Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/18/84 +MMDF+MULTI+2.11; site uel
Path: utzoo!watmath!clyde!burl!ulysses!allegra!mit-eddie!think!harvard!seismo!mcvax!ukc!qtlon!uel!glenn
From: glenn@uel (Glenn Wright )
Newsgroups: net.unix-wizards,net.lang.c
Subject: Re: AIM Benchmarks
Message-ID: <518@uel>
Date: Tue, 3-Dec-85 06:25:18 EST
Article-I.D.: uel.518
Posted: Tue Dec  3 06:25:18 1985
Date-Received: Fri, 6-Dec-85 08:17:44 EST
References: <567@scc.UUCP>
Organization: Unix Europe Ltd, London UK
Lines: 33
Xref: watmath net.unix-wizards:15970 net.lang.c:7310

> *** 
> 
> 	I am interested in finding out what people
> think of the AIM benchmarks.  How useful are they
> to you?  How would you improve them?
>
 
I have used the AIM benchmarks for testing numerous systems. They 
take sometime to set up as they are mainly based on shell scripts
which have to be tailored to produce banners and printouts to your
requirements. As far as the tests go they are comprehensive within 
each individual test, but the problem is (as with many benchmarks)
bundling these results together and judging the overall results.

As I remember the tests cover the following:
	terminal i/o
	disk block movements
	multi-user (many different processes)
	matrix multiply (diiferent sizes)
	timing of the cc command
	A compute bound program test
	...

The AIM bencmarks (I feel) compare favourably with the "Byte"
magazine tests, but remember they are only useful when comparing
two systems with EXACTLY the same test!

	
	 
-- 
Glenn Wright
============
UNIX Europe Ltd, London UK.		{mcvax!ukc!}uel!glenn