Path: utzoo!attcan!uunet!zephyr.ens.tek.com!uw-beaver!mit-eddie!mintaka!olivea!apple!agate!violet.berkeley.edu!ilan343
From: ilan343@violet.berkeley.edu
Newsgroups: comp.sys.ibm.pc.hardware
Subject: Re: Cyrix CX-803D87-20 Coprocessor
Message-ID: <1990Sep27.060417.23408@agate.berkeley.edu>
Date: 27 Sep 90 06:04:17 GMT
References: <14244@shlump.nac.dec.com> <1990Aug4.220844.7349@water.waterloo.edu> <128@thor.UUCP>
Sender: usenet@agate.berkeley.edu (USENET Administrator)
Organization: University of California, Berkeley
Lines: 16

In article <128@thor.UUCP> scjones@thor.UUCP (Larry Jones) writes:
>There are now three sources of plug-compatible coprocessors.  (The
>Weitek is faster and more expensive, but it's not plug-compatible --
>you need a special socket [or adapter board] and special software to
>use it.  Standard 80287/80387 code can't use it.)
>
>
>Cyrix makes clones of both the 80287 and 80387.  These are slightly
>faster than Intel's parts and a little bit cheaper.
>
>IIT also makes clones of bothe the 80287 and 80387.  These are a
>good bit faster than Intel's parts for about the same price.

I have seen adds for Cyrix FastMath coprocessor (80387 compatible) that
claims 3X speedups.  Are there any benchmarks to support this?
Also, is compatibility an issue?