Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10 5/3/83 based; site hou2f.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!mhuxr!mhuxt!houxm!hou2f!jja
From: jja@hou2f.UUCP (J.ARLETH)
Newsgroups: net.micro
Subject: 286 vs. 68k
Message-ID: <475@hou2f.UUCP>
Date: Thu, 2-May-85 17:41:19 EDT
Article-I.D.: hou2f.475
Posted: Thu May  2 17:41:19 1985
Date-Received: Fri, 3-May-85 05:22:36 EDT
Organization: AT&T Bell Labs, Holmdel NJ
Lines: 45


>Did any of the benchmarks you refered to happen to require more than 64K
>of data? Probably not, because you would be ashamed of the results instead
>of proud of them if that were the case. 80*86's make nice LSI-11 replacements,
>but I wish Intel would stop trying to convince people that are actually useful
>for real work. Todays jobs require more than 64K of data space, and the 80*86
>is useless for that. Maybe if Intel designs a new chip instead of upgrading
>the 4004, you might hit upon something good. Then again, maybe not, the 432
>never quite made it. So I guess Intel should stick to doing 4004 upgrades.
>You will always have a market full of people more interested in quick delivery
>and backwards compatibility, and who don't care at all about performance.
>
>Why am I flaming so hard at Intel? Because I own a 186 based system, and I'm
>sick and tired of working around its limitations.
>--
>
>			  John McNamee
>		..!decvax!philabs!sbcs!bnl44!jpm
>			 jpm@BNL44.ARPA
>
>

		Yeah!!
			What he said.

	It is actually amusing to look at the benchmarks that have been 
published comparing these two processors.  Intel insists on either
comparing the two processors of the same clock rate or comparing short
string copy programs.  Another set of benchmarks that I've seen
are based on a number of COLBOL programs.  

			And another thing.

	Backward compatability and state-of-the-art performance are 
obvious trade offs.  Intel's 286 (has a mode which) is backwards 
compatabile with its 8086 processor.  It paid a price for this.
Motorola paid a different price (software incompatability) by designing
the 68000 without 6809 compatability constraints.  This is why it yields
better performance for most 'real' jobs.


			Jim Arleth
			hou2f!jja

...The usual disclaimers.