Date:	Sun, 19 Mar 2000 11:01:45 +0200 (IST)
From:	Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
X-Sender:	eliz AT is
To:	Javier Bardal Prieto <u0204956 AT golum DOT inforg DOT uniovi DOT es>
cc:	djgpp-workers AT delorie DOT com
Subject:	Re: Fastest integer type
In-Reply-To:	<200003171950.UAA20742@golum.inforg.uniovi.es>
Message-ID:	<Pine.SUN.3.91.1000319110049.13844B@is>
MIME-Version:	1.0
Reply-To:	djgpp-workers AT delorie DOT com
Errors-To:	dj-admin AT delorie DOT com
X-Mailing-List:	djgpp-workers AT delorie DOT com
X-Unsubscribes-To:	listserv AT delorie DOT com

On Fri, 17 Mar 2000, Javier Bardal Prieto wrote:

> The compiler options are:
> 
> gcc -O9 -fomit-frame-pointer -funroll-loops -m386 -mcpu=i386 -march=i386
> -ffast-math bench.c -obench

Doesn't this disable some Pentium-specific optimizations related to
scheduling instructions for simultaneous execution?

What about alignment: do architecture-specific switches affect
code/data/stack alignment?

Also, what version of GCC and Binutils did you use?

> DO NOT CHANGE IT!!!!
> gcc generates highly optimized code with this settings, and changing it
> to a specific processor model only has 2% of speed up. So for a uniform
> testing I recommend to do not change it.

If the effect of changing the optimization options is so minimal, why
do you worry so much about people playing with them?  OTOH, if some of
the options have a significant effect on the speed, perhaps we need to
know about them?

> I have problems posting the benchmark source, bench.c is 17kb and djgpp
> mail-list seems to reject it.

Not because of size.

> begin 644 bench.zip

Ouch, why not plain text?..

- Raw text -