13 May
2004
13 May
'04
4:49 a.m.
On Thu, 13 May 2004, Jochen Kunz wrote:
Peter Wallace
Mesa Electronics
On Thu, 13 May 2004 07:25:20 -0700 (PDT)
"Peter C. Wallace" <pcw(a)mesanet.com> wrote:
I have experience in porting NetBSD to new hardware platforms. If it is
an already supported CPU the work is not that hard.
Then the Motorola 5200 might be about right... maybe overkill (400 or so MHz)
but its still pretty cheap
Give me
hardware and I'll care about the NetBSD port.
Thats a dangerous offer... I may not be that expensive nowadays. A MIPS or
ARM SOC with PCI for
I/O expansion [...]
I would prefer some PowerPC based stuff. I don't know
about the embedded
MIPS chips, but I know how fast a StrongARM or Xscale is and that a
(quite old) 300 MHz PPC604e is much faster. PowerPC is very well
supported in NetBSD and GCC[1]. It is continuously improved thanks to
the macppc users.
Maybe it is posible to use an already existing evaluation board and
interface it to the UniBus...
Right,
CPUs without PCI are even cheaper.
I would
not use non-PCI devices. If all I/O beside the UniBus hangs of
PCI, all you need is a machine dependent PCI backend in NetBSD to give
you acces to all PCI devices already supported by NetBSD. That way you
need only to implement a single driver for the PCI bus, not one for
every device on your board. Discrete mosfets with RC at the gates (say sot23
if surface mount)
would give you all the drive you needed without excessive edge rates.
The input side could be done with Schmitt trigger CMOS parts or even
quad comparators if you wanted to get fancy about input threshold...
Well, when
you say. Discrete analog stuff is not my territory. ;-)
[1] GCC likes to generate sub-optimal code for "non mainstream RISC"
CPUs and you have to be prepared for compiler bugs on "non mainstream
CPUs".
--
tschüß,
Jochen
Homepage: http://www.unixag-kl.fh-kl.de/~jkunz/