>> 3.  Ditto the SGI/MIPS architectures.
>
>I haven't heard rumor here about SGI's, but we do run on lots of MIPS
>hardware already, so it would be a matter of technical information and 
>will on the part of an interested person.

We run on mips r2k/r3k, on r4k/r4400/r4600. The locore code is known
to work on the QED cores. We have both little- and big- endian ports
in the tree. As perry says, all that's really needed is technical info
and time on the part of a skilled/interested person.

The information you'd need includes details of how bootloaders work,
to build bootblocks; and of device chips, i/o bus chips, and the
addresses they're wired up at. Then it's a simple matter of
cookie-cutting the existing MI drivers.  (And if IRIX uses ELF or
ECOFF, you can use native-OS bootloaders at least to begin
with. That's what I did on pmax.)


I've heard rumours of _interest_ from the direction of NASA Ames, but
nothing more concrete.

(Historically, SGI also tended to be very tightlipped about their
graphics hardware, which'd be a big obstacle to getting graphics
working; you may well want to emulate IRIX so as to run the SGI
product X/GL servers, but that's just a guess).