Perfect; thanks!  Commit coming shortly.

On Jan 6, 2011, at 1:56 AM, Samuel Thibault wrote:

> Jeff Squyres, le Thu 06 Jan 2011 03:39:28 +0100, a écrit :
>> * \note HWLOC_MEMBIND_STRICT is only supported with this function on
>> * some systems.  If it is specified on an unsupported system, -1 is
>> * returned and errno is set to ENOSYS.
> 
> Err, not exactly: there is currently no system where hwloc_get_membind
> is supported without STRICT but not supported with STRICT.
> 
>> * If HWLOC_MEMBIND_STRICT is not specified, or if all the process'
>> * (or thread's) pages have the same memory binding policy,
> 
> policy _and nodeset_ (i.e. actually just one node).
> 
>> \p nodeset
>> * is cleared.  Each page is then examined and has the bit set in \p
>> * nodeset corresponding to the NUMA node where it is located.  The
>> * resulting \p nodeset represents the set of NUMA nodes containing
>> * all the pages in the process (or thread).
> 
> Samudl
> _______________________________________________
> hwloc-devel mailing list
> hwloc-de...@open-mpi.org
> http://www.open-mpi.org/mailman/listinfo.cgi/hwloc-devel


-- 
Jeff Squyres
jsquy...@cisco.com
For corporate legal information go to:
http://www.cisco.com/web/about/doing_business/legal/cri/


Reply via email to