Open MPI logo

Hardware Locality Users' Mailing List Archives

  |   Home   |   Support   |   FAQ   |   all Hardware Locality Users mailing list

Subject: Re: [hwloc-users] Istopo question
From: Dan Eaton (daniel.eaton_at_[hidden])
Date: 2010-01-28 14:09:34


> Does Linux announce things correctly in /proc/cpuinfo in that case?
Yes.

I've now also run the two versions (stable + patched nightly release) on a dual-socket AMD Opteron 2216 machine. No surprise, the x86 path doesn't return any cache info :) but gets the hierarchy right.

The stable version (using the OS path I'm guessing?) does correctly print the hierarchy though.

Also, consistent with Brice's assertion, libnuma doesn't work at detecting NUMA (it reports none) on either of the NUMA machines I've tested it on (Nehalem, Opteron)... I hadn't tried that before, and just assumed that whenever I got to trying it out it would work out of the box on whatever Linux version that supported it.

Ahh Linux, when is that ever the case? ;)

Kind regards,
Dan

-----Original Message-----
From: hwloc-users-bounces_at_[hidden] [mailto:hwloc-users-bounces_at_[hidden]] On Behalf Of Samuel Thibault
Sent: Thursday, January 28, 2010 12:05 PM
To: Hardware locality user list
Subject: Re: [hwloc-users] Istopo question

Dan Eaton, le Thu 28 Jan 2010 11:57:08 -0700, a écrit :
> I see the comment TODO: AMD version in topology-x86.c ... do you have any idea when your module might support that?

When I get the time to parse AMD's documentation :)

> I've attached the debug output of lstopo with HT enabled (looks good!).

Good :)

Does Linux announce things correctly in /proc/cpuinfo in that case?

Samuel
_______________________________________________
hwloc-users mailing list
hwloc-users_at_[hidden]
http://www.open-mpi.org/mailman/listinfo.cgi/hwloc-users