Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!linus!philabs!cmcl2!seismo!ut-sally!utastro!jk
From: jk@utastro.UUCP (John Krist)
Newsgroups: net.micro.ns32k
Subject: Benchmarks - Intel vs 32 bitters
Message-ID: <801@utastro.UUCP>
Date: Tue, 10-Jun-86 11:57:00 EDT
Article-I.D.: utastro.801
Posted: Tue Jun 10 11:57:00 1986
Date-Received: Sat, 14-Jun-86 05:37:57 EDT
Organization: U. Texas, Astronomy, Austin, TX
Lines: 14
Keywords: segmentation vs. linear addressing


      For those who keep pointing at the higher ratings for Intel
  chips on the Dhrystone compared to the 32k ratings, I challenge you
  to do any benchmark using arrays, looping, etc. with a number greater
  than 64k.  Once you hit that magic number, the segmentation of the 
  Intel chips comes into play and down goes the speed.  Try doing the
  Sieve of Eratosthenes (sp?) with the number of primes being 80,000.      
  Then let's talk about their speed versus the 32k.

      Also, remember that one benchmark means nothing.  Why use a Cray
  for Lotus 1-2-3 instead of a PC?

  John Krist
  U. Texas Astronomy Dept.