Lpc2000

> Cost-wise the Xilinx Spartan 3E looks interesting, as it allows one 
to use
> generic (low cost) SPI memory for config of part and application post
> config.  Xilinx offers low cost eval boards, but beyond the eval 
software -
> $$$.
> 
> 
> Joel
>

I have used the Spartan 3 Starter board quite a bit as well as the 
B5X300 from www.burched.com.  My largest project was to implement the 
T80 core (Z80) in the B5X300, add a couple of CF drives and run CP/M 
2.2.  PS/2 keyboard input and 25x80 text mode for VGA output.  All of 
this works very well and I have a totally silent, if obsolete, 
computer.

I don't think I consider the Xilinx WebPack ISE as 'eval' software.  
This is a MAJOR development platform.  Sure, it doesn't have the wiz-
bang features of some of the more expensive systems but it does a 
wonderful with VHDL.  And, it makes using the devices possible for the 
average hobbyist.

I intend, at some point, to build an IBM 1130 into the Spartan 3 
Starter Board.  Not a particularly interesting computer, unless it was 
your very first, which it was.

But, there is no way in the world to implement something as 
sophisticated as a modern microcontroller with an FPGA and get it done 
anytime soon.  There are a lot of cores around but nothing as complete 
as an LPC.  And a truly custom core will have exactly nothing for a 
tool chain.  In some ways the device selection will be driven by the 
availability of inexpensive tools - particularly for hobbyists such as 
myself.  If development tools aren't free, I have to move to a 
different device.  My budget is exactly $0.

On the scale of things, there is no point in using a 48 pin ATmega128 
when a PIC 16F88 will do the job.  That '88 has a lot of gadgets for 
an 18 pin pkg.  And there is no point in using an LPC if the ATmega128 
can do the job.  First of all, the world is based on 5V and the 3.3V 
spec of the LPC gets in the way.  Particularly for A/D.  But, if you 
need the horsepower, the LPC is a great device.

And so it goes, up the scale to ARM9, X Scale, Pentium, and on and 
on.  At every increasing level, complexity increases, development 
takes longer, tools cost more - everything goes up.  Including the 
amount of information/knowledge required to make the thing work.  
Linux on the Intel PXA255 (www.gumstix.com) is an example.

There is no single, universal, one size fits all, processor.  If such 
a thing could be built, there wouldn't be hundreds, if not thousands, 
of processors around (counting each PIC and AVR flavor as a unique 
device).

I use the LPC because development boards are cheap and software is 
free.  If either condition changed, I would move to something else.  
And, even at that, when I get up in the morning I am as likely to play 
with a PIC as an LPC.  It depends on what the project demands.

Richard