Date: Sat, 7 Jul 2001 16:36:45 -0500 From: Alfred Perlstein <bright@sneakerz.org> To: Jonah Kuo <jonahk@ms39.hinet.net> Cc: freebsd-smp@freebsd.org Subject: Re: mp is slower than sp kernel. Message-ID: <20010707163645.B88962@sneakerz.org> In-Reply-To: <006301c10728$f9707e70$2e53a8c0@profe>; from jonahk@ms39.hinet.net on Sun, Jul 08, 2001 at 05:08:30AM %2B0800 References: <006301c10728$f9707e70$2e53a8c0@profe>
next in thread | previous in thread | raw e-mail | index | archive | help
* Jonah Kuo <jonahk@ms39.hinet.net> [010707 16:09] wrote: > Hello, > > If this has been answered, I'm sorry! > > This is what I observed and I'm really not sure how would this happen, > but I'm sure my mp kernel is faster several months ago. So asking you > professionals would be my first try, thanks! > > The following is my tail'ed output of make world for sp and mp, respectively: ... > > This two 'make world' are using same copy of anything: > > * smp box. (pentium200-mmx * 2) asus mb. It's only a test box, nobody > uses it except me. > * /etc/make.conf > * kernel config > * source, which I cvsup'ed at Jul 7 16:06 > * I always do a 'chflags -R noschg and rm -rf /usr/obj/*' before > making world. How do expect to gain more performance when running "make world" unless you tell make to run more than one job in parallel? Try doing "make -jN world" where N is some number from 4 to 12 or so, you should see the improvement. As far as the reasoning for the slowdown for non-parallel compiles, the reason for that is that managing two processors is more expensive than the single processor model, so there's some overhead to take care of. -Alfred To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-smp" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20010707163645.B88962>