Message-ID: <921@phs.UUCP>
Date: Thu, 30-Aug-84 18:42:52 EDT
Article-I.D.: phs.921
Posted: Thu Aug 30 18:42:52 1984
Date-Received: Mon, 3-Sep-84 11:00:41 EDT
Organization: Duke Physiology
Lines: 17


	I recently had a chance to try out some programs written in C
(CI-86) on an AT with the 80287 floating point chip.  One program did
number crunching operations primarily and we tested versions with and
without the floating point chip.  Both versions had been compiled
on an XT.  The non-80287 version ran about 3x faster on the AT, compared
to the XT -- about the same speed improvement as other programs IBM
demonstrated.  The 80287 version ran only about 10% faster on the
AT, compared with the XT!  Inquiries to IBM brought the comment that
the internal speed of the 80287 is essentially the same as the 8087.
Pity... -- we had hoped the AT would be faster in number crunching
operations!

	Mailen Kootsey
	Duke University
	phs!jmk