From owner-freebsd-stable@FreeBSD.ORG Sun Sep 14 22:29:13 2003 Return-Path: Delivered-To: freebsd-stable@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 45AC216A4BF for ; Sun, 14 Sep 2003 22:29:13 -0700 (PDT) Received: from carver.gumbysoft.com (carver.gumbysoft.com [66.220.23.50]) by mx1.FreeBSD.org (Postfix) with ESMTP id A8B4443F75 for ; Sun, 14 Sep 2003 22:29:12 -0700 (PDT) (envelope-from dwhite@gumbysoft.com) Received: by carver.gumbysoft.com (Postfix, from userid 1000) id 97A9F72DA3; Sun, 14 Sep 2003 22:29:12 -0700 (PDT) Received: from localhost (localhost [127.0.0.1]) by carver.gumbysoft.com (Postfix) with ESMTP id 922CE72DA2; Sun, 14 Sep 2003 22:29:12 -0700 (PDT) Date: Sun, 14 Sep 2003 22:29:12 -0700 (PDT) From: Doug White To: Irvine Short In-Reply-To: <20030915070012.U36360@fling.sanbi.ac.za> Message-ID: <20030914221953.H97248@carver.gumbysoft.com> References: <20030913092804.S46465@fling.sanbi.ac.za> <20030913123257.C51554@fling.sanbi.ac.za> <20030915070012.U36360@fling.sanbi.ac.za> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII cc: freebsd-stable@freebsd.org Subject: Re: Large memory issues on 4-STABLE X-BeenThere: freebsd-stable@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Production branch of FreeBSD source code List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Mon, 15 Sep 2003 05:29:13 -0000 On Mon, 15 Sep 2003, Irvine Short wrote: > Yes, I understand that. What I am saying is, is that the general story out > there is that with a 32 bit operatng system no process can address more > than 4GB of RAM. Fine. David said that in FreeBSD I cannot in practicality > address more than about 3GB of RAM. Also fine. > > However we have a situation where if I set MAXDSIZ to 2048 or above then > things break, so FreeBSD right now has an effectivce limit of 2GB per > process. > > Is this to be considered a bug or a feature? I'd have to say feature. The kernel placement and user/kernel boundary sizing is part of FreeBSD and the 4GB limit for both kernel and user is a restriction of the processor architecture. Its possible to do some tuning and crunching and maybe eek out 3GB user addressible space, but thats all you're going to get. > Then, with the PAE support, lets say I have 8GB of RAM. Will those same > design limitations restrict me to 3GB of RAM, or 2GB, or will I be able to > have a single process addressing 4GB all by itself? Since you need to have the kernel inside that 4GB space somewhere, and 4GB of RAM chews up a ton of memory in page tables, I would say no. PAE is not 64 bit addressing. PAE is 16 banks of 32 bit address spaces. ("36 bit" == 4 bits of bank and 32 bits of address) As David said, on the 8GB system you could have two large processes taking up 2.5GB each, but not one 5GB process. > This is relevant to the work we're doing - some of my users actually > really do need this amount of memory. If this is the case then you should consider a 64 bit architecture, like ia64, sparc64, or amd64 (Opteron). On these systems you can allocate as much memory as you can cram into the system*; 64 bits goes a long way. :) -- Doug White | FreeBSD: The Power to Serve dwhite@gumbysoft.com | www.FreeBSD.org * There's probably some limits when you start using fractional terabytes to represent the amount of memory you're using, but such limits would be surmountable given time and equipment.