Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!sdd.hp.com!usc!snorkelwacker.mit.edu!bloom-beacon!eru!hagbard!sunic!mcsun!i2unix!inria!ftc!ndoduc
From: ndoduc@framentec.fr (Nhuan Doduc)
Newsgroups: comp.sys.ibm.pc.misc
Subject: Re: BENCHMARKS !!!!
Message-ID: <1552@ftc.framentec.fr>
Date: 26 Nov 90 16:22:22 GMT
References: <36221@cup.portal.com>
Sender: news@framentec.fr
Lines: 27

In <36221@cup.portal.com> mmm@cup.portal.com (Mark Robert Thorson) writes:
>do you think about them?  Specifically, the Landmark, Power Meter, MHz,
>and Norton SI benchmarks.  Are there others which should be added
>to that list?
Landmark is not bad except that your x86 will "run at 116MHz" especially if
you have cache. SI is like dhrystone: unavoidable. PM is the best if not the
"less-worse", still there are plenty others: PCTech, PCLabs, Demo from Compaq..
and of course we'll not forget Byte's

>What makes a PC benchmark different from a benchmark for workstations,
>like the SPEC benchmark?  Do PC benchmarks measure I/O?  Do they provide
>a single number, or separate numbers like the SPEC benchmark suite?
Power Meter gives you a set of numbers, CoreTest is a good test about disk (we'
ve found correlation with our SSBA's I/O part). Workstation benchmark basically
should not be different, in principle, to , say, PC benchmark (except such
trivialities like multitasking, multiuser issues)

>I suppose all PC benchmarks run in real-address mode, so that eliminates
>performance differences due to the processor architecture, which is
>something SPEC is designed to reveal.
As far as I know, all PC benchmarks DO run in real mode (the doduc (!#@!) does
run in protected mode too -self advertisement- !#@!).

--nh
Nhuan DODUC, 
Framentec-Cognitech, Paris, France, ndoduc@framentec.fr or ndoduc@cognitech.fr,
Association Francaise des Utilisateurs d'Unix, France, doduc@afuu.fr