Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/18/84; site rochester.UUCP
Path: utzoo!linus!philabs!cmcl2!seismo!rochester!dibble
From: dibble@rochester.UUCP (Peter C. Dibble)
Newsgroups: net.micro.68k
Subject: 68020 Benchmarks
Message-ID: <10973@rochester.UUCP>
Date: Fri, 9-Aug-85 21:31:46 EDT
Article-I.D.: rocheste.10973
Posted: Fri Aug  9 21:31:46 1985
Date-Received: Mon, 12-Aug-85 07:32:57 EDT
Distribution: net
Organization: U. of Rochester, CS Dept.
Lines: 34

A few months ago somebody posted a set of C (language) benchmarks:
sort, fib-numbers, seive, and a floating point exercise.  I sent
sort, fib and prime (with a knight's tour program that showed up
here about the same time) to Gimix.  They ran them on a 68010
machine that they have around and on the single board 68020 system
they are preparing for sale (Late Sept?).  I don't have times
for the knight's tour, but here are the results for the other 
three:

Program     68010        68020
Sort         42 Sec        10.5 Sec
Fib          48 Sec        12.7 Sec
Seive        4 Sec          0.9 Sec

All tests used OS-9 and Microware's C compiler.
The Gimix machine runs at 12.5 MHz with one wait state.
The 68010 machine was a Mizar (sp?)

I ran the knight's tour for a 6x6 board on my 6809 system
in about 10 minutes (a 5x5 board took only about 2 seconds
so I didn't try a 7x7 board).  I expect to have a time for the tour
on the 68020 next week.  If anyone has times for other processors 
I'd like to hear about them.

Two side issues:
I understand that I'm about to be quoted in UNIX World saying that OS-9 
performance is far worse than Unix's(tm :-).  I did say something
like that.  My 6809 system is not as fast as the 780 I use.
In any case, on a 68020, OS-9 evidently does NOT perform
poorly.

I haven't mentioned any non-Motorolla processors here.  I
enjoyed the 80*/68* discussion, but I'm not trying to start
it again!