From owner-freebsd-current  Sun Oct 13 17:09:16 1996
Return-Path: owner-current
Received: (from root@localhost)
          by freefall.freebsd.org (8.7.5/8.7.3) id RAA12761
          for current-outgoing; Sun, 13 Oct 1996 17:09:16 -0700 (PDT)
Received: from haywire.DIALix.COM (haywire.DIALix.COM [192.203.228.65])
          by freefall.freebsd.org (8.7.5/8.7.3) with ESMTP id RAA12756
          for <freebsd-current@freebsd.org>; Sun, 13 Oct 1996 17:09:11 -0700 (PDT)
Received: (from news@localhost) by haywire.DIALix.COM (8.7.6/8.7.3) id IAA05660 for freebsd-current@freebsd.org; Mon, 14 Oct 1996 08:09:01 +0800 (WST)
Received: from GATEWAY by haywire.DIALix.COM with netnews
	for freebsd-current@freebsd.org (problems to: usenet@haywire.dialix.com)
To: freebsd-current@freebsd.org
Date: 14 Oct 1996 00:09:01 GMT
From: peter@spinner.DIALix.COM (Peter Wemm)
Message-ID: <53s0at$358$1@haywire.DIALix.COM>
Organization: DIALix Services, Perth, Australia.
References: <380.845216513@critter.tfs.com>
Subject: Re: make -j#
Sender: owner-current@freebsd.org
X-Loop: FreeBSD.org
Precedence: bulk

In article <380.845216513@critter.tfs.com>,
	phk@critter.tfs.com (Poul-Henning Kamp) writes:
> 
> Now that we have the possibility, I guess I should mention that most
> compiles go a couple of percent faster if you use the parallel
> feature of make(1).
> 
> Try using:
> 	make -j3
> for your kernel makes.

I find that using -pipe is a better overall option for a single-cpu compile.
When I set in /etc/make.conf:
CFLAGS= -O -pipe
COPTFLAGS= -O -pipe
..  I find that a simple 'make' is faster than any setting of -j# ..

This is probably more dependent on the machine in question than anything,
including things like cache speed, ram speed, disk speed and architecture
(ie: scsi vs. ide etc).

I found that the job-complete polling was too chunky though, It seemed that
make was taking up to half a second to start a new job after the last one
had finished.  Dropping one of the select timeout parameters from 500000
usec to 50000 usec made a lot of difference, but it still wasn't quite as
quick as simply -pipe.

However, on the smp kernel it certainly is nice! :-)

Cheers,
-Peter