Newsgroups: comp.sys.mac.advocacy,comp.sys.mac.misc,comp.lang.dylan,alt.computer.consultants
Path: cantaloupe.srv.cs.cmu.edu!rochester!cornellcs!newsstand.cit.cornell.edu!news.kei.com!newsfeed.internetmci.com!in1.uu.net!wb3ffv!mail!eshea
From: Ed Shea <eshea@mail.bcpl.lib.md.us>
Subject: Re: Angry, very angry, at Apple Computer Corp. 
Sender: usenet@abs.net
Message-ID: <Pine.SOL.3.91.951115125249.19934A-100000@mail>
X-Sender: eshea@mail
In-Reply-To: <48b1rr$fn6@nyx.cs.du.edu> 
Date: Wed, 15 Nov 1995 18:03:40 GMT
X-Nntp-Posting-Host: mail.bcpl.lib.md.us
Content-Type: TEXT/PLAIN; charset=US-ASCII
References: <463pvk$gca@elaine35.Stanford.EDU>  <Pine.SOL.3.91.951110151445.1974A-100000@mail> <48b1rr$fn6@nyx.cs.du.edu> 
Mime-Version: 1.0
Organization: ABSnet Internet Services, Inc. - info@abs.net - (410)-361-8160
Lines: 48
Xref: glinda.oz.cs.cmu.edu comp.sys.mac.advocacy:88087 comp.sys.mac.misc:120294 comp.lang.dylan:5922

On 14 Nov 1995, Joel Mawhorter wrote:
>: Ed Shea wrote:
> : Just playing a little devil's advocate here on benchmarks.  First, saying 
> : something is 50% better than something else tells me little.  Tell me 
> : absolutes. My wife always says "But I got it for 25% off."  I'd like to 
> : know how many dollars are going out of the checking account.
> 
> : As far as your point is concerned, Aron, it proves at best that the 
> : Mac/Photoshop combination is faster.  But is what percentage, if any, of 
> : that improvement is due to the software.  Is the Mac version optimized 
> : for the PowerPC chip? Is the PC version 386 code (i.e. not optimized for 
> : the Pentium)?
> 
> However that is what is available.  You can't make arguments based on 
> vaporware (a version of Photoshop optimized for the Pentium).
> 
> : Obviously a if you're running only Photoshop that might be enough.  If I 
> : have to run other software combinations, I'd need a better benchmark.  
> 
> :  Ed Shea
> : eshea@mail.bcpl.lib.md.us
> 
> -Joel
> 
> 
You miss the point entirely, Joel.  Simply saying, as was said at the 
outset that a Mac of a given configuration is 50% more faster on a given 
piece of software than a PC tells me nothing about the machines, except 
that I can run that piece of software on that Mac faster than I can on 
that PC.  To determine the validity of the "benchmark" as a measure of 
the machine's capabilities I do have to know the variables, including 
the optimization of the software.  That way I can judge how much of the 
performance gain is due to the software rather than the machine. I am not 
making arguments based on vaporware, but rather on the software that is 
available.  Take a look at Intel's P6 which, according to some reports is 
slower running 16 bit software than the Pentium, but providees a 
performance gain running 32 bit software.  If I benchmark the chips using 
16 bit software only, I might conclude that the Pentium is a more 
powerful chip.  But if I look at the software it is running and the fact 
that the P6 is optimized for 32 bit performance I conclude otherwise.  

to put it all in a less subtle manner, beenchmarks are worthless unless 
you know the details - including thee nature of the software forming the 
basis of the test.

Ed Shea
eshea@mail.bcpl.lib.md.us

