Fabian Wein writes:
>># hwloc-bind node1:1 hwloc-ps | grep hwloc
>>13425 NUMANode:1 hwloc-ps
>
> I don't understand what you mean:
>
> opt/hwloc-1.11.1/bin/hwloc-bind
> /opt/hwloc-1.11.1/bin/hwloc-bind: nothing to do!
>
> /opt/hwloc-1.11.1/bin/hwloc-bind node1:1
> /opt/hw
On 11/03/2015 06:13 PM, Dave Love wrote:
Fabian Wein writes:
There is an old OpenFOAM installation which includes and old open-mpi,
this might
cause the trouble.
OpenFOAM should definitely be built against the system MPI (and, in
general, you should avoid bundled libraries wherever possible
Fabian Wein writes:
> There is an old OpenFOAM installation which includes and old open-mpi,
> this might
> cause the trouble.
OpenFOAM should definitely be built against the system MPI (and, in
general, you should avoid bundled libraries wherever possible IMHO).
> I also suspect that sourcing
> Am 02.11.2015 um 15:58 schrieb Dave Love :
>
> Fabian Wein writes:
>
>>> Am 30.10.2015 um 21:45 schrieb Jeff Squyres (jsquyres) :
>>>
>>> Oh, that's an interesting idea: perhaps the "bind to numa" is
>>> failing -- but perhaps "bind to socket" would work.
>>>
>>> Can you try:
>>>
>>> /opt/
Fabian Wein writes:
>> Am 30.10.2015 um 21:45 schrieb Jeff Squyres (jsquyres) :
>>
>> Oh, that's an interesting idea: perhaps the "bind to numa" is
>> failing -- but perhaps "bind to socket" would work.
>>
>> Can you try:
>>
>> /opt/openmpi-1.10.0-gcc/bin/mpiexec -bind-to numa -n 4 hostname
>>
Fabian Wein writes:
> On 10/30/2015 02:48 PM, Dave Love wrote:
>> Fabian Wein writes:
>>
>>> Is this a valid test?
>>>
>>>
>>> /opt/openmpi-1.10.0-gcc/bin/mpiexec -n 4 hostname
>>> leo
>>> leo
>>> leo
>>> leo
>>
>> So, unless you turned off the default binding -- to socket? check the
>> mpirun m
> Am 30.10.2015 um 21:45 schrieb Jeff Squyres (jsquyres) :
>
> Oh, that's an interesting idea: perhaps the "bind to numa" is failing -- but
> perhaps "bind to socket" would work.
>
> Can you try:
>
> /opt/openmpi-1.10.0-gcc/bin/mpiexec -bind-to numa -n 4 hostname
> /opt/openmpi-1.10.0-gcc/bin/
Oh, that's an interesting idea: perhaps the "bind to numa" is failing -- but
perhaps "bind to socket" would work.
Can you try:
/opt/openmpi-1.10.0-gcc/bin/mpiexec -bind-to numa -n 4 hostname
and
/opt/openmpi-1.10.0-gcc/bin/mpiexec -bind-to socket -n 4 hostname
> On Oct 30, 2015, at 12:02 P
Brice -- I know this started on the hwloc list and then bounced over here, but
we're running out of ideas.
Got any clue what is happening here? From the OMPI config logs that Fabian
sent, it looks like hwloc built with libnuma support properly...?
> On Oct 30, 2015, at 4:34 PM, Ralph Castain
I honestly have no ideas…best I can see, it looks like hwloc feels that it
cannot perform that operation and returns an error.
> On Oct 30, 2015, at 1:31 PM, Jeff Squyres (jsquyres)
> wrote:
>
> He's building and running on a single server (leo). From the configure
> output, all the numa li
He's building and running on a single server (leo). From the configure output,
all the numa libs and headers are available on this leo server.
> On Oct 30, 2015, at 11:09 AM, Ralph Castain wrote:
>
> I think Dave has probably hit the problem - that node may well not have a
> “numa” object. Y
On 10/30/2015 04:09 PM, Ralph Castain wrote:
I think Dave has probably hit the problem - that node may well not have a
“numa” object. You also might check that node “leo” has libnuma on it
I have a shared memory system, I run on the machine I build the stuff.
Fabian
On 10/30/2015 02:48 PM, Dave Love wrote:
Fabian Wein writes:
Is this a valid test?
