From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S263953AbTLOVE4 (ORCPT ); Mon, 15 Dec 2003 16:04:56 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S263969AbTLOVE4 (ORCPT ); Mon, 15 Dec 2003 16:04:56 -0500 Received: from pop.gmx.net ([213.165.64.20]:33941 "HELO mail.gmx.net") by vger.kernel.org with SMTP id S263953AbTLOVEz (ORCPT ); Mon, 15 Dec 2003 16:04:55 -0500 X-Authenticated: #4512188 Message-ID: <3FDE21EF.6070608@gmx.de> Date: Mon, 15 Dec 2003 22:04:47 +0100 From: "Prakash K. Cheemplavam" User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.6b) Gecko/20031208 Thunderbird/0.4 X-Accept-Language: en-us, en MIME-Version: 1.0 To: linux-kernel@vger.kernel.org Subject: [OT] Exists? kernel/library simulating cluster/multi-processor machine Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 8bit Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Hi, for my diploma I am currently working on a library called "load balancing class" originally developed by Frank Meisgen. As the name suggests its duty is to balance the load on a heterogenous cluster. The original library was written for sun sparc machines and I am proting it and enhancing it. Though I have started doing it for Windows2000, it also runs on Linux (which is the goal, but as Windows seems to make more troubles, I develop it there). A age simulation /population simulation already works and now I am busy debugging a fast sat solver (originally written by Max Böhm for his "plb" and converted to lbc by Frank Meisgen). My question is, where there existes some sort of modified kernel or library (perhaps a modified mpi libraray, as lbc uses mpi to communicate) which simulates a cluster or multi-processor machines on a single cpu machine (or a "bigger" cluster on a cluster). Why? Of course it won't do anything good performance wise, But I want to see how well the lbc works and getting access to big clusters isn't easy esp if you want to tweak. Starting more than one process on one machine won't help, as the processes with natrurally balance each other. Thanks, Prakash