Date: Sat, 22 Feb 2014 15:36:19 -0500 From: Lowell Gilbert <freebsd-questions-local@be-well.ilk.org> To: Matthias Apitz <guru@unixarea.de> Cc: Kostas Oikonomou <k.oikonomou@att.net>, freebsd-questions@freebsd.org Subject: Re: slowdown because of gettimeofday change between 9.2 and 10? Message-ID: <444n3q6g1o.fsf@lowell-desk.lan> In-Reply-To: <20140222181449.GA2475@La-Habana> (Matthias Apitz's message of "Sat, 22 Feb 2014 19:14:49 %2B0100") References: <5307E64C.3040108@att.net> <5308D37B.50908@att.net> <20140222171602.GA2265@La-Habana> <20140222181449.GA2475@La-Habana>
next in thread | previous in thread | raw e-mail | index | archive | help
Matthias Apitz <guru@unixarea.de> writes: > El d=EDa Saturday, February 22, 2014 a las 06:16:03PM +0100, Matthias Api= tz escribi=F3: > >> > 2) I checked on another system which runs the same application, but >> > under 9.2. It turns out that the application issues the same calls to >> > gettimeofday, also on a continuous basis, and, furthermore the numbers >> > of calls on the two systems over a period of 15 seconds are almost the= same. >> >=20 >> > So there must be some other difference between 9.2 and 10.0 that causes >> > the slowdown. >>=20 >> In my case it is: >>=20 >> $ uname -a >> FreeBSD La-Habana 10.0-CURRENT FreeBSD 10.0-CURRENT #0 r250588M: Sat >> Jun 22 14:15:48 CEST 2013 >> guru@La-Habana:/usr/obj/usr/home/guru/head/src/sys/GENERIC i386 >>=20 >> The app in question (kabc) was compiled based on the ports at this time. > > $ truss -f -o tr kaddressbook > kdecore (KAction): WARNING: KActionCollection::KActionCollection( > QObject *parent, const char *name, KInstance *instance ) > $ fgrep gettimeofday tr | wc -l > 2262 FreeBSD has *always* had a slower (but, it looks like, more accurate) gettimeofday() implementation than some other systems. This is an infamous slowdown for Java.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?444n3q6g1o.fsf>