/opt/openmpi-1.10.0-gcc/bin/mpiexec -n 4 hostname
leo
leo
leo
leo
So, unless you turned off the default binding -- to socket? check the
mpirun man page -- it worked, but the "numa" level failed. I don't kno
I think Dave has probably hit the problem - that node may well not have a
“numa” object. You also might check that node “leo” has libnuma on it
> On Oct 30, 2015, at 6:48 AM, Dave Love wrote:
>
> Fabian Wein writes:
>
>> Is this a valid test?
>>
>>
>> /opt/openmpi-1.10.0-gcc/bin/mpiexec -n
Fabian Wein writes:
> Is this a valid test?
>
>
> /opt/openmpi-1.10.0-gcc/bin/mpiexec -n 4 hostname
> leo
> leo
> leo
> leo
So, unless you turned off the default binding -- to socket? check the
mpirun man page -- it worked, but the "numa" level failed. I don't know
if that level has to exist, a
Yes, I think that test is valid. 1.10.0 is a good version to test.
Ralph: it all looks good to me, but that message is coming from ORTE. Can you
have a look?
Sent from my phone. No type good.
> On Oct 30, 2015, at 12:33 AM, Fabian Wein wrote:
>
>> On 10/30/2015 02:26 AM, Jeff Squyres (jsquy
On 10/30/2015 02:26 AM, Jeff Squyres (jsquyres) wrote:
Your Open MPI build looks good -- it seems to have found all the right libnuma
stuff during configured, etc.
The error message you're seeing indicates that the embedded hwloc is telling
Open MPI that it doesn't have binding support, which
Your Open MPI build looks good -- it seems to have found all the right libnuma
stuff during configured, etc.
The error message you're seeing indicates that the embedded hwloc is telling
Open MPI that it doesn't have binding support, which would be unusual since you
have both libnuma and libnuma
On 10/29/2015 09:29 PM, Dave Love wrote:
I guess the problem is not having hwloc/libnuma1 on the compute host, as
opposed to the build host. Test like this on the host, though the
component name may be different in 1.10:
I Have shared-memory system (4 package opteron, 8 numa nodes, 48
cores),
If your hwloc was built without libnuma support, you will get the
following error:
$ hwloc-bind --membind node:0 -- echo foo
hwloc_set_membind 0x000f failed (errno 38 Function not implemented)
Some days ago I installed hwloc and observed "libnuma memory support:
no". Then I
installed libn
"Jeff Squyres (jsquyres)" writes:
> +1
>
> If you're compiling Open MPI from source, you need the -devel package
> so that the libnuma header files are installed (and therefore Open MPI
> [i.e., the hwloc embedded in Open MPI] can include those header files
> and then compile support for libnuma)
Le 29/10/2015 21:04, Fabian Wein a écrit :
>> If you're compiling Open MPI from source, you need the -devel package
>> so that the libnuma header files are installed (and therefore Open
>> MPI [i.e., the hwloc embedded in Open MPI] can include those header
>> files and then compile support for li
If you're compiling Open MPI from source, you need the -devel package so that
the libnuma header files are installed (and therefore Open MPI [i.e., the hwloc
embedded in Open MPI] can include those header files and then compile support
for libnuma).
Can I also test the numa capability via a
Is there any hint in the configure output?
Yes. Please send all the info listed here:
http://www.open-mpi.org/community/help/
I attached the files. I built it again, this time with gcc instead of
intel
This is how I call mpiexec to make sure I get the right one
make MPIEXEC="/opt/o
On Oct 29, 2015, at 9:26 AM, Fabian Wein wrote:
>
> Is there any hint in the configure output?
Yes. Please send all the info listed here:
http://www.open-mpi.org/community/help/
--
Jeff Squyres
jsquy...@cisco.com
For corporate legal information go to:
http://www.cisco.com/web/about/doin
On 10/29/2015 03:01 PM, Ralph Castain wrote:
You need to also install libnuma-devel, I’m afraid - just has to do with how
they chose to split the library
sudo apt-get install libnuma-devel
...
E: Unable to locate package libnuma-devel
sudo apt-get install libnuma-dev
...
libnuma-dev is alread
+1
If you're compiling Open MPI from source, you need the -devel package so that
the libnuma header files are installed (and therefore Open MPI [i.e., the hwloc
embedded in Open MPI] can include those header files and then compile support
for libnuma).
> On Oct 29, 2015, at 7:01 AM, Ralph Ca
You need to also install libnuma-devel, I’m afraid - just has to do with how
they chose to split the library
> On Oct 29, 2015, at 2:01 AM, Fabian Wein wrote:
>
> Hello,
>
> I'm new to the list, maybe the issue was somewhere else already solved. I
> found a similar question in the mailing li
28 matches
Mail list logo
