Path: utzoo!utgpu!news-server.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!snorkelwacker!apple!brutus.cs.uiuc.edu!ux1.cso.uiuc.edu!sparc1!test
From: test@sparc1
Newsgroups: comp.sys.ibm.pc
Subject: Re: 8087 Coprocessor
Message-ID: <1990Jul27.143309.12059@ux1.cso.uiuc.edu>
Date: 27 Jul 90 14:33:09 GMT
References: <169@evtprp0b.UUCP> <1990Jul11.172043.525@bigsur.uucp> <532@metaware.metaware.com>
Sender: usenet@ux1.cso.uiuc.edu (News)
Distribution: usa
Organization: University of Illinois at Urbana
Lines: 20

johnv@metaware.metaware.com (John Vinopal) writes:

>> Stuff about 8087's overheating deleted.

>It is common (risky) practice to overdrive chips.  There were 25Mhz 386 boxes
>long before the i386-25 was released.  Why then can this not be done with the
>387 chips?  I have heard that at least with the 287s the chips are all basically
>the same but they have been verified at different clock speeds.  Thus an 8Mhz
>is no different than a 10Mhz, but for the fact that the 10 is guarenteed to 
>work at 10.  Could someone expand on why a 25Mhz 387 could not be used in a
>33Mhz box?  If overheating is the only problem then whats the deal, install a
>cooling system.
>	
>						johnv@metaware.com
     
Chips which are rated for a specific speed are tested by driving them at differ-ent speeds; when a speed is reached where the chip no longer works reliably,
it's assigned the speed at which it last functioned properly.  If you run a chipfaster than its rating (which, as you say, is a common occurrence), you're prob-ably safe if you only push it a couple of Mhz.  I wouldn't recommend buying a
16 Mhz 386 and trying to run it at 33 Mhz, however :-).

16 Mhz 386, and trying to run it at 33 Mhz, however :-).