From: Bjorn De Meyer <bgdmeyer AT eduserv2 DOT rug DOT ac DOT be>
Newsgroups: comp.os.msdos.djgpp
Subject: Re: large variation in run-time
Date: Mon, 4 May 1998 17:14:08 +0200
Organization: University of Ghent, Belgium
Lines: 82
Message-ID: <Pine.GSO.3.96.980504162711.23787A-100000@eduserv2.rug.ac.be>
References: <354DAFD3 DOT 8D277F06 AT CPRO DOT DLO DOT NL>
NNTP-Posting-Host: eduserv2.rug.ac.be
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: <354DAFD3.8D277F06@CPRO.DLO.NL>
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp
Precedence: bulk

On Mon, 4 May 1998, Johan van Ooijen wrote:

> Hi,
> I am performing Monte Carlo simulation studies that require millions of
> runs. In order to predict how long each session is going to take I use
> trial sessions of 1000 runs. I find that it not possible to predict the
> duration of 10^6 runs. 
Well, statistically speaking, a sample of only 1000 out of a population of
many millions is quite small. I suggest you run 20 tests of 10000,
each with different parameters (if that doesn't take too long)


> For instance, two sessions I predict to take 34
> and 68 hours, respectively (one should be exactly twice as long as the
> other), and they took 10 and 43 hours. It is fine that the sessions take
> shorter, but it is strange that the second is not even close to taking
> twice as long as the other. So this leaves me with the uncertain feeling
> of the correctness of the results, although they appear to be what I
> expect.
> 
Well, it all depends on the parameters. If your alogrythm is correct, then
praise yourself happy you have DJGPP ^_^!. Of course, it's hard for me to
judge without the source code...


> So the question is, whether there is an explanation for these large
> variations in run-time when using go32 ?
> 

I'd rather go for a variation in your parameters wich you did not
anticipate in your samples. OR, your CPU might be warming up due to
these intensive calculations, and thus become somewhat irrattic.


> The software is pure ANSI C, no warnings are given by the compiler (gcc)
> whatsoever. The program uses little memory, it does millions of
> malloc()'s and free()'s, but total used memory doesn't grow (so that is
> OK) and there is no swapping activity.

Well, in protected mode you always should look twice for memory leaks as
the functions wich you can use to check the available memory *might* not
be returning the true amount of free memory.

> I run under MSDOS 6.22, with 
>    device=himem.sys
>    device=emm386.exe ram x=d800-dfff
>    dos=high,umb
> and go32 says:
>    go32 version 1.12.maint3
>    ...
>    VCPI (Expanded) memory available: 29312 Kb
>    Swap space available: 131040 Kb
> 
> Any help is appreciated. Please also mail me directly:
>    j DOT w DOT vanooijen AT CPRO DOT DLO DOT NL
> 
> Thanks,
> Johan van Ooijen

Hope this helps a little.
Cya! En groeten uit Belgie!

> 
> 

Bjorn De Meyer.
IA #i^4
Homepage: http//studwww.rug.ac.be/~bgdmeyer
E-mail: bjorn DOT demeyer AT rug DOT ac DOT be 
Chemistry student at RUG, option polymers.
Coordinator of the "Tales of Artah: First Sword" CRPG project.         
  
 "I've shaven. I'm happy. I have no religion. Don't pray for me." 
                                                Illusion of Gaia 2