Path: utzoo!attcan!uunet!olivea!mintaka!bloom-picayune.mit.edu!athena.mit.edu!acook
From: acook@athena.mit.edu (Andrew R Cook)
Newsgroups: comp.sys.ibm.pc.hardware
Subject: Re: coprocessor... how fast???
Message-ID: <1991Jan28.150353.25420@athena.mit.edu>
Date: 28 Jan 91 15:03:53 GMT
References: <1991Jan27.172506.15133@ncsuvx.ncsu.edu>
Sender: news@athena.mit.edu (News system)
Reply-To: acook@athena.mit.edu (Andrew R Cook)
Organization: Massachusetts Institute of Technology
Lines: 27

In article <1991Jan27.172506.15133@ncsuvx.ncsu.edu>, big@hobbes.catt.ncsu.edu (Alan Porter) writes:
|> 
|> Forgive me, but I am still new at this game.
|> How fast must a math coprocessor be to work and to be
|> effective?
|> 
|> I have a 286-16.  Do I need a 287-16 or a 287-8 or what?
|> Will a slower speed still work, but not as effectively?
|> 
|> 
|> big@hobbes.catt.ncsu.edu              Alan Porter
|> NCSU Computer Theme Program
|> Box 21536 NCSU / Raleigh, NC 27607 USA / +1 (919) 737-6121

The general rule of thumb for 86 and 286 machines is that the coprocessor
runs at 2/3's of the main CPU speed.  For your machine, you'll want a 12Mhz
coprocessor.  Getting a lower rated coprocessor is asking for trouble.  The
computer has no way of telling what speed your coprocessor is running at.  
The clock speed is set at something by the manufacturor.  If the MPU cannot
keep up with the set rate, it will trip up and can cause faulty answers, or
in worst case, lock-up you machine.  

Check out AMD's coprocessor.  It's only $99, and fully gauranteed to work.
I got one, and it works great.

Andy Cook
acook@athena.mit.edu