Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!mailrus!wuarchive!wugate!uunet!cbmvax!daveh
From: daveh@cbmvax.UUCP (Dave Haynie)
Newsgroups: comp.sys.ibm.pc
Subject: Re: Compaq find problem with chip
Message-ID: <8374@cbmvax.UUCP>
Date: 1 Nov 89 17:14:12 GMT
References: <788@awdprime.UUCP>
Organization: Commodore Technology, West Chester, PA
Lines: 27

in article <788@awdprime.UUCP>, ron@woan.austin.ibm.com (Ronald S. Woan) says:
> Keywords: defect in all '486 chips thus far!
> Xref: cbmvax comp.arch:12529 comp.sys.intel:1056 comp.sys.ibm.pc:41838

> In article <443@ssp1.idca.tds.philips.nl>, dolf@idca.tds.PHILIPS.Grunbauer) writes:
>> What is the actual problem ? What chips 80486 and/or 80487 ?
>> Could someone give more details or did they only say there is a problem
>> and not what the precise problem is ?

> According, second hand, to the San Jose Mercury news 27 Oct 89, a defect 
> causes incorrect calculations given a rare sequence of floating point
> operations. Ghastly!

The story I heard was that any integer operation immediately following the 
COS-SIN function (whatever they call the function that returns both Sine and
Cosine) results in an incorrect result from the on-chip FPU.  Intel is
apparently trying to minimize the impact of this by claiming that it's rare
to interleave floating point and integer instructions this way.  Which is
real strange, since that's exactly what you'd expect to do in order to 
cover stalls in the floating point pipeline.  This information also come to
me second hand, from two separate sources though.

> 						Ron
-- 
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