From:	antonios DOT proios AT wmich DOT edu
Date:	Tue, 30 Nov 1999 10:55:39 -0500 (EST)
Subject:	Re: pgcc died when ...
In-reply-to:	<19991126022911.A4036@cerebro.laendle>
To:	Marc Lehmann <pcg AT opengroup DOT org>
Cc:	antonios DOT proios AT wmich DOT edu, pgcc AT delorie DOT com, antonios DOT proios AT wmich DOT edu
Message-id:	<Pine.PMDF.3.96.991130105019.557872038D-100000@WMICH.EDU>
MIME-version:	1.0
Reply-To:	pgcc AT delorie DOT com
X-Mailing-List:	pgcc AT delorie DOT com
X-Unsubscribes-To:	listserv AT delorie DOT com

> On Thu, Nov 25, 1999 at 06:18:12PM -0500, antonios DOT proios AT wmich DOT edu wrote:
> > My reply address: 99PROIOS AT WMICH DOT EDU
> 
> if that is your reply address then why don't you use it? ;)

I did, i just post it in the message in order to convenience those who are
cut and paste freaks ;)


> > Reading specs from /usr/lib/gcc/i586-pc-linux-gnu/pgcc-2.91.66/specs
> > gcc version pgcc-2.91.66 19990314 (egcs-1.1.2 release)
> 
> This is a very old version. Chances are good that this has been fixed in
> cvs (or in 2.95.2)

hmmmm. this is the one i got right off the pgcc web site.. I will look for
the one you mention: 2.95.2 
I am alittle confused: should i look for pgcc or egcs-1.1.x ?
I heard the two were merged, but i am not certain on this..


> > I wanted to try the newer one because I have been having crashes with any
> > kernel compiled with -mpentiumpro -march=pentiumpro  when trying to run
> 
> Apart from testing compatibility it is not a good idea to compile the
> kernel with pgcc (or at leats not with high optimization levels). It will
> only grow in size.

Yes, i understand this, except I am a speed freak... (and yes i know that
larger size will inversely affect locality and therefore cache performance
in certain situations, but i DID notice speeds ups in my programs due to
faster kernel system call executions (my supposition), and that's why i
wanted to do this..).. Maybe i am mistaken but since I got Celeron CPU's
if i enable the 686 instructions, will pgcc use the MMX and FPU registers
for function calls? (instead of stack)  This should in many cases speed up
things (maybe not that mouch, but by some)..

Thanks for the info

- Raw text -