Path: utzoo!censor!geac!torsqnt!news-server.csri.toronto.edu!cs.utexas.edu!sdd.hp.com!zaphod.mps.ohio-state.edu!unix.cis.pitt.edu!pitt!willett!ForthNet
From: ForthNet@willett.pgh.pa.us (ForthNet articles from GEnie)
Newsgroups: comp.lang.forth
Subject: Forth Engines / Harris
Message-ID: <2124.UUL1.3#5129@willett.pgh.pa.us>
Date: 17 Dec 90 00:23:33 GMT
Organization: String, Scotch tape, and Paperclips.  (in Pgh, PA)
Lines: 67

Category 6,  Topic 15
Message 224       Sat Dec 15, 1990
F.SERGEANT [Frank]           at 17:21 CST
 
 Clyde Phillips writes
 CP> I'd also welcome any discussion of creating "generic" FORTH 
 CP> engines using FPGA's that WOULD BE affordable
  Yes, this sound very interesting to me.  I considered it and would  still
like to design my own "for fun."  Every time I've looked it over  and priced
the Xilinx chips or whatever, it has seemed that it would  still be cheaper to
buy a NOVIX or RTX or SC32, especially if I  considered the design time.  Do
you know any current prices (in  singles) for any PLDs within the range of
"4000 gates" or more?  It  would seem they should get cheap because of their
possible large  market, but they didn't seem cheap a year or so ago when I
last looked.
  Your idea of a generic or PD design answers my objection about the  design
costs, more or less.
  I even liked the idea of Phil Koopman's 16 bit WISC based on 74xxXX  chips
(and 74xx181 ALUs).  I'm tired, though, of wire-wrapping, and  probably won't
do much more unless I get well paid for it.  That was  one of the appealing
aspects of the large programmable logic chips. I  could do much the same
thing, but "wire-wrap" them very quickly with a  PLD programmer.
  Marty suggests that we're unlikely to get the speed we'd like, and I 
believe him.  It is a shame, because it is an idea that "ought to  work."  I'm
hoping to hear from you and others on how it could really  work.
  I think break-throughs often start out as just such an idea.  "You  know, it
OUGHT to be possible to do such & such."  "How?"  "I don't  know yet, it's
just a feeling so far."  Like the electric light bulb,  perhaps.
  Anyway, I have another one of these "there ought to be a way ..."s to  toss
out.  I don't really see how it can work yet, but it feels like  there ought
to be a way.  We've got all these mass-produced CHEAP  microprocessors, such
as the Z80, 6802, Z8, Super-8, even 68000s.  You  can probably find some of
these from time to time for under a dollar  each in singles!  There OUGHT to
be a way to combine them to make a  super powerful processor super cheap!  I'm
not necessarily talking  about mixing families.  Perhaps a board with 8 6802s
on it.  Maybe we'd  pay a little (a lot?) more to use one-time programmable
Z8s etc so we  don't have to wire up separate EPROMs, etc.  I think the
68HC11s are  overpriced at this time for combining as I've suggested.
  There are problems.  Do we connect them in parallel so each processor  works
on 8 bits of a wider data-item, say 4 across - giving 32 bits?   If so, we
have the terrible delay waiting for the carry between each  stage.  We might
as well let a single processor do its own multiple  precision math.  Or, would
we "pipe-line" it, so that even though the  32-bit operation takes 4 (or more
likely 8 or more) times as long (as  the others must read something external
to get the carry), as soon as  each processor finished its part of the 32-bit
op it would be free to  start the next one.  So, here, I don't really see
where we've gained  anything.
  Instead, perhaps the stacks and registers (for the Forth machine) are  NOT
kept inside the cheap micros, but accessed through external I/O.   Maybe this
offers a speed advantage somehow.  If we have to put  together very much
hardware, we are back to the situation where it is  cheaper to pay the price
for a commercial Forth engine.  Perhaps the  separate micros could specialize
somehow, rather than working in  parallel on different parts of the large
word.  All of this leads us  into the maze of complexity of parallel
processing.
  At this stage it is just a dream.
 CP> The Zilog thing is does anyone have FORTH's for the Super Z-8
  Yeah, I have a Forth running on a Super 8 that I've bread boarded.   Other
than some glue chips it consists of the Super8 chip, a 32K static  RAM chip,
and an 8K EPROM into which I put a version of Pygmy Forth.  I  use my PC XT as
a terminal and disk server for it, over a serial line.   I haven't touched it
in probably a year.  As soon as I get caught up  (ha ha) I'm going to do some
more work with it.
  -- Frank
-----
This message came from GEnie via willett through a semi-automated process.
Report problems to: dwp@willett.pgh.pa.us or uunet!willett!dwp