Date: Wed, 26 Aug 2015 23:41:25 +0300 From: Slawa Olhovchenkov <slw@zxy.spb.ru> To: Stanislav Sedov <stas@freebsd.org> Cc: freebsd-hackers@FreeBSD.org Subject: Re: How to application sample profiling? Message-ID: <20150826204125.GK21849@zxy.spb.ru> In-Reply-To: <9BFE7A4D-CCA3-4B44-9E2D-F6F7DAB78F07@freebsd.org> References: <20150823221312.GD21849@zxy.spb.ru> <9BFE7A4D-CCA3-4B44-9E2D-F6F7DAB78F07@freebsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Wed, Aug 26, 2015 at 01:25:25PM -0700, Stanislav Sedov wrote: > > > On Aug 23, 2015, at 3:13 PM, Slawa Olhovchenkov <slw@zxy.spb.ru> wrote: > > > > I have multi-thread application fully utilise all CPU cores. > > I am need to determine lines give mostly CPU loads. > > How I can do this? > > > > `pmcstat -P BU_CPU_CLK_UNHALTED -t application_pid -n 500000 -O sample.out` > > consume to many CPU (performance drop x100 times) and mostly show > > himself in kernel and pmc locks and none for application. > > pmcstat -S BU_CPU_CLK_UNHALTED -- same here. > > > > Did you try something like `pmcstat -S instructions -T` by any chance? > That should give some quick clues as to which functions are being hot > at least. I am try '-S instructions' w/o -T, no difference. > Which FreeBSD version are you using by the way? 10-STABLE
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20150826204125.GK21849>