Re: [hwloc-users] Zero cache line size on Power7?

2013-02-27 Thread Brice Goglin
I think I've seen cases where the device-tree contains 0 for such line sizes. I guess we should document that the line size is 0 means unknown. Can you send the tarball generated by hwloc-gather-topology ? (send it only to, in a private email) Brice Le 27/02/2013 23:11, Erik Schnetter a écrit

[hwloc-users] Zero cache line size on Power7?

2013-02-27 Thread Erik Schnetter
I am running hwloc 1.7a1r5312 on a Power7 architecture. I find there a level 2 cache with a cacheline size of 0. Is this to be expected? The documentation doesn't say that determining the cacheline size may fail. I query the cache parameters from my application with these results: Cache (unknown

Re: [hwloc-users] rebind subarray

2013-02-27 Thread Aulwes, Rob
I'll give that a try. Thanks Brice! From: Brice Goglin mailto:brice.gog...@inria.fr>> Reply-To: Hardware locality user list mailto:hwloc-us...@open-mpi.org>> List-Post: hwloc-users@lists.open-mpi.org Date: Wed, 27 Feb 2013 22:08:48 +0100 To: mailto:hwloc-us...@open-mpi.org>> Subject: Re: [hwloc-

Re: [hwloc-users] rebind subarray

2013-02-27 Thread Brice Goglin
Hello Rob, That's exactly what set_area_membind() was meant for. That said, you can only move entire pages (usually 4kbytes) to another domain. So you can only move half of your array if you array is larger than a single page. You likely need something like this: [... initialize the hwloc top

[hwloc-users] rebind subarray

2013-02-27 Thread Aulwes, Rob
Hi, I'm running OpenMP tests and I see a slow down going from 8 to 16 threads. I'm suspecting that the arrays I'm using are all allocated on a single NUMA domain, so that threads 9-15 encounter a slowdown when accessing those arrays, because they have to cross domains. Suppose that I've alloc