From owner-freebsd-smp Tue Sep 28 19:44:29 1999 Delivered-To: freebsd-smp@freebsd.org Received: from pop3-3.enteract.com (pop3-3.enteract.com [207.229.143.32]) by hub.freebsd.org (Postfix) with SMTP id 85DE014E8D for ; Tue, 28 Sep 1999 19:44:26 -0700 (PDT) (envelope-from dscheidt@enteract.com) Received: (qmail 94506 invoked from network); 29 Sep 1999 02:44:25 -0000 Received: from shell-1.enteract.com (dscheidt@207.229.143.40) by pop3-3.enteract.com with SMTP; 29 Sep 1999 02:44:25 -0000 Date: Tue, 28 Sep 1999 21:44:24 -0500 (CDT) From: David Scheidt To: Daniel O'Connor Cc: freebsd-smp@freebsd.org, "Adam D. Marks" Subject: Re: make question In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-smp@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org On Wed, 29 Sep 1999, Daniel O'Connor wrote: > > > depend on what the limit on performance is. In my machine, the limit is > > almost always disk performance. > > Well I did some benchmarks of doing make buildworld for -current on a -current > box. > > I went from 1 to 20 in steps of 2. From memory the best resulsts where about -j > 12, but that ate a LOT of memory :) > > If you actually want to use your computer while doing a compile then -j 4 is > probably OK. > > The system I did it on was a dual PII-350 with an IDE disk and 128 meg of RAM. I have a dual PII-400, with 256MB. One Ultra2 SCSI disk, and two IDE disks. I get about ~55 minute buildworlds, with idle CPU most of the time, and ~100 disk transactions per disk involved. Memory isn't a factor, either. David To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-smp" in the body of the message