Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!sun-barr!olivea!uunet!sjsca4!poffen
From: poffen@sj.ate.slb.com (Russ Poffenberger)
Newsgroups: comp.sys.ibm.pc.misc
Subject: Re: Math Coprocessors
Message-ID: <1991Feb26.201401.22738@sj.ate.slb.com>
Date: 26 Feb 91 20:14:01 GMT
References: <1991Feb19.214611.31858@slate.mines.colorado.edu> <84380001@hpfcmdd.hp.com>
Reply-To: poffen@SunOS (Russ Poffenberger)
Organization: Schlumberger Technologies, ATE division, San Jose, Ca.
Lines: 26

In article <84380001@hpfcmdd.hp.com> edm@hpfcmdd.hp.com (Ed Moore) writes:
>The MHz rating on an 80387 chip is the speed at which Intel certifies it
>will run.  If you put a 16 MHz chip into a 20 MHz computer, the clock in the
>computer forces the chip to run at 20 MHz.  That should work _most_ of the
>time.  But if Intel's testing could certify the chip at 20 MHz they would
>have sold it as a 20 MHz chip; something didn't work right some of the time.


This isn't necessarily true. What a vendor certifies as a particular speed
rating on a chip make be because of several factors. Some chip designs are
rated up to a certain speed. An example is the 80386/33 which is a different
chip altogether than slower versions. (not by logic design, but chip geometries
such as the die shrink size.)

Another factor is the packaging. Higher speed generates more heat. Some
packaging is not certified to dissipate enough heat at higher speeds.

Even for identical chips/packaging, vendors will sometime test chips up to a
certain speed based only on market demand at the time. If they are short on
25Mhz versions, they will test only to 25Mhz, and not spend the extra time
or effort to certify them to a higher speed, even though they may work.

Russ Poffenberger               DOMAIN: poffen@sj.ate.slb.com
Schlumberger Technologies       UUCP:   {uunet,decwrl,amdahl}!sjsca4!poffen
1601 Technology Drive		CIS:	72401,276
San Jose, Ca. 95110             (408)437-5254