[ a bunch of MIPS Co. gadgets ]

> if there are developers that would benefit from these, i'm willing to put
> together a few working systems and ship them. if there's someone in the
> san francisco bay area that would like to coordinate the effort and accept
> a "drop shipment" of hardware, that would be even better.

Oooh, I'd fly to SFO to receive some of them. :-)

> i have actually installed riscos on one of these -- it's best described as
> a "pure sysV port, complete with pure sysV bugs". i found it lacking, to
> say the least.

I'd had no change to touch with RiscOS 5 so far (I'm bit surprised it
existed ever).  I've been tracking UMIPS/RiscOS evolvement for years,
and the last RiscOS of my experience was 4.52.  It was an interesting
port of UNIX loaded with awefully burdensome efforts to make it
happen.  It was very regretful that RiscOS was a victim of
technological/commercial wigwags at the age like SMI/AT&T alliance.
RiscOS 4.52 was a certain combination of SVR3 + BSD networking stack
(native socket) + NFSSRC.

> i also have a number of mips magnum3000 systems (R3000) and some Rs2030
> (R2000A) if anyone's interested. i have framebuffers, they take generic
> memory, and the io chips are all "standard" eg amd7990 ethernet, emulex
> ncr5380 scsi, etc. i suspect that a port wouldn't take much effort.

Oooh, you have RS2030.  It was a UNIX of the folklore age of RISC
processors.  It was an evil machine (sorry) made as if it was designed
at a garage factory.  Every IO is routed via NEC V50 (forget the
specific number), and without the IOP internals/APIs it'd be
relatively hard to port NetBSD to it.  (Sumitomo Elec.'s R3000
SUMIstation is the direct descendent of RS2030 with less odd
hardwares).  On the other hand, R3000 Magnum was a good computer with
a smart IOASIC, named "Rambo," which can achieve near best performance
of the RISC age.

From the memories which were fading after near decade involvement.

Tohru Nishimura