Mailing-List: contact cygwin-help AT cygwin DOT com; run by ezmlm
List-Subscribe: <mailto:cygwin-subscribe AT cygwin DOT com>
List-Archive: <http://sources.redhat.com/ml/cygwin/>
List-Post: <mailto:cygwin AT cygwin DOT com>
List-Help: <mailto:cygwin-help AT cygwin DOT com>, <http://sources.redhat.com/ml/#faqs>
Sender: cygwin-owner AT cygwin DOT com
Mail-Followup-To: cygwin AT cygwin DOT com
Delivered-To: mailing list cygwin AT cygwin DOT com
Reply-To: Cygwin List <cygwin AT cygwin DOT com>
Message-Id: <6.0.1.1.0.20040323221912.03c80e78@127.0.0.1>
X-Sender: 
Date: Tue, 23 Mar 2004 22:20:31 -0500
To: Robert Mecklenburg <mecklen AT comcast DOT net>,
   Cygwin Users List <cygwin AT cygwin DOT com>
From: Larry Hall <cygwin-lh AT cygwin DOT com>
Subject: Re: GNU Make performance question
In-Reply-To: <16480.64460.156000.338305@gargle.gargle.HOWL>
References: <16480 DOT 64460 DOT 156000 DOT 338305 AT gargle DOT gargle DOT HOWL>
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"

At 10:09 PM 3/23/2004, you wrote:
>I am writing a document on the performance of gnu make and various
>makefiles and came across something I can't explain.
>
>To measure the speed of gnu make $(subst...) versus sed I wrote this
>simple makefile:
>
>  p1000 := ...<1000 character string with periodic semicolons>...
>  # This simple assignment 10000 times.
>  x := $(subst ;, ,$(p1000))
>  ...
>
>I then run this makefile 10 times and average the times.  I ran the
>makefile on Windows XP with Cygwin and a Linux system with RedHat 9:
>
>  Windows XP     1.8 GHz/P4/512MB   82644 assignments / second
>  Linux     450 MHz/P2/256MB    111111 assignments / second
>
>As you can see the puny 450 MHz P2 managed to kick Windows ass.  I'm
>at a total loss to explain why, though.  For instance,
>
>* Both systems were idle, with no memory hogging apps running (the
>  Windows machine was freshly booted)
>
>* The test runs the makefile only 10 times, for only 10 process
>  create/loads and 100000 assignments
>
>* The test seems to be entirely cpu bound, with both executables
>  compiled by gcc (albeit different versions)
>
>Could this be entirely explained by the difference in process creation
>times?  I would have thought that the 4 times clock rate and beefier
>ram would have adequately compensated.
>
>Note, this is not - in any way - a complaint about performance.  I'd
>just like to understand the reasons.


Some of it, at least, is likely the overhead of Cygwin.  You might want to 
try a completely native version for Windows if you're looking for a more
direct comparison.



--
Larry Hall                              http://www.rfk.com
RFK Partners, Inc.                      (508) 893-9779 - RFK Office
838 Washington Street                   (508) 893-9889 - FAX
Holliston, MA 01746                     


--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Problem reports:       http://cygwin.com/problems.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/