Date: Thu, 20 Nov 1997 22:00:18 +0100 From: Tor Egge <Tor.Egge@idi.ntnu.no> To: sos@FreeBSD.dk Cc: jkh@time.cdrom.com, hasty@rah.star-gate.com, toor@dyson.iquest.net, sprice@hiwaay.net, smp@csn.net, current@FreeBSD.ORG Subject: Re: make world time???/ Message-ID: <199711202100.WAA13101@pat.idi.ntnu.no> In-Reply-To: Your message of "Mon, 17 Nov 1997 08:32:10 %2B0100 (MET)" References: <199711170732.IAA09348@sos.freebsd.dk>
next in thread | previous in thread | raw e-mail | index | archive | help
> Same observation here, I just went dual on my P6 system, and > it hardly got any better in a make world. I notice that it spends > ALOT of its time in system, so our megalock might be the culprit... This is due to how CPU usage accounting is currently handled. 5 make worlds gave the following result with a dual P6 machine: make.world1: 9950.08 real 1818.02 user 7944.37 sys make.world2: 9995.30 real 1857.03 user 7778.47 sys make.world3: 9840.08 real 1736.67 user 7857.49 sys make.world4: 9823.49 real 1726.97 user 7874.01 sys make.world5: 9839.31 real 1731.59 user 7861.62 sys After some changes to statclock and hardclock, I got more reasonable numbers: make.world1: 9941.28 real 5616.14 user 3932.16 sys make.world2: 9914.21 real 5632.65 user 3973.46 sys make.world3: 9779.39 real 5621.38 user 3939.89 sys make.world4: 9770.91 real 5622.68 user 3942.34 sys - Tor Egge
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199711202100.WAA13101>