not follow the HPSDR group... Mercury has a FPGA...
>
>
>
> Following it casually. I figured that was the case, so then, what
> does the PS3 really buy you?  You're not going to be doing a HDTV
> decoder in software, presumably, nor a COFDM or ATSC demodulator.  I
> grant I can think of possible applications for lots of CPU cycles,
> but I have a hard time thinking of a reasonable SDR application (of
> interest to more than, say, a dozen people) where something like a
> PS3 would actually be the "best" solution.
>
> The SDR1000 has sold, probably, more than 1000 copies, so it's
> clearly meeting a fairly wide need. Partly this is because it
> leverages very inexpensive computing power available in the form of
> consumer PCs.  Potentially the PS3, being a mass market consumer
> device, can provide a very attrative MIPS per dollar/euro ratio, but
> I'm not sure it could be effectively utilized.. there not being
> sufficient need for such power, in a amateur SDR, in moderate
> volumes. Some day, say 5-10 years hence, there will be some SDR
> application with similar popularity and interest, but by then, the
> PS3 will be history, and the set-top-box will have moved on to
> something else, and it's not clear that development for the PS3 would
> port forward painlessly (if you even wanted it to.. XP and Agile and all that).
>
> I can see the PS3 being a (very) interesting platform to tinker with,
> for a very limited number of people who are willing to spend the time
> and effort to port whatever is needed, etc.  But those people would
> be (and are) tinkering, regardless of whether the PS3 existed.  The
> hard core (or occupationally supported) tinkerer who really wants to
> fool with cell processors might even fork out the substantial cash to
> get a different eval platform.
>
>
> The high performance computing community has gone around this sort of
> thing many, many times.  Every time some new whiz bang specialized
> consumer computational widget comes out, there's folks who try to
> leverage it for HPC.  Pretty soon, though, they fall back to the off
> the shelf commodity PC style computers (albeit packaged differently
> these days, to get the density up), because they are truly more
> flexible, easier to support (both in terms of failures, and in terms
> of available software), and because it's cheap to add more PCs to the cluster.
>
> That said, I am hopeful that finally, there will be some high
> performance compute engine that isn't a pain in the rear to work
> with.  I'm just a bit jaded and cynical. {Hey, I had a Intel PSC-1
> SugarCube, and I had some transputer cards, and a 68K with the
> 68888(?) FPU, as well as Weitek 1167 and x87 FPCs, so I've been
> through the special purpose computer thing before...)
>
> Jim

I don't think anyone is really thinking about hooking their SDR-1000
up to a PS3 and I agree there would not be much to be gained by doing
that.  When the PS3 becomes more widely available it will be a way for
those interested experimenters (and even affordable for hobbyists) to
be able to hack around with a Cell processor.  Right now the other
Cell options are way too expensive for experimenter/hobbyists.  I
don't really give a hoot about the practicalities, otherwise I'd go
buy a YaeKenCom Ham Radio Appliance and just operate - I have fun
hacking around with things like this.

Phil N8VB

 1165673356.0