From: everettm@merlin.think.com (Mark Everett) Subject: Re: MIPS R3000 board to run Linux, anyone? Date: 16 Jun 1993 10:53:57
In article <1vn67m$ibm@senator-bedfellow.MIT.EDU> s9091124@aix00.csd.unsw.OZ.AU (Mark Pulmano) writes:
[thread not included because I don't know how (I'm new to this)]
If you're using GNUS in emacs, after 'f' for follow-up you can ctl-C ctl-Y where
"ctl" is holding the control key down. This assumes you haven't changed your
keymap. (You probably haven't, yet :-).
I've thought about other processors stuck into my pc as well, and I think that
we shouldn't restrict ourselves to a R3000. We should consider chip pricing
and availability as well. There was a message a while back (in comp.sys.???)
which said Fujitsu Sparclite's (don't know how different it is to a real Sparc)
could be had for <US$20, and i960's & Am29000's for <US$50 in another message.
I really know nothing about i960s. the Am29200 (I think) has memory management
and floating point on chip. It would be more expensive ($200 as a guess) but
not bad for the extra stuff. Its an interresting architecture: 128 registers
shadowing the stack and another 64 global registers. Plus, you can "protect"
registers in sets of 16 (also good for not forcing a reload of all 128 local
registers on task switch). The interrupt latency is one cycle, I believe,
in one of the interrupt code formats.
The above information could be wildly wrong (an unreliable memory) and I
apologise for it's generality and vagueness, but the point is we should look
at the pro's and con's of other chips eg. apparently the i860 has similar
memory management to a 486 (I think...).
The i860 is essentially a dead architecture. I have a Stardent station, made by
a division of Okidata, which went out of business rather than porting SVR4 to
yet another architecture when Intel decided not to do other spins. Besides,
the chip handles interrupts by throwing up and letting the software clean up
for it. Not a pretty sight.
I think the idea of another processor board for a pc is a great idea, though
I would like to extend it to a multi-processor board (R4000's are better
than R3000's for this because they have atomic test-and-set). I would very
much like to be involved in a project of this sort.
ps. where can you get chip pricing and specs from?
(preferably internet sources, because I live in Australia)
Australia has a big enough economy that most of these companies have sales
offices there :-).
-everettm