ward@cfa.harvard.EDU (Steve Ward) (11/24/88)
Here is my sideline appraisal of what is happening in 32K net land
with respect to hardware projects:
1. There is one project out there somewhere, coordinated by Richard
Rodman. Dave Rand can give pointers to this project. It is an '016
(please correct me if I err) cpu board which plugs into PC I/O bus as
a PC I/O card.
2. Several people have designed and wirewrapped 32k boards, but none
have been produced as printed circuit boards, except that the item 1.
project may be producing or about to produce PCB's. Note that I am
excluding commercial activities and only talking about more or less
volunteer PD efforts.
3. There seems to be a split into more or less two camps of 32k
'netters with respect to building and buying a 32k board:
A. the "small and inexpensive" camp. This camp wants to be
able to take adantage of PC hardware (I/O, periphs, chassis)
because they either have some or can buy it cheap. For this
camp, spending $1k on the 32k hardware alone (not counting PC
stuff) represents a hard ceiling; preferably < $500.00 !!
B. the "big and costly" or even "medium and moderate cost" camp.
This camp wants lots of performance and is willing to pay for
it. It is unclear what the price range is, but $2k-$4k seems
to be the range, with a few talking even higher prices.
Camp A will never convince camp B to switch camps, and vica versa. It
seems to be more an issue of money than anything else, and therefore
convincing probably means financial grants. I will gladly build a
big system if someone out there will bankroll me :-)
I personally am going ahead with a *small camp* 32k board project. I
encourage others to go ahead with their project, whatever it is. I am
of the opinion that forward motion will be slow unless somebody just
decides to do something and does it. I don't mean that every project
needs a solitary person, but hopefully each project will have somebody
that is going to design and build something, heck or high water, and
solicits help and input along the way. This is what I am doing. My
project is of the small camp variety, but I welcome all comments and
help.
I think the "big camp" needs somebody to get out in front, and
hopefully someone who will commit to the project all the way through
to the artwork (gerber plot file for me :-). I guess the make or
break point is when you find out if you have enough people and money
to make some boards, but you really are not at that point until you
have the artwork in hand, ready for PCB fab. If you can't get that
far, you can only talk in hypothetical terms about who may be willing
to pay for a board. Sure, you have to discuss what is most likely
to be acceptable to most people, but then you have to go do it. A
"big camp" point man is needed.
BTW, the main reason I am going ahead with a small 32k board project
is because I want one, no matter what happens elsewhere, and the only
way to guarantee getting a board at least moderately conforming to what I
want is to build it myself. I have no illusions about building the
board the 32k 'netters will desire - interest on the part of others
is nice, but secondary. Still, I am trying to work with others on this
just because more progress can be made and faster, with additional
help. Clearly my 32k board will only appeal, at most, to the "small
camp."
There is a clear common ground for these two camps: 32K SOFTWARE!
In response to earlier postings, several people have volunteered
to maintain 32K software archives at their sites. Others have
volunteered software. Shortly there will be a posting announcing
the details, including how/where to retrieve the software and to
contribute it. Hopefully software is a true common ground and
everyone will support the sharing concept!
Regards to all Steve W. cfa@harvard.edu ...harvard!cfa!wardpcg@aber-cs.UUCP (Piercarlo Grandi) (11/26/88)
In article <1285@cfa.cfa.harvard.EDU> ward@cfa.harvard.EDU (Steve Ward) writes:
3. There seems to be a split into more or less two camps of 32k
'netters with respect to building and buying a 32k board:
A. the "small and inexpensive" camp. This camp wants to be
able to take adantage of PC hardware (I/O, periphs, chassis)
because they either have some or can buy it cheap. For this
camp, spending $1k on the 32k hardware alone (not counting PC
stuff) represents a hard ceiling; preferably < $500.00 !!
B. the "big and costly" or even "medium and moderate cost" camp.
This camp wants lots of performance and is willing to pay for
it. It is unclear what the price range is, but $2k-$4k seems
to be the range, with a few talking even higher prices.
Well, I think I could belong to camp B, but I want camp A AT compatibility.
That is why I tend not to agree with the 32032 AND the 32532 proponents;
On one hand the 32032 is too cheap (and does not have crucial bus sizing
feature), and on the other hand a 10-15 MIPS 32532 means VERY HIGH prices if
you really want to exploit it -- then one buys Heurikon. A 386 level
32332 is in my view the best compromise.
I think that the discussion is really between people that want a system
where they can run something like UNIX, i.e. a 386-like box, and
people that wants a small sort of thing, mainly to experiment with a 32k,
i.e. an Atari-like box, maybe running MINIX.
Among the former, system people, the desire to be able to put things
into an AT clone is understandable; AT components are cheap, when you
need a lot of things to build a system. Otherwise we could just buy
a SUN...
On the other hand the gizmo people are not concerned; after all the Atari
is just a motherboard and a plastic box. It is more cost effective to not
care about compatibility etc...
There may be a third camp, those that suggest a 32532 system with NuBus.
Well, Why not a VME based Heurikon ? Why not a NeXT ? They come quite cheap...
Enough for this. We may eventually want to count ourselves...
As to software, Ron Guilmette has got the whole GNU suite working for
the 32k series; in practice one only needs a kernel at that point. Anybody
has a stopgap one until I get NONE done ? :-) :-)
Somebody had asked about the Ciarcia's 286 motherboard-on-a-board; it is
described, with full schematics, in Byte Septemeber & October 1987. Note
that it is not a coprocessor board, like the Definicon or Opus or Zaiaz.
It is a full motherboard, that will just need an AT bus backplane to run.
Also, why nobody gives any serious thought to replacing the 386 with a
daughterboarded 32332? I think that could be done... I think that
replacing a 286 with a daughterboarded 32332 in 386SX like capacity could
be done, and probably much more interesting.
--
Piercarlo "Peter" Grandi INET: pcg@cs.aber.ac.uk
Sw.Eng. Group, Dept. of Computer Science UUCP: ...!mcvax!ukc!aber-cs!pcg
UCW, Penglais, Aberystwyth, WALES SY23 3BZ (UK)