Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/5/84; site omen.UUCP
Path: utzoo!watmath!clyde!cbosgd!ihnp4!houxm!vax135!cornell!uw-beaver!tektronix!reed!omen!caf
From: caf@omen.UUCP (Chuck Forsberg WA7KGX)
Newsgroups: net.micro
Subject: Re: 68020 benchmarks??
Message-ID: <159@omen.UUCP>
Date: Wed, 15-May-85 14:32:18 EDT
Article-I.D.: omen.159
Posted: Wed May 15 14:32:18 1985
Date-Received: Fri, 17-May-85 00:15:49 EDT
References: <272@petfe.UUCP>
Organization: Omen Technology, Portland
Lines: 22

The unmentioned gotcha in Intel's benchmark comparisions is the ages old
apples-vs-oranges comparision trick.

Most 68k C compilers use 32 bit ints and true 32 bit addresses.

Most 8086 and 80286 C compilers don't support this combination.

For a fair comparision, you MUST compile with the "HUGE" model which
allows for objects >64k.  That means Lattice Large Model or Microsoft
HUGE model.

Even these aren't fair comparision as the stack space is still limited to
64k on the 8086 chips but not on 68k.

I have an up to date comparision of sieve benchmarks (slightly modified
from the BYTE code) on Telegodzilla as "bench/siev.doc".  The source is
available as "bench/siev.c".  Telegodzilla is at 503-621-3746 1200, 300
bps.  Hit RETURN untilthe system recognizes your speed.
-- 
Chuck Forsberg WA7KGX	..!tektronix!reed!omen!caf
Omen Technology Inc 17505-V NW Sauvie IS RD Portland OR 97231
Voice: 503-621-3406	Modem: 503-621-3746 (Hit CR's for speed detect)