Date:	Thu, 18 Sep 1997 10:36:35 +0200 (MET DST)
From:	Jan Hubicka <hubicka AT ta DOT jcu DOT cz>
To:	Thomas Djafari <tdjafari AT nordnet DOT fr>
cc:	djgpp AT delorie DOT com
Subject:	Re: The numer 1 compiler, DJGPP or MSVC Here's a good rating comparision
In-Reply-To:	<5vomfv$bch$1@news.internetsat.com>
Message-ID:	<Pine.LNX.3.95.970918103015.7937A-100000@tabor.ta.jcu.cz>
MIME-Version:	1.0

> the unrolling loop function of djgpp is much more intelligent that the
> watcom's (which unrolls loops only if there's a very low count). djgpp
> unrolled the loop 10 times
Well, I did some tests with latests egcs(pre gcc3.0) and it has new loop
unrooling code, wich seems to make about 20% speedup in case of many
simple loops :) Maybe it should be interesting to make comparsion between
gcc and egcs(I should do that at my linux)
> 
> I also took all the previous tests and changed everything to floats (with
> (int) casts before each array to run the first one) and this time.... the
> watcom blown away djgpp with no possible comparison  : 6 times faster !
> BUT the watcom has pentium-fpu optimization which is not known to djgpp
Well, I was making tests at FP loop of Mandelbrot Set and got quite a
oposite (wetcom 4 times slower) tip 1: Use -ffast-math GCC switch
tip 2: Use long double instead (Gcc tries to fill iso standards and FP at
intels is incredibly slow in case of floats - long double is fastest.
Watcom usually generates slightly inexact code in case it helps...
-ffast-math usually helps, long double (or double) helps a lot (because of
machine description of Intel lies to GCC about intel hardware, it will
most likely change in egcs soon)

Please send me your code so I will be able to do comparsions..

BTW egcs has pentium/pentiumpro optimizations too. Looking forward for
egcs compiled for djgpp :)
> 
> I'll continue my tests, it seems that djgpp has improved a lot since my
> first test. but the gcc I've with my linux (slackware 2.0.0) SUUUUUUUCKS, it
> generates the worst code I've ever seen,  maybe there's a newer version ?
> (please tell me)
well, generated code changed very slightly from 2.5.8 to 2.6.2 and almost
no changes to 2.7.1 as long as I remember. So you was comparing exactly
same code...I am sorry. Maybe incorrect switches?
> 
> I'll make more complete tests and publish both results and source code, so
> we'll be able to establish comparisons with msvc and bc++ too.
OK
Send me it by email :)

Honza
> 
> 
> 
> 
> 
> 
> 
> 

------------------------------------------------------------------------------
                   Have you browsed my www pages? Look at:
                       http://www.paru.cas.cz/~hubicka
      Koules-the game for Svgalib,X11 and OS/2,  Xonix-the game for X11
      czech documentation for linux index, original 2D computer art and
              funny 100 years old photos and articles are there!

- Raw text -