Bruce ONeel <edoneel at sdf.lonestar.org> wrote:
> This looks cute...
> 
>  http://www.strategic-test.com/embedded_systems/pxa255-cpu_ethernet_dimm_module.html
> 
>    Processor Intel XScale PXA255 400 MHz
>    Memory 64 MByte SDRAM (32-bit @ 100 MHz)
>    32 MByte flash memory (32 bit)
 ...
> all on a DIMM sized module.

Thanks for the link.  I took a look at the site.   At 70 mW standby,
1300 mW active, it is way outside my power budget.  I couldn't find any
prices on the site (which means it is also way outside my financial
budget ;) ).

I'm looking for a much smaller, cheaper, lower power uP, but hopefully a
32-bit CPU, with several serial ports and a flat address space.  I am
sick of segmented architectures.  Ignoring the off-chip peripherals it
would be driving, I think the uP needs to use in the area of 3 mW active
and well under a mW on standby.  It needs access to perhaps several MB
of static RAM (not necessarily on-chip).  It will need to talk to a
PCMCIA (or compact flash) flash disk. I like the idea of the main
program running in RAM, so that on boot-up, the program is loaded from
the card disk into RAM.  The program size right now is estimated as
being in the area of 64KB.  (The Texas Instruments MSP430 family sounds
appealing, although I haven't looked closely yet.  60KB of on-board
flash for the program code in a $10 chip (in singles from Digi-Key). 
I'd still like to be able to run code in RAM, and the MSP430 is 16-bits
but I would prefer 32.

I have no hopes of it running Squeak directly.  I expect to use (my own)
Forth, although I have been considering more and more of using Squeak as
a fancy "macro assembler" (as Jon has done for the PIC 

http://www.huv.com/uSeeker/smalltalk/pic.html

or as mentioned on this list some time ago by Alan or Dan).

Also, if I can rant about Hitachi and its obsolete H8/532 uP a moment: I
was (and currently still am) stuck with this chip on the project; I
didn't choose the chip.  I don't know if we will be able to change but I
am working toward that possibility.  Anyway, aside from the limited and
self-inconsistent documentation, one of the most irritating things about
it is that the Hitachi engineers seem to think it is OK for the on-chip
serial port to *send out a character everytime the chip comes out of
standby mode*.  I am aghast.  Not to mention that its 1 MB address space
is not flat but is accessed by 64KB segments.  Did I mention how I felt
about segmented architectures?  ;)


-- Frank