From owner-freebsd-questions Tue Aug 11 16:22:24 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id QAA26191 for freebsd-questions-outgoing; Tue, 11 Aug 1998 16:22:24 -0700 (PDT) (envelope-from owner-freebsd-questions@FreeBSD.ORG) Received: from littlebear.cs.du.edu (littlebear.cs.du.edu [130.253.192.1]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id QAA26178 for ; Tue, 11 Aug 1998 16:22:22 -0700 (PDT) (envelope-from leut@cs.du.edu) Received: from chocolate (chocolate [130.253.192.91]) by littlebear.cs.du.edu (8.8.8/8.8.8) with SMTP id RAA27683 for ; Tue, 11 Aug 1998 17:21:57 -0600 (MDT) Date: Tue, 11 Aug 1998 17:21:57 -0600 (MDT) From: Scott Leutenegger X-Sender: leut@chocolate To: freebsd-questions@FreeBSD.ORG Subject: raw disk question Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-questions@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG Hi, Want to help convert a solaris user to FreeBSD? Question: Does FreeBSD support raw disks access? i.e. a read/write to/from disk does NOT access the file cache in memory, but rather 100% of the accesses go-to/come-from disk. If so, how do I create the raw disk partition? Note: I am trying to decide whether to install solarix x86 or FreeBSD on a cluster of PCs to build a parallel database application (parallel multi-dimensional indexing) I know solaris x86 supports this, but have been lead to believe FreeBSD will be faster, easier to install/maintain, and comes loaded with most of the software (gcc/g++/sockets/perl) I need. (please correct me if I am wrong) Further Explanation: (not really relevant, explains why raw disk access desired) Our project has been developed on Sun's running solaris, but due to the cost tradeoff we bought 10 PCs to build a cluster. We do our own buffer pool (memory) caching. When a read/write is exectued I want the data to come/go straight from/to disk, ignoring any file system caching. Bye the way, no, I do not think I can implement a better caching policy. Rather, I need this functionality to enable measurements for the number of disk I/O for various memory sizes. The applications do run faster using a file rather than raw I/O (since you get many buffer hits), but I need the flexibility and predictablility of a raw disk. If you are really curious, check out the methodology section of: "The Effect of Buffering on the Performance of R-Trees" (postscript only) at www.cs.du.edu/users/leut/pubs.html Thanks, ---Scott Professor Scott T. Leutenegger Phone: 303-871-2821 Math and CS Dept, Univ of Denver FAX: 303-871-3010 2360 S. Gaylord Street email: leut@cs.du.edu Denver, CO 80208-0189 http://www.cs.du.edu/users/leut Professor Scott T. Leutenegger Phone: 303-871-2821 Math and CS Dept, Univ of Denver FAX: 303-871-3010 2360 S. Gaylord Street email: leut@cs.du.edu Denver, CO 80208-0189 http://www.cs.du.edu/users/leut To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-questions" in the body of the message