Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: Notesfiles; site smu.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!mhuxj!ihnp4!inuxc!pur-ee!uiucdcs!smu!mike
From: mike@smu.UUCP
Newsgroups: net.micro.pc
Subject: Re: AT arithmetic problem (FORTRAN).
Message-ID: <15000004@smu.UUCP>
Date: Thu, 20-Sep-84 16:30:00 EDT
Article-I.D.: smu.15000004
Posted: Thu Sep 20 16:30:00 1984
Date-Received: Wed, 26-Sep-84 01:16:30 EDT
References: <13362@sri-arpa.UUCP>
Lines: 15
Nf-ID: #R:sri-arpa:-1336200:smu:15000004:000:584
Nf-From: smu!mike    Sep 20 15:30:00 1984

<>

I'm not sure how your FORTRAN proogram worked, but it probably did
something flakey to detect whether an 87 was in place.  The 8086
should not really be able to figure this out.  Because the
processor/coprocessor interface between the 286 and 287 is so much
different than that between the 86 and 87, I am not surprised that it
does not work.  The 286 does have the capability of trapping upon
recognition of an ESC instruction via a couple of bits in the MSW.  I
have heard, however, that current AT software doesn't support that
kind of thing.

Mike McNally
...convex!smu!mike