Path: utzoo!utgpu!attcan!uunet!ncrlnk!ncr-sd!hp-sdd!hplabs!nsc!grenley
From: grenley@nsc.nsc.com (George Grenley)
Newsgroups: comp.sys.nsc.32k
Subject: Re: 32K "do-it-yourself" kits...
Keywords: The Ultimate Scrooge...
Message-ID: <7594@nsc.nsc.com>
Date: 7 Nov 88 18:13:06 GMT
References: <2613@sultra.UUCP> <7532@nsc.nsc.com> <1039@stratus.UUCP>
Reply-To: grenley@nsc.nsc.com.UUCP (George Grenley)
Organization: National Semiconductor, Sunnyvale
Lines: 85

I'm happy to see Mr. Smith helping out here.  As a former NSC employee, he has
detailed knowledge of Life Here in 32000 Land.

In article <1039@stratus.UUCP> gbs@stratus.UUCP (George B. Smith) writes:
>In article <7532@nsc.nsc.com> grenley@nsc.nsc.com.UUCP (George Grenley) writes:
>>In article <2613@sultra.UUCP> dtynan@sultra.UUCP (Der Tynan) writes:
>>>OK, so I've been following all these postings about so-and-so who makes an
>>>"ultra-cheap" 32K experimental system.  Well, so far, 'ultra-cheap' has been
>>>open to varied interpretation.  
  (GRG talking:)
>>I still maintain that if cheap is your primary goal, you can't beat the 532DK,
>>at a whopping $532.  Yes, half a k-buck is not trivial, but you have
>>something which, with work, can be a real OS9/Minix/GNU/whatever system.
  (GBS talking:)
>I suggest that the 532DK, otherwise known as the 532 Designer Kit, would *not*
>be an appropriate starting point for a system.  There is no bus, there is only
>128k of static RAM with no provision to expand it, there is no provision for
>adding a mass storage device, i.e. disk of any kind, there is no provision
>to add networking, etc, etc.  The 532DK is just what George Grenley stated
>before, a demo board which proves that it is possible to wire a 532 up to
>other chips and run small benchmarks generated by expensive 32k software
>tools which must run on expensive host systems.  There is a "wire wrap area"
>on the board which George must be referring to when he says that "with work"
>the 532DK "can be a real OS9/Minix/GNU/whateve system".  I would say a *lot*
>of work that would be of questionable value.

I agree.  I said a lot of work would be required.  However, for a hobbyist on
a budget, it would not be too difficult to add a DRAM controller and a couple
of meg of DRAM; and also add a SCSI port.   Here at the ol' factory we have
a list of app notes we're working on; these subjects are to be covered, as
quick as we can get to them.  (By the way, while we cannot commit to anything,
we do sometimes work with outside engineers to write such app notes.)

This would make a pretty good single or small multi user system.  Not a
NeXT or anything, but the cost is reasonable.
 

>Now the CG16 group is working on something that would be a good starting
>point for a 32k PC system.  Hey Dave, yeah, Dave Rand, care to share some
>details with the net?

I do not believe this is a good alternative, for several reasons.  First, the
32CG16 has no MMU capabilities, so any Unix type box is out.  Second, while the
BitBlt capabilities of the CG16 make it quite fast at bit diddling, in terms
of general performance, it is more like a 15 mhz 68000 - not that fast compared
to other CPUs on the market.  Although I haven't benchmarked it, I would guess
that a fast 286-clone machine running Microport Unix (or similar) would be
at least as fast, maybe faster.

Why build a slow machine?  Most of the cost of any Unix-capable machine is 
going to be memory (32 1 meg RAMS = $500-$1000) and disk (40 meg = $500).
So, why not hook it up to a GREAT CPU?

>>Der, the realities of corporate bookkeeping are such that we can't do
>>ANYTHING in low volume for $100.

>National does not want to cater to the low cost , low volume market.
>Now my question is of the chicken or the egg variety, is it because
>they in fact *can't* do it or they just don't want to?  Let's put it
>another way, if they thought they *could* do it, would they find a way
>to get it done?

As my esteemed former collegue knows, wanting to do something and doing it
are different.  NSC, with the overhead of a billion dollar semi plant capacity,
simply cannot afford to do business "retail".  If some other company wanted
to do a PC class machine, we would certainly be interested in seeing them
be successful, but frankly I doubt that you can knock off either the DOS
camp or the Apple camp.  Sun might; but they've got a running start.  NeXT
might; they've got Steve Jobs.  But I don't believe the low end of the computer
market can support that many families - the three that exist now (DOS/Intel,
Mac/68020, Unix/68020-sparc) are enough, if not too much.

>>Regards,
>>George Grenley

>Regards,
>the other George

>George B. Smith                  Disclaimer: don't hold anyone else
>Stratus Computer, Inc.                       accountable but me.
>gbs@stratus

Best Regards,
the first George
Nat'l Semi