Path: utzoo!utgpu!news-server.csri.toronto.edu!bonnie.concordia.ca!uunet!zaphod.mps.ohio-state.edu!uakari.primate.wisc.edu!dali.cs.montana.edu!masscomp!peora!tarpit!bilver!wbeebe
From: wbeebe@bilver.uucp (Bill Beebe)
Newsgroups: comp.sys.intel
Subject: Re: 287 on a 386 machine
Message-ID: <1991Jan12.014902.28338@bilver.uucp>
Date: 12 Jan 91 01:49:02 GMT
References: <1991Jan9.004654.21586@research.canon.oz.au> <916@tuura.UUCP>
Organization: W. J. Vermillion - Winter Park, FL
Lines: 20

In article <916@tuura.UUCP> risto@tuura.UUCP (Risto Lankinen) writes:
>
>The socket is a bit peculiar:  There are pin holes in square arrangement for
>80387, and in two parallel rows for 80287, some of which are shared at the
>same place.  Also, the 287 is 2/3 the frq of the CPU (like 12 MHz <-> 16 MHz)
>as it often is with 287.
>
>I don't know how much speed would be gained by using 387 instead.

 In the beginning, when the 80386 was first introduced, Intel did not have
the 80387 ready. So they published an ap note that explained how to
interface the 80287 to the 80386. Many systems houses did this at first to
ship product. Some were foresighted enough to include sockets for both
the 287 and 387.

 The 80287 is currently rated at 10 MHz. max. The 80387 _starts_ at 16 MHz
and goes up to 33 MHz. On a 16 MHz 386 system, the 387 will run 3 to 5 
times faster than the 287. Part of this is the higher clock, part is
the optimized microcode and hardware of the 387. Keep in mind that
the 387 has trigonometric functions, and the 287 does not.