From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <gcc-return-101715-listarch-gcc=gcc.gnu.org@gcc.gnu.org>
Received: (qmail 25046 invoked by alias); 31 Aug 2004 10:44:45 -0000
Mailing-List: contact gcc-help@gcc.gnu.org; run by ezmlm
Precedence: bulk
List-Archive: <http://gcc.gnu.org/ml/gcc/>
List-Post: <mailto:gcc@gcc.gnu.org>
List-Help: <http://gcc.gnu.org/ml/>
Sender: gcc-owner@gcc.gnu.org
Received: (qmail 24869 invoked from network); 31 Aug 2004 10:44:43 -0000
Received: from unknown (HELO Cantor.suse.de) (195.135.220.2)
  by sourceware.org with SMTP; 31 Aug 2004 10:44:43 -0000
Received: from extimap.suse.de (extimap.suse.de [195.135.220.6])
	(using TLSv1 with cipher EDH-RSA-DES-CBC3-SHA (168/168 bits))
	(No client certificate requested)
	by Cantor.suse.de (Postfix) with ESMTP id B0AA1B48E1E;
	Tue, 31 Aug 2004 12:43:09 +0200 (CEST)
Received: from stevenb.home.suse.de (70-90.ipact.nl [82.210.90.70])
	(using TLSv1 with cipher RC4-MD5 (128/128 bits))
	(Client did not present a certificate)
	by extimap.suse.de (Postfix) with ESMTP
	id 44DD08C029; Tue, 31 Aug 2004 12:43:09 +0200 (CEST)
From: Steven Bosscher <stevenb@suse.de>
To: Karel Gardas <kgardas@objectsecurity.com>,
	Paolo Bonzini <bonzini@gnu.org>
Subject: Re: Compilation performance comparison of gcc3.4.1 and gcc3.5.0 2004-08-30 on MICO sources
Date: Tue, 31 Aug 2004 10:55:00 -0000
User-Agent: KMail/1.5.4
Cc: GCC Mailing List <gcc@gcc.gnu.org>
References: <Pine.LNX.4.43.0408311227330.1006-100000@thinkpad.gardas.net>
In-Reply-To: <Pine.LNX.4.43.0408311227330.1006-100000@thinkpad.gardas.net>
Organization: SUSE Labs
MIME-Version: 1.0
Content-Type: text/plain;
  charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
Content-Disposition: inline
Message-Id: <200408311243.11380.stevenb@suse.de>
X-SW-Source: 2004-08/txt/msg01605.txt.bz2

On Tuesday 31 August 2004 12:28, Karel Gardas wrote:
> On Tue, 31 Aug 2004, Paolo Bonzini wrote:
> > >>>1) typecode.cc: 40% regression on O1 while 7% speedup on O2
> > >>
> > >>Can you show us the time report for the 40% regression?
> >
> > Also for 3.4.1?
>
> Sure!

Hmm...  No obvious hot spots eh?

Looks like the tree optimizers are to blame.  We spend roughly the same
amount of time in the post-GIMPLE passes, and we spend >7.5s in the tree
optimizers.  The total slowdown you measured was ~8.9s.  The other 1.4s
are spent in expand as shown in the previous message:

3.4.1:  expand                :   0.79 ( 6%) usr   0.03 ( 1%) sys   0.78 ( 5%) wall
3.5.0:  expand                :   2.08 ( 9%) usr   0.07 ( 4%) sys   2.51 (10%) wall

Hmm, we should probably disable at least flag_thread_jumps and
flag_loop_optimize at -O1, and perhaps consider disabling some
of the more expensive (parts of the) tree optimizers...   And
of course see if it makes sense to disable a few RTL optimizers.

So, looks like a tuning problem to me, not really a slowdown that
indicates something algorithmic being really wrong.

Gr.
Steven