Date: Sat, 1 May 2004 12:25:58 -0400 From: Alexandre "Sunny" Kovalenko <Alex.Kovalenko@verizon.net> To: freebsd-current@freebsd.org Subject: Re: Is it possible to make -lpthread program to use 100% CPU? Message-ID: <20040501122558.6139dec1.Alex.Kovalenko@verizon.net> In-Reply-To: <40911C45.8090909@fer.hr> References: <20040429111355.7eb83170.Alex.Kovalenko@verizon.net> <40911C45.8090909@fer.hr>
next in thread | previous in thread | raw e-mail | index | archive | help
On Thu, 29 Apr 2004 17:16:21 +0200 Ivan Voras <ivoras@fer.hr> wrote: > Alexandre Sunny Kovalenko wrote: > > > Alternatively, if there is a need to test a patch or time a program > > in this environment, it could easily be accomplished -- it is a non- > > production box. > > Could you turn off Hyperthreading and report any changes? > > > _______________________________________________ > freebsd-current@freebsd.org mailing list > http://lists.freebsd.org/mailman/listinfo/freebsd-current > To unsubscribe, send any mail to "freebsd-current-unsubscribe@freebsd.org" I have updated my system to the latest CTM mailing on Friday (4/30) morning (EST). This includes version 1.176 of kern_thread.c. I have turned off hyperthreading and got result, which muddles me rather -- try as I might, with -lpthread I could not get past 50% CPU utilization (as reported by 'top') and number of computations per second matching that (number is very close to that for running single-threaded). Top report with thread view enabled shows that both KSE are bound to CPU 0 (if I am interpreting C column properly). Program, built with -lthr, maxed CPU for two threads and more with consistent increase in number of computations. Machine is dual 2.4GHz Xeon, HT capable. When I have enabled Hyperthreading again, the main impact of the update seems to be that 'top' reports 100% CPU utilization with either -lthr or -lpthread for four and more threads, however number of computations per second for program built with -lpthread is about 30% lower then same for program built with -lthr (~270,000 for -lpthread vs. ~370,000 for -lthr). Results are consistent across short runs (~1 minute per run) and long runs (~1 hr per run). Any suggestions, including RTFM (with pointer to FM) would be appreciated greatly. Again, if someone needs something tested on this box, let me know. Oh, yes -- last round of tests was round with SCHED_ULE only. -- Alexandre "Sunny" Kovalenko.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20040501122558.6139dec1.Alex.Kovalenko>