From owner-freebsd-questions@FreeBSD.ORG Fri Jun 11 21:13:20 2010 Return-Path: Delivered-To: freebsd-questions@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34]) by hub.freebsd.org (Postfix) with ESMTP id E02B31065678 for ; Fri, 11 Jun 2010 21:13:20 +0000 (UTC) (envelope-from talon@lpthe.jussieu.fr) Received: from shiva.jussieu.fr (shiva.jussieu.fr [134.157.0.129]) by mx1.freebsd.org (Postfix) with ESMTP id 75D5D8FC14 for ; Fri, 11 Jun 2010 21:13:20 +0000 (UTC) Received: from parthe.lpthe.jussieu.fr (parthe.lpthe.jussieu.fr [134.157.10.1]) by shiva.jussieu.fr (8.14.4/jtpda-5.4) with ESMTP id o5BLDHkC081500 ; Fri, 11 Jun 2010 23:13:18 +0200 (CEST) X-Ids: 168 Received: from niobe.lpthe.jussieu.fr (niobe.lpthe.jussieu.fr [134.157.10.41]) by parthe.lpthe.jussieu.fr (Postfix) with ESMTP id E12678A073; Fri, 11 Jun 2010 23:13:16 +0200 (CEST) Received: by niobe.lpthe.jussieu.fr (Postfix, from userid 2005) id CEFED2C; Fri, 11 Jun 2010 23:13:16 +0200 (CEST) Date: Fri, 11 Jun 2010 23:13:16 +0200 From: Michel Talon To: freebsd-questions@freebsd.org, Martin Cracauer Message-ID: <20100611211316.GA521@lpthe.jussieu.fr> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline User-Agent: Mutt/1.4.2.3i X-Miltered: at jchkmail2.jussieu.fr with ID 4C12A6EE.000 by Joe's j-chkmail (http : // j-chkmail dot ensmp dot fr)! X-j-chkmail-Enveloppe: 4C12A6EE.000/134.157.10.1/parthe.lpthe.jussieu.fr/parthe.lpthe.jussieu.fr/ Cc: Subject: Re: lang/cmucl broken on amd64? X-BeenThere: freebsd-questions@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: User questions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 11 Jun 2010 21:13:21 -0000 Martin Cracauer wrote: > Interesting you have lower performance in SBCL. Are you comparing a > 64 bit SBCL with a 32 bit CMUCL? Is your SBCL binary (whichever > bitcount) compiled with thread support? I have a 32 bits machine, and i was using the FreeBSD sbcl port without changing any compiling option. It is sufficient to run a number of maxima examples (*) to see that they run frequently faster with cmucl (gcl was also similarly speedy) than with sbcl (sometimes considerably faster). I think having seen similar assertions in maxima mailing list. (*) for example this computation is appropriate http://maxima.sourceforge.net/docs/manual/en/maxima_54.html#SEC233 batch("grobner.demo") Another thing to consider is that the cmucl compiler is now able to emit sse2 instructions, and this gives a quite substantial gain in numeric computations under maxima, an example being eigens_by_jacobi on a large matrix, which gets a considerable speed boost. -- Michel TALON