[hpsdr] Mercury

[John B. Stephensen]

External RAM would be useful for iterative (turbo) decoding. Perhaps one or
two large static RAMs could be put on the Mercury board.

73,

John
KD6OZH

----- Original Message ----- 
From: "KD5NWA" <kd5nwa at cox.net>
To: <hpsdr at hpsdr.org>
Sent: Sunday, June 11, 2006 05:10 UTC
Subject: Re: [hpsdr] Mercury


> ***** High Performance Software Defined Radio Discussion List *****
>
> I ended buying the Agilent Spartan III kit for $99 but paid an extra
> $50 to kick it up from the X3S200 to a X3S1000 so I could play with
> CPU architecture and DSP filtering.
>
> It's a pretty nice board besides all the built in RAM it has 512KB of
> 32 bit external 10ns RAM. I was surprised that it came with the
> software to generate DSP filters but it requires a copy of MatLab R13
> or R14 which I can't afford at this time.
>
> With the clock doubling feature you can clock the part to 250MHz
> external clock doubled internally to 500MHz the Altera is nowhere
> there in speed. A great part for doing CPU and DSP work, then there
> is Virtex 5 parts now you got some incredible crunching power and $$$.
>
>
> At 11:08 PM 6/10/2006, you wrote:
> >That's also an issue. I need to clock the filter engines at 147.456 MHz.
> >This required pipelined 18x25-bit multipliers plus 3-stage pipelined
> >accumulators on the Xilinx XC3S400. The high precision coefficients
support
> >large amounts of decimation in each filter. A slower clock would require
> >processing I and Q samples in parallel, rather than serially as I do now.
> >The number of multipliers then doubles.
> >
> >73,
> >
> >John
> >KD6OZH
> >
> >----- Original Message -----
> >From: "KD5NWA" <kd5nwa at cox.net>
> >To: "John B. Stephensen" <kd6ozh at comcast.net>
> >Sent: Sunday, June 11, 2006 03:54 UTC
> >Subject: Re: [hpsdr] Mercury
> >
> >
> > > They are also faster parts.
> > >
> > > At 10:43 PM 6/10/2006, you wrote:
> > > >***** High Performance Software Defined Radio Discussion List *****
> > > >
> > > >The RAM is a problem. I need lots of it for filters, the FFT and
> >traceback
> > > >memory. The extra RAM in the smaller Xilinx parts is a big help for
DSP.
> > > >
> > > >The largest Xilinx part available in a non-BGA package now appears to
be
> >the
> > > >XC3S500E and is $26 (1-24 pieces) at Nu Horizons. This provides more
than
> > > >twice the memory of the Altera, part plus a few more multipliers and
> >logic
> > > >elements.
> > > >
> > > >73,
> > > >
> > > >John
> > > >KD6OZH
> > > >
> > > >----- Original Message -----
> > > >From: "Lyle Johnson" <kk7p at wavecable.com>
> > > >To: "John B. Stephensen" <kd6ozh at comcast.net>
> > > >Cc: "Philip Covington" <p.covington at gmail.com>; <hpsdr at hpsdr.org>
> > > >Sent: Sunday, June 11, 2006 03:24 UTC
> > > >Subject: Re: [hpsdr] Mercury
> > > >
> > > >
> > > > > > It would be nice to have at least as many resources as a Xilinx
> >XC3S400
> > > >(16
> > > > > > 18k dual-port RAMs, 16 18x18 multipliers and 7200 logic
elements).
> > > > >
> > > > > The Altera EP2C8 is the largest Cyclone II device I can find in a
> > > > > non-BGA package.  It has 165,888 bits of SRAM (36 blocks of 4K
bits),
> >18
> > > > > 18x18 multipliers and 8256 logic elements.  $24 in singles from
the
> > > > > Altera online store.
> > > > >
> > > > > 73,
> > > > >
> > > > > Lyle KK7P
> > > > >
> > > >
> > > >_______________________________________________
> > > >HPSDR Discussion List
> > > >To post msg: hpsdr at hpsdr.org
> > > >Subscription help:
http://lists.hpsdr.org/listinfo.cgi/hpsdr-hpsdr.org
> > > >HPSDR web page: http://hpsdr.org
> > > >Archives: http://lists.hpsdr.org/pipermail/hpsdr-hpsdr.org/
> > >
> > >
> > > Cecil Bayona
> > > KD5NWA
> > > www.qrpradio.com
> > >
> > > "Windows the worlds most successful software virus"
> > >
>
>
> Cecil Bayona
> KD5NWA
> www.qrpradio.com
>
> "Windows the worlds most successful software virus"
>
> _______________________________________________
> HPSDR Discussion List
> To post msg: hpsdr at hpsdr.org
> Subscription help: http://lists.hpsdr.org/listinfo.cgi/hpsdr-hpsdr.org
> HPSDR web page: http://hpsdr.org
> Archives: http://lists.hpsdr.org/pipermail/hpsdr-hpsdr.org/


 1150008245.0