Open MPI logo

Open MPI User's Mailing List Archives

  |   Home   |   Support   |   FAQ   |   all Open MPI User's mailing list

Subject: Re: [OMPI users] bindings not reported in openmpi-1.9a1r27359
From: Ralph Castain (rhc_at_[hidden])
Date: 2012-09-24 10:57:42


The warning is telling you that we don't see libnuma on that machine, and therefore we cannot bind memory to your local process. You can turn the warning off by adding

-mca hwloc_base_mem_bind_failure_action silent

As for why you get different outputs, I honestly don't know - I can't replicate that behavior (and I've now tried on multiple installations), which makes me suspicious that it has something to do with your setup. One thing you could check is that your path and ld_library_path on linpc1 is pointing to the right version of OMPI when remotely invoked - i.e., that the orted being started on linpc1 when you execute mpirun on tyr is in fact the orted from the desired version of OMPI. It looks to me like maybe that isn't correct.

On Sep 24, 2012, at 5:04 AM, Siegmar Gross <Siegmar.Gross_at_[hidden]> wrote:

> Hi,
>
>> Hmmm...works fine for me on Linux x86_64
>>
>> [rhc_at_bend001 svn-trunk]$ mpirun -n 4 -host bend002 -report-bindings -map-by core -bind-to core date
>> [bend002:04730] MCW rank 3 bound to socket 1[core 2[hwt 0-1]]:[../../../BB/../..]
>> [bend002:04730] MCW rank 0 bound to socket 0[core 0[hwt 0-1]]:[BB/../../../../..]
>> [bend002:04730] MCW rank 1 bound to socket 0[core 1[hwt 0-1]]:[../BB/../../../..]
>> [bend002:04730] MCW rank 2 bound to socket 0[core 2[hwt 0-1]]:[../../BB/../../..]
>> Sun Sep 23 10:39:46 MDT 2012
>> Sun Sep 23 10:39:46 MDT 2012
>> Sun Sep 23 10:39:46 MDT 2012
>> Sun Sep 23 10:39:46 MDT 2012
>>
>> This is with 1.9a1r27360
>
> I get a warning on my Linux machine (Sun Ultra 40, 2 dual-core AMD
> Opteron 280 processors, openSuSE Linux 12.1) and the output depends
> on the location where I run the command.
>
> tyr fd1026 149 ssh linpc1 mpicc -showme:version
> mpicc: Open MPI 1.9a1r27361 (Language: C)
>
> tyr fd1026 150 mpirun -np 4 -host linpc1 -report-bindings -map-by core -bind-to core date
> --------------------------------------------------------------------------
> WARNING: a request was made to bind a process. While the system
> supports binding the process itself, at least one node does NOT
> support binding memory to the process location.
>
> Node: linpc1
>
> This is a warning only; your job will continue, though performance may
> be degraded.
> --------------------------------------------------------------------------
> Mon Sep 24 13:52:18 CEST 2012
> Mon Sep 24 13:52:18 CEST 2012
> Mon Sep 24 13:52:18 CEST 2012
> Mon Sep 24 13:52:18 CEST 2012
>
> tyr fd1026 151 ssh linpc1
> linpc1 fd1026 96 mpirun -np 4 -host linpc1 -report-bindings -map-by core -bind-to core date
> --------------------------------------------------------------------------
> WARNING: a request was made to bind a process. While the system
> supports binding the process itself, at least one node does NOT
> support binding memory to the process location.
>
> Node: linpc1
>
> This is a warning only; your job will continue, though performance may
> be degraded.
> --------------------------------------------------------------------------
> Mon Sep 24 13:53:30 CEST 2012
> Mon Sep 24 13:53:30 CEST 2012
> Mon Sep 24 13:53:30 CEST 2012
> [linpc1:21580] MCW rank 2 bound to socket 1[core 0[hwt 0]]: [./.][B/.]
> [linpc1:21580] MCW rank 3 bound to socket 1[core 1[hwt 0]]: [./.][./B]
> [linpc1:21580] MCW rank 0 bound to socket 0[core 0[hwt 0]]: [B/.][./.]
> [linpc1:21580] MCW rank 1 bound to socket 0[core 1[hwt 0]]: [./B][./.]
> Mon Sep 24 13:53:30 CEST 2012
>
>
> Any suggestions why I get a warning and why I get different outputs?
> Thank you very much for any help in advance. Hopefully I'm not annoying
> you with all my emails. Please let me know if it is too much.
>
>
> Kind regards
>
> Siegmar
>
>
>> On Sep 23, 2012, at 3:32 AM, Siegmar Gross <Siegmar.Gross_at_[hidden]> wrote:
>>
>>> Hi,
>>>
>>> yesterday I installed openmpi-1.9a1r27359 and it doesn't report process
>>> bindings on Solaris 10 x86_64 and Linux x86_64. As I showed in my previous
>>> email for openmpi-1.7, it worked for openmpi-1.6.2.
>>>
>>> tyr fd1026 114 mpiexec -np 2 -host sunpc0 -report-bindings \
>>> -map-by core -bind-to core date
>>> Sun Sep 23 12:29:25 CEST 2012
>>> Sun Sep 23 12:29:25 CEST 2012
>>>
>>> tyr fd1026 115 mpiexec -np 2 -host linpc0 -report-bindings \
>>> -map-by core -bind-to core date
>>> --------------------------------------------------------------------------
>>> WARNING: a request was made to bind a process. While the system
>>> supports binding the process itself, at least one node does NOT
>>> support binding memory to the process location.
>>>
>>> Node: linpc0
>>>
>>> This is a warning only; your job will continue, though performance may
>>> be degraded.
>>> --------------------------------------------------------------------------
>>> Sun Sep 23 12:29:36 CEST 2012
>>> Sun Sep 23 12:29:36 CEST 2012
>>>
>>> tyr fd1026 116 mpiexec -np 2 -host tyr -report-bindings \
>>> -map-by core -bind-to core date
>>> [tyr.informatik.hs-fulda.de:24039] MCW rank 0 bound to socket 0[core 0[hwt 0]]: [B][.]
>>> [tyr.informatik.hs-fulda.de:24039] MCW rank 1 bound to socket 1[core 1[hwt 0]]: [.][B]
>>> Sun Sep 23 12:29:48 CEST 2012
>>> Sun Sep 23 12:29:48 CEST 2012
>>>
>>>
>>> Kind regards
>>>
>>> Siegmar
>>>
>>> _______________________________________________
>>> users mailing list
>>> users_at_[hidden]
>>> http://www.open-mpi.org/mailman/listinfo.cgi/users
>>
>>
>