home .. forth .. misc mail list archive ..

Re: Hardware/Software


Forwarded message:
>From sherwin@filoli.com Tue Jan 30 12:04:29 1996
>Message-Id: <199601300204.SAA24428@sunspot.filoli.com>
>Content-Type: text/plain
>Mime-Version: 1.0 (NeXT Mail 3.3 v118.2)
>From: Sherwin Gooch <sherwin@filoli.com>
>Date: Mon, 29 Jan 96 18:04:32 -0800
>To: Wayne Morellini <waynem1@cq-pan.cqu.edu.au>
>Subject: Re: Hardware/Software
>Cc: misc@pisa.rockefeller.edu (misc)
>Reply-To: Sherwin_Gooch@filoli.com
>References: <199601290400.OAA08526@cq-pan.cqu.edu.au>
>
>Wayne,
>
>What do I think?  I think you've completely missed the boat:
>
>The sole reason that MISC processors are of interest is that they  
>minimize the real-estate required to implement a processor.
>
>Why is this of interest?  Because you can put a maximum number of  
>processors in minimal real-estate [read:  on a single chip].  This  
>enables fine-grained computing.
>
>If you need to worry about interfacing I/O devices to individual  
>MISC cells, you have the wrong architecture.
>
>Think about it.
>				Sherwin Gooch
>


Rational discussion, reasoning or debate.

Say that I give somebody a bare chip to sell to the public, not say a 80X86
or something they could use as a plug in upgrade.  Firstly the people
that matter, the public with the money, might be curious but would not be
very interested.  Secondly before we got that far the dealers would think we
were crazy or something at least, possibly idiots.  We need to package the
chips in sell able products, that need extra components to work.  The point
with Misc is not only does it take less real estate (a minor point when a
million transistors can be had for below $10, or more to our example a 100,
000 transistors (286).  But more importantly it uses less power, consumes
less code space (memory costs), costs less and is faster.  Remember a fast
32-bit Disc core that is already popular and in existence uses around 27K
transistors (Arm 7) and has chips with all periphial chips integrated
(I/O, PC-Card, Virtual Memory system, Graphic etc) for I believe around
350k transistors, the extensive features probably more than match all the
planned cells on the list put on the lang.forth newsgroup last year, about
Computer Cowboys chip fabrication.  So until we have more cells and finished
designs with them in we need to add for functionality.  Having listed the
obvious other benefits of Misc above that fit in with the publicly declared
intentions of Mr Ting and Jeff, who in literature (electronic also) have
mentioned PDA's, Video Games and Networking amongst others,  we need to add
this functionality to archive these things.  We are not likely to have chips
with keyboards, LCD'S, IO, Disk Interface (rather than drive) integrated
on them to sell directly to the public as a finished product, unfortunately
if I or anybody else wants to add descent 16-Bit graphics, IO ports, memory
systems etc (depending on the application) we need to do it ourselves, as
the cells are not available.  Even though the chip with out added
functionality as it is may suit your area of the market for others the story
is different. I suggest we try another boat :)   


I fully agree with your comments though, about fitting the maximum number of
processors in the minimum real-estate.


We have potential here not only to promote Forth, Stack based chips, Misc,
Chucks designs, but for our own mutual interest, so is there some other
comments.



Wayne Morellini

waynem1@cq-pan.cqu.edu.au


--