Date: Sun, 22 Mar 2015 08:28:07 -0700 From: Harrison Grundy <harrison.grundy@astrodoggroup.com> To: freebsd-arch@freebsd.org Subject: Re: RFC: Simplfying hyperthreading distinctions Message-ID: <550EDF87.2070101@astrodoggroup.com> In-Reply-To: <20150322100744.5b390591@ernst.home> References: <1640664.8z9mx3EOQs@ralph.baldwin.cx> <54FA1180.3080605@astrodoggroup.com> <1526311.uylCbgv5VB@ralph.baldwin.cx> <20150320123823.GA49621@zxy.spb.ru> <550DC564.5020802@freebsd.org> <20150321214336.334eaea5@nonamehost.local> <20150322100744.5b390591@ernst.home>
next in thread | previous in thread | raw e-mail | index | archive | help
On 03/22/15 02:07, Gary Jennejohn wrote: > On Sat, 21 Mar 2015 21:43:35 +0200 > Ivan Klymenko <fidaj@ukr.net> wrote: > >> __ Sat, 21 Mar 2015 12:24:20 -0700 >> Alfred Perlstein <alfred@freebsd.org> __________: >> >>> John, >>> >>> Just a quick note on this, hopefully it's not too off-topic... >>> >>> We need to detect if HTT or SMT is enabled, right now there are no >>> sysctl nodes to detect this and instead we have to parse xml out of >>> the scheduler... >>> >>> Does it make sense to have a basic sysctl tree for this? >>> >>> hw.cpu.threading.smt=0 >>> hw.cpu.threading.htt=0 >>> >>> or something? >>> >> >> I am sorry that I interfere >> Why then not use kern.smp.topology for this purpose? >> > > Because it's only present in the ULE scheduler, BSD doesn't > have it. > Extracting topology detection out of the scheduler and finalizing that API may be a good idea anyway, given the NUMA work going on elsewhere. Things besides the scheduler can benefit from the information, and it'll simplify the scheduler a bit. Any thoughts on handling it that way? --- Harrison
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?550EDF87.2070101>