From owner-freebsd-ports@FreeBSD.ORG  Fri May 23 20:08:14 2003
Return-Path: <owner-freebsd-ports@FreeBSD.ORG>
Delivered-To: freebsd-ports@freebsd.org
Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125])
	by hub.freebsd.org (Postfix) with ESMTP id D778F37B401
	for <freebsd-ports@FreeBSD.org>; Fri, 23 May 2003 20:08:14 -0700 (PDT)
Received: from web13407.mail.yahoo.com (web13407.mail.yahoo.com
	[216.136.175.65])
	by mx1.FreeBSD.org (Postfix) with SMTP id 54D4643F75
	for <freebsd-ports@FreeBSD.org>;
	Fri, 23 May 2003 20:08:14 -0700 (PDT)
	(envelope-from giffunip@yahoo.com)
Message-ID: <20030524030814.59783.qmail@web13407.mail.yahoo.com>
Received: from [200.91.194.224] by web13407.mail.yahoo.com via HTTP;
	Sat, 24 May 2003 04:08:14 BST
Date: Sat, 24 May 2003 04:08:14 +0100 (BST)
From: "=?iso-8859-1?q?Pedro=20F.=20Giffuni?=" <giffunip@yahoo.com>
To: Nakata Maho <chat95@mbox.kyoto-inet.or.jp>
In-Reply-To: <20030524.093848.115902776.chat95@mbox.kyoto-inet.or.jp>
MIME-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Transfer-Encoding: 8bit
cc: freebsd-ports@FreeBSD.org
Subject: Re: Serious bug in math/superlu port (with fix)
X-BeenThere: freebsd-ports@freebsd.org
X-Mailman-Version: 2.1.1
Precedence: list
List-Id: Porting software to FreeBSD <freebsd-ports.freebsd.org>
List-Unsubscribe: <http://lists.freebsd.org/mailman/listinfo/freebsd-ports>,
	<mailto:freebsd-ports-request@freebsd.org?subject=unsubscribe>
List-Archive: <http://lists.freebsd.org/pipermail/freebsd-ports>
List-Post: <mailto:freebsd-ports@freebsd.org>
List-Help: <mailto:freebsd-ports-request@freebsd.org?subject=help>
List-Subscribe: <http://lists.freebsd.org/mailman/listinfo/freebsd-ports>,
	<mailto:freebsd-ports-request@freebsd.org?subject=subscribe>
X-List-Received-Date: Sat, 24 May 2003 03:08:15 -0000

 --- Nakata Maho <chat95@mbox.kyoto-inet.or.jp> wrote: > Hi,
> 
...
> my /etc/make.conf is just like follwings(partially, of
> course):
> 
> CXXFLAGS+= -pipe -O2 -mcpu=pentiumpro -march=pentiumpro
> CFLAGS+=   -pipe -O2 -mcpu=pentiumpro -march=pentiumpro
> FFLAGS+= -pipe -O4 -mcpu=pentiumpro -march=pentiumpro
> -fstrength-reduce -funroll-loops -fexpensive-optimizations
> -ffast-math -malign-double
> 
OK, FFLAGS is not documented in the example make.conf, dumb me. 

> Anyway, it should be much better to use math/atlas instead of
> using BLAS,
> which is the fastest blas implimentation, and lapack is also
> paritally
> implimented. You want to use atlas on sparc64 platform? yes,
> it doesn't compile, reported by kris...
> 

It's faster but isn't size important too?? In the case of the
superlu patch I changed BLAS from BUILD_DEPENDS to RUN_DEPENDS
precisely to hint out that even when you need to link it with
BLAS, you can use any BLAS. 

Well... I don't know what to think of BLAS.. it's an important
port but there are not many ports using it so there is no policy
unto which should be the "official" BLAS. FWIW, there's also an
optimized version of BLAS for pentiums somewhere. (I recall I
posted the URL some years ago, but nobody ported it.)

cheers,

    Pedro.

__________________________________________________
It's Samaritans' Week. Help Samaritans help others. 
Call 08709 000032 to give or donate online now at http://www.samaritans.org/support/donations.shtm