From mboxrd@z Thu Jan 1 00:00:00 1970 Date: Wed, 22 Nov 2000 15:28:14 +0000 From: "Stephen C. Tweedie" Subject: Re: max memory limits ??? Message-ID: <20001122152814.B7417@redhat.com> References: <3A1BCC05.4080608@SANgate.com> <20001122161104.C28963@mea-ext.zmailer.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20001122161104.C28963@mea-ext.zmailer.org>; from matti.aarnio@zmailer.org on Wed, Nov 22, 2000 at 04:11:04PM +0200 Sender: owner-linux-mm@kvack.org Return-Path: To: Matti Aarnio Cc: BenHanokh Gabriel , linux-mm@kvack.org List-ID: Hi, On Wed, Nov 22, 2000 at 04:11:04PM +0200, Matti Aarnio wrote: > > - does using HIGHMEM results with performance penalty ? > > Of course, it causes extra mapping operations on machines > needing it to support larger memory (intel PAE36 featured > hardware). User processes can access all what is mapped > into them at the same time -- All programs, kernel included > are limited to 32 bit addresses, but kernel can juggle maps > to reach areas not mapped in its address space at some moment. The other performance implications are: HIGHMEM currently results in extra copies if you are on Intel and using a 4G or 64G config, whenever you perform disk IO to memory above 1GB. If you are using the 64G config (PAE36), then the page tables double in size, occupying more of your cache. Those are likely to be the two biggest performance costs in enabling highmem in an i686 kernel. Cheers, Stephen -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@kvack.org. For more info on Linux MM, see: http://www.linux.eu.org/Linux-MM/