From owner-freebsd-hackers  Tue Jul  9 01:03:47 1996
Return-Path: owner-hackers
Received: (from root@localhost)
          by freefall.freebsd.org (8.7.5/8.7.3) id BAA13516
          for hackers-outgoing; Tue, 9 Jul 1996 01:03:47 -0700 (PDT)
Received: from ra.dkuug.dk (ra.dkuug.dk [193.88.44.193])
          by freefall.freebsd.org (8.7.5/8.7.3) with SMTP id BAA13498
          for <hackers@FreeBSD.org>; Tue, 9 Jul 1996 01:03:43 -0700 (PDT)
Received: (from sos@localhost) by ra.dkuug.dk (8.6.12/8.6.12) id KAA29489; Tue, 9 Jul 1996 10:02:13 +0200
Message-Id: <199607090802.KAA29489@ra.dkuug.dk>
Subject: Re: A faster compiler? ( Re: gcc lies? )
To: doug@sun1paztcn.wr.usgs.gov (Doug Wellington)
Date: Tue, 9 Jul 1996 10:02:13 +0200 (MET DST)
Cc: hasty@rah.star-gate.com, hackers@FreeBSD.org, doug@sun1paztcn.wr.usgs.gov
In-Reply-To: <9607090729.AA18149@sun1paztcn.wr.usgs.gov> from "Doug Wellington" at Jul 9, 96 00:29:13 am
From: sos@FreeBSD.org
Reply-to: sos@FreeBSD.org
X-Mailer: ELM [version 2.4 PL24]
Content-Type: text
Sender: owner-hackers@FreeBSD.org
X-Loop: FreeBSD.org
Precedence: bulk

In reply to Doug Wellington who wrote:
> 
> Previously:
> >Well, I sure hope someone does a performance analysis on gcc-1.42 vs gcc-2.xx
> >or a faster compiler comes along .
> 
> I know that there is always the argument for consistent code generation,
> but you might try looking at lcc.  An article in Linux Journal quotes
> times to compile lcc with lcc and with gcc.  Lcc compiled itself in 36
> seconds and gcc compiled lcc in 68 seconds with default optimization and
> 130 seconds for maximum optimization.  Interestingly, the lcc compiled
> program ran in 36 seconds and the gcc program ran in 30.  (So gcc gains
> about 20% speed for the almost 400% compile time...)
> 
> Check out http://www.princeton.edu/software/lcc/

I've played alot with lcc, there is a port to FreeBSD on:

ftp://ra.dkuug.dk/pub/LCC

It does quite a nice job, but it has some serious limitations that
makes it a bit awkward in the FreeBSD world. The main offender
here is that we use long long's which lcc doesn't support. I've
played with adding support for long long's but it is not pretty :(

I've got a lcc-4.0 beta though that is "real" 64bit, when I have
time I'll add a x86 backend to that one, if the authors doesn't
beat me to it....


-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Soren Schmidt             (sos@FreeBSD.org)             FreeBSD Core Team
               So much code to hack -- so little time.