From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <gcc-return-103462-listarch-gcc=gcc.gnu.org@gcc.gnu.org>
Received: (qmail 19488 invoked by alias); 30 Sep 2004 06:58:29 -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 19454 invoked from network); 30 Sep 2004 06:58:27 -0000
Received: from unknown (HELO mtagate1.de.ibm.com) (195.212.29.150)
  by sourceware.org with SMTP; 30 Sep 2004 06:58:27 -0000
Received: from d12nrmr1607.megacenter.de.ibm.com (d12nrmr1607.megacenter.de.ibm.com [9.149.167.49])
	by mtagate1.de.ibm.com (8.12.10/8.12.10) with ESMTP id i8U6wQfQ150296;
	Thu, 30 Sep 2004 06:58:26 GMT
Received: from d12ml102.megacenter.de.ibm.com (d12av02.megacenter.de.ibm.com [9.149.165.228])
	by d12nrmr1607.megacenter.de.ibm.com (8.12.10/NCO/VER6.6) with ESMTP id i8U6wP08114828;
	Thu, 30 Sep 2004 08:58:25 +0200
In-Reply-To: <OFD6E082D0.AE5C8974-ONC2256F1E.00465975-C2256F1E.00488D05@il.ibm.com>
Subject: Re: SMS scheduling
To: Canqun Yang <canqun@nudt.edu.cn>
Cc: gcc@gcc.gnu.org
Message-ID: <OFC326A99F.D1A154FA-ONC2256F1F.00258145-C2256F1F.00264CCE@il.ibm.com>
From: Mostafa Hagog <MUSTAFA@il.ibm.com>
Date: Thu, 30 Sep 2004 08:50:00 -0000
MIME-Version: 1.0
Content-type: text/plain; charset=US-ASCII
X-SW-Source: 2004-09/txt/msg01649.txt.bz2


BTW, there are several improvement to SMS that
are postponed to post-gcc4.0, these enhancement include:
1. Use of loop-info structure to detect loops.
2. Use of alias sets to remove memory dependencies.
3. Remove register anti-dependencies.
Patches for these improvements are ready.
We are also working on using the register pressure utility
in SMS.

Mostafa.

Mostafa Hagog wrote on 29/09/2004 15:12:26:

>
> Canqun Yang <canqun@nudt.edu.cn> wrote on 28/09/2004 11:06:29:
>
> > Hi, all
> >
> > I tested the Swing Modulo Scheduling of GCC. For
> > simple program, the numerical caculation of PI, it
> > achieves significant speedup on IA64. But, for little
> > bit complex programs, the SMS can hardly work out a
> > successful schedule.
> >
> > The algorithms implemented by Ayal and Mostafa are
> > correct. It seems that the SMS itself is wrong. The
> > schedule priority order calculated by SMS is much
> > different from normal MinDist algorithm.
>
> SMS prioritizes the nodes using critical patch based
> heuristic - the main idea is that instructions that are
> on the critical path are less flexible in means of
> scheduling.
> Why do you think that the priority order should be
> according to MinDist algorithm (what is the minimum
> distance in this case)? Can you provide an example
> that supports this?
>
> > Besides this,
> > SMS is not sensitive to II. Is SMS really wrong?
>
> The node ordering step in SMS is not sensitive to II
> (as mentioned before it is based on critical path
> heuristic).  However, the scheduling step is sensitive
> to II - failing to schedule the nodes of a loop within
> II cycles will lead to try scheduling the nodes again
> within II + 1 cycles.
>
>
> Mostafa.
>
>
>
>
>