Message-Id: <200001071056.EAA12627@lakdiva.slt.lk>
From: "Kalum Somaratna aka Grendel" <kalum AT newmail DOT net>
To: djgpp AT delorie DOT com
Date: Fri, 7 Jan 2000 04:56:42 +0600
MIME-Version: 1.0
Content-type: text/plain; charset=US-ASCII
Content-transfer-encoding: 7BIT
Subject: Re: Why did ID choose DJGPP for Quake?
In-reply-to: <gc4d4.271$CJ1.9007@dfiatx1-snr1.gtei.net>
X-mailer: Pegasus Mail for Win32 (v3.12)
Reply-To: djgpp AT delorie DOT com
Errors-To: dj-admin AT delorie DOT com
X-Mailing-List: djgpp AT delorie DOT com
X-Unsubscribes-To: listserv AT delorie DOT com
Precedence: bulk

On 6 Jan 00, at 17:23, Damian Yerrick wrote:

> 
> "Eli Zaretskii" <eliz AT is DOT elta DOT co DOT il> wrote:
<snip>
> 
> > The only reason we don't compile with -mpentium is because
> > we still didn't switch to GCC 2.9x for building the library, and
> > versions of GCC before 2.9x didn't support Pentium-specific
> > optimizations.
> 
> Then I guess I'll have to recompile everything from sources myself
> once the next DJGPP version is released, as one of my systems is
> 486SX-based.
> 

I think Eli meant that DJGPP would be compiled using the -m586 
switch wich would not generate pentium specific code. It would be 
still generating 386 instructions specially timed for the pentium (the 
-march=i586 would generate pentium specific insns). So AFAIK the 
code generated with the -m586 switch should run without 
generating invalid opcodes on your 486 but it might be a bit slower 
than compiling with the -m486 option.

But the starange thing is that on my old 486DX2 I found that code 
compiled with pentium speciific optitmisations -m586 is actually 
faster than code from the m486 switch.

Regards,
Grendel