From owner-freebsd-hackers Thu May 16 16:48:00 1996 Return-Path: owner-hackers Received: (from root@localhost) by freefall.freebsd.org (8.7.3/8.7.3) id QAA17089 for hackers-outgoing; Thu, 16 May 1996 16:48:00 -0700 (PDT) Received: from schizo.cdsnet.net (schizo.cdsnet.net [204.118.244.32]) by freefall.freebsd.org (8.7.3/8.7.3) with ESMTP id QAA17084 for ; Thu, 16 May 1996 16:47:58 -0700 (PDT) Received: from localhost (mrcpu@localhost) by schizo.cdsnet.net (8.7.5/8.6.12) with SMTP id QAA02224 for ; Thu, 16 May 1996 16:48:38 -0700 (PDT) X-Authentication-Warning: schizo.cdsnet.net: mrcpu owned process doing -bs Date: Thu, 16 May 1996 16:48:38 -0700 (PDT) From: Jaye Mathisen To: hackers@freebsd.org Subject: A MMAP observation Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-hackers@freebsd.org X-Loop: FreeBSD.org Precedence: bulk I've been using MSQL a bit to play around with some stuff, and w/o question, *disabling* mmap() in MSQL lets the application run much faster than *with* mmap. The DB is about 20MB's, and with MMAP (on a 64MB box), it pages a fair amount, and the process gets way up there. W/o mmap, running the same queries and such, the CPU time is way up there, but hardly any paging activity. It also seems to reach a steady state much quicker, where the Wired/Cache/Buf/Free numbers hardly change. I guess w/o more empirical evidence, what it says to me is that at first run, the FreeBSD VM system is doing a darn good job, especially caching. The same app run on BSD/OS just kills it. I don't know why, but there are a few machine differences that may be coming in to play as well. Of course, if somebody would make ODBC clients for FreeBSD so I could let it talk to my NT SQL server, then I wouldn't have to mess with this at all. :)