On Tue, 24 Apr 2007, Jeff Squyres wrote:
> On Apr 24, 2007, at 10:03 AM, Bert Wesarg wrote:
>
> > Finaly a found a machine with HTT enabled (btw: what is 'HMT'?)
>
> "HMT" = Jeff is having a bad morning; he can't type, read error
> messages, and should strongly consider having significant amounts of
> caffeine before replying any further.
Hmm, I found "Hardware Multi Threading"
>
> > but without the node directory
> >
> > this reminds we, that we need some sort of smart fallback, if some
> > directories or files are missing
>
> Agreed.
>
> Right now, if the (socket,core) stuff is unavailable, you'll get a
> graceful failure from all the topology functions.
>
> Is it possible to have less than all of the (socket,core)
> information, the node ID information, and HTT information? Do we
> need to make the have_topo_information() function a bit more fine-
> grained? (i.e., indicate which, if any, of the topology information
> is available) And/or make *some* of the functions fail (gracefully)
> if the requested information is unavailable?
I think we should emulate missing information:
* no node info: put all processor ids into node0
* no thread info: only one thread
:
:
and so on
>
> I do not follow the history of the Linux kernel enough to know when
> this stuff was introduced, nor what distros are doing to back port, etc.
For me, this stuff is also new, and I follow the kernel development with
an half eye. And I have search for this topology information one in a
while.
> --
> Jeff Squyres
> Cisco Systems
>
> _______________________________________________
> plpa-users mailing list
> plpa-users_at_[hidden]
> http://www.open-mpi.org/mailman/listinfo.cgi/plpa-users
>
|
![[Flag]](../../../../../community/mirrors/flag-us.gif){kind=small}