Date: Fri, 23 Feb 2007 08:35:58 -0800 From: "Kunze, Aaron" <aaron.kunze@intel.com> To: "Daniel Eischen" <deischen@freebsd.org>, "Robert Watson" <rwatson@freebsd.org> Cc: atmblr@gmail.com, freebsd-smp@freebsd.org Subject: RE: Setting CPU affinity to process( Freebsd smp kernel) Message-ID: <07DDDFCFB8BE0A43BCA52E743373DBDC03102190@orsmsx416.amr.corp.intel.com> In-Reply-To: <Pine.GSO.4.64.0702231107580.29991@sea.ntplx.net> References: <07DDDFCFB8BE0A43BCA52E743373DBDC030C5D5A@orsmsx416.amr.corp.intel.com> <20070223151158.Q88189@fledge.watson.org> <Pine.GSO.4.64.0702231107580.29991@sea.ntplx.net>
next in thread | previous in thread | raw e-mail | index | archive | help
Thanks for the info. The Linux equivalent would be sched_setaffinity which takes a bitmask as input, allowing the user to define which processors will run a particular thread. Here's a link: http://ibm5.ma.utexas.edu/cgi-bin/man-cgi?sched_setaffinity+2 > > There's a potential for=20 > > conflict between the kernel's use of pinning and binding for kernel=20 > > synchronization and the user space affinity model, which will be=20 Can you elaborate on this? Some of my colleagues and I are considering tackling this and would like to avoid such pitfalls, if possible. Aaron -------------------------- Aaron Kunze Advanced Visual Computing aaron.kunze@intel.com --PGP Key ID: 0x81124B7C-- --NOT SPEAKING FOR INTEL-- -------GO BOILERS!!------- =20 > -----Original Message----- > From: Daniel Eischen [mailto:deischen@freebsd.org]=20 > Sent: Friday, February 23, 2007 8:09 AM > To: Robert Watson > Cc: Kunze, Aaron; atmblr@gmail.com; freebsd-smp@freebsd.org > Subject: Re: Setting CPU affinity to process( Freebsd smp kernel) >=20 > On Fri, 23 Feb 2007, Robert Watson wrote: >=20 > > On Wed, 21 Feb 2007, Kunze, Aaron wrote: > > > >> Does anyone know if this will change any time soon? For=20 > example, is=20 > >> anyone working on exposing affinity to user-space applications via=20 > >> extensions of the pthreads interface? > >>=20 > >> Sorry to reply to such an old thread... > > > > I know of no work along these lines currently, but it's something a=20 > > lot of people would like to see happen. There's a potential for=20 > > conflict between the kernel's use of pinning and binding for kernel=20 > > synchronization and the user space affinity model, which will be=20 > > entirely avoided if done right. :-) For now, it's quite=20 > easy to add a=20 > > sysctl/syscall that allows user space to send the kernel=20 > scheduler's=20 > > notion of thread binding, but this isn't really the right=20 > approach. =20 > > As I understand it, some systems support setting CPU=20 > affinity for a thread as a set of CPUs it is willing to run on ? >=20 > I know Solaris has processor_bind(2) and pset_bind(2): >=20 > http://docs.sun.com/app/docs/doc/816-5167/6mbb2jaeu?a=3Dexpand#P >=20 > -- > DE >=20
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?07DDDFCFB8BE0A43BCA52E743373DBDC03102190>