🚀 go-pugleaf

in article <21709@cup.portal.com>, mslater@cup.portal.com (Michael Z Slater) says:

> (This is a continuation of an earlier posting, which was truncated by
> accident just as it was getting to the heart of the matter)

> 2. If a coprocessor is something like an 8087 that executes from the
>    same instruction stream as the main processor, then what is a good name
>    for an independent processor that works with another processor?
>    Attached processor?  Slave processor?

I've never seen any formalism on the names myself, but I question whether the
term "coprocessor" is the appropriate one for something like an 8087.  At
least if the 8087 works anything like Motorola "coprocessors", such devices
become basic extensions to the CPU model you're working with, and at least
from the programming prespective, there's no way you can tell that they
are at all distinct from the main CPU.  The term "coprocessor" for such a
device never worked for me; maybe it's appropriate, but not specific enough.
I'd guess that there are really three different things that might be called
coprocessors.  The CPU-extension processor, like a 68882, is one kind.  The
second would be a helper processor of some kind.  In the Amiga system we have
a thing like this, a video coprocessor called the "Copper" for short.  This
thing executes it's own instruction stream and does video operations to free
up the CPU from having to deal with some of the drudgery of managing a video
display.  This is really a slave device -- it couldn't do anything useful
without a main CPU to help it out.  The third kind of coprocessor might exist
as a helper or an equal in a multiple CPU system, but it's a full fledged 
CPU in it's own right.  

As for the Intel 80860, it sure seems to me that the device fits this last
category.  It might be put into a system as a slave to the main CPU, perhaps
managing a video display or something similar.  But it's a full fledged CPU
in it's own right, it'll apparently even run UNIX, albeit poorly.  I kind of
wonder if this was the original design goal for the part, but that's 
apparently what it turned out to be.

> Michael Slater, Microprocessor Report, mslater@cup.portal.com
> 415/494-2677   fax: 415/494-3718
-- 
Dave Haynie Commodore-Amiga (Systems Engineering) "The Crew That Never Rests"
   {uunet|pyramid|rutgers}!cbmvax!daveh      PLINK: hazy     BIX: hazy
                    Too much of everything is just enough