Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!mailrus!cs.utexas.edu!ut-emx!dell!mustang!jrh
From: jrh@mustang.UUCP (jrh)
Newsgroups: comp.unix.i386
Subject: Re: Cyrix FasMath CX-83D87 80387 replacement with 386/Unix
Message-ID: <4529@dell.dell.com>
Date: 30 Nov 89 19:05:12 GMT
References: <1989Nov29.215150.53@jdyx.UUCP>
Sender: news@dell.dell.com
Reply-To: cs.utexas!bigtex!dell!mustang!jrh (jrh)
Distribution: usa
Organization: Dell Computer Corp.
Lines: 22

In article <1989Nov29.215150.53@jdyx.UUCP>, tpf@jdyx.UUCP (Tom Friedel) writes:
> Has anyone used this chip with 386/ix?  From the manual, "Cyrix
> CX-83D87 is a CMOS VLSI integrated circuit that is pin compatible
> and software compatible with the 80387, yet ... achieves 4 to 10
> times greater performance ... by implementing floating point primitives
> operation in hardware."  
> Given one floating point intensive process (a ray trace) I am 
> wondering what performance difference I would see?
> 

Well, there is an article in Info World (11/27/89) reporting that Intel is
disputing the compatibility claims of coprocessors from other vendors.  A
couple cited in the article were the IIT-2C87 and the Cyrix Fasmath 83D87.
Interestingly, the article states that Cyrix has released a report claiming
to have "identical numerical and status results" to the chip compared to the
GET THIS >>>>> Intel 80287.  Notice that the comparison reported is not to the
80387.  I personally don't know either way, but that's the guts of the article.

+------------------------------------------------------------------------+
| James R. Howard                | Insert Standard Disclaimer Here...    |
| cs.utexas.edu!dell!mustang!jrh |                                       |
+------------------------------------------------------------------------+