Sender:	root AT squid DOT netplus DOT net
Message-ID:	<37C39977.647AD260@netplus.net>
Date:	Wed, 25 Aug 1999 02:21:27 -0500
From:	Steve Bergman <steve AT netplus DOT net>
X-Mailer:	Mozilla 4.51 [en] (X11; I; Linux 2.2.5-15 i586)
X-Accept-Language:	en
MIME-Version:	1.0
To:	pgcc AT delorie DOT com
Subject:	Re: Is "pgcc -O2 -m486" equivalent to "egcs -O2 -m486" in compile speed?
References:	<37C1C15D DOT BF6C1A AT netplus DOT net> <19990824214057 DOT D5486 AT cerebro DOT laendle>
Reply-To:	pgcc AT delorie DOT com

Marc Lehmann wrote:
> 
> On Mon, Aug 23, 1999 at 04:47:09PM -0500, Steve Bergman <steve AT netplus DOT net> wrote:

 
> No, but pgcc -O2 should be _very_ close in compile speed to egcs -O2,
> likewise -O. Without optimization they both should be equally and very
> fast.
> 

Hi,

Thanks for the reply.

I did grab egcs 2.91.66 (The redhat build) and compared against pgcc
2.91.66 (The mandrake build) with "make modules" for my 2.2.11 kernel.  

I compared egcs with standard options (-O2 -m486, etc) to pgcc with the
same options and pgcc with the mandrake options: -O2 -mcpu=pentium
-march=pentium -ffast-math -fno-rtti -fno-exceptions -fexpensive
optimizations.

I was surprised and pleased to find that pgcc does quite well in all
cases.  It is only about 4% slower with the vanilla options.  And only
7% slower (than egcs and vanilla options) when using the mandrake
options.  i.e. fairly trivial.  I thought that there would be more of a
difference.  I'm also glad to hear that -O2 is pretty conservative. 
Although I am excited about pgcc, (It's irritating to see MSVC
performing so well ;-) I have always felt that when something goes
wrong, and even if it is not pgcc's fault, that it's another possibility
to be ruled out.  To be honest, I have never tracked a problem down to
pgcc.  There have been a couple of things that went "unresolved" but
everything I have been able to track down has been caused something else
entirely.  


-Thanks,
-Steve Bergman

- Raw text -