Date: Mon, 2 Apr 2001 10:07:42 +0200 (IST)
From: Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
X-Sender: eliz AT is
To: "Orlando P. Hevia" <heviaop AT ssdfe DOT com DOT ar>
cc: Andris Pavenis <pavenis AT lanet DOT lv>, djgpp AT delorie DOT com
Subject: Re: g772953 and djdev203
In-Reply-To: <Pine.A41.4.05.10104020814040.61050-100000@ieva06.lanet.lv>
Message-ID: <Pine.SUN.3.91.1010402100047.26715I-100000@is>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Reply-To: djgpp AT delorie DOT com
Errors-To: nobody AT delorie DOT com
X-Mailing-List: djgpp AT delorie DOT com
X-Unsubscribes-To: listserv AT delorie DOT com
Precedence: bulk


On Mon, 2 Apr 2001, Andris Pavenis wrote:

> At first it's much more usefull to send it to DJGPP mailing list than to
> me only.

Indeed.

> try profiling Your application (compile and link Your application
> using command line optioon -pg, after running Your application use gprof)
> and compare results You're getting with djdev202 and djdev203.

Unforunately, profiling doesn't work in djdev202, so the only way to do 
that is to replace the itimer.c module in the v2.02 library with the 
version from djdev203.zip.

> > When I updated to djdev203, the speed decreased in near 
> > 15%. It is a surprising result for me.

Please describe what does the program do, and what library functions does 
it call.

> > Of course, I will continue using djdev202.

You might wish to reconsider: djdev202.zip had a few nasty bugs in it.  
One is with the profiling, mentioned above; but there are others.  Is 
speed so much important to you that you are willing to sacrifice 
stability and reliability?

I'm actually surprised you had such a significant performance hit, since 
djdev203.zip is a bug-fix release: there are no significant changes in 
the code wrt djdev202.zip.  So I suspect that your program somehow 
depends on small changes in code/data location in memory; if that's true, 
you can sustain such a performance hit by simple code modifications, even 
if you stick to djdev202.