[mpich-discuss] issue setting process-core mapping with hydra

sreeram potluri potluri at cse.ohio-state.edu
Wed Nov 10 07:49:32 CST 2010


Pavan,

While configuring MPICH2, I had used a default configure, if this
information is helpful.

Sreeram Potluri

On Wed, Nov 10, 2010 at 8:42 AM, sreeram potluri <potluri at cse.ohio-state.edu
> wrote:

> Pavan,
>
> Attached is some information about the machines I am trying this on.
>
> We have SLURM (2.1.11) installed on this cluster. Do you know of any
> interactions that this could have?
>
> Thank you
> Sreeram Potluri
>
> On Wed, Nov 10, 2010 at 8:29 AM, Pavan Balaji <balaji at mcs.anl.gov> wrote:
>
>> Sreeram,
>>
>> I'm not able to reproduce this on any of our platforms with trunk, so it
>> looks like it's something specific to your hardware. What system is this
>> (OS, /proc/cpuinfo, whatever else might be relevant)?
>>
>>  -- Pavan
>>
>>
>> On 11/09/2010 05:23 PM, sreeram potluri wrote:
>>
>>> Hi,
>>>
>>> I am observing an issue when specifying process-core binding using
>>> mpiexex.hydra. I am using the trunk version of MPICH2. When I use the
>>> following command
>>>
>>> ../bin/mpiexec.hydra -binding user:0,1 -n 2 -f hosts ./hellow
>>>
>>> it prints out the following error. However, the tests complete fine. I
>>> assume it uses some preset binding.
>>>
>>>
>>> ----------------------------------------------------------------------------------------------------
>>> Restrict topology cpusets to existing PU and NODE objects
>>> Propagate offline and disallowed cpus down and up
>>>
>>> ----------------------------------------------------------------------------------------------------
>>>
>>> This error does not happen with the version of hydra available here :
>>>
>>>
>>> http://www.mcs.anl.gov/research/projects/mpich2/downloads/tarballs/1.3/hydra-1.3.tar.gz
>>>
>>> My hosts file is as follows:
>>>
>>> [potluri at wci-login examples]$ cat hosts
>>> wci37
>>> wci37
>>>
>>> Please suggest if I am missing something here. Thanks in advance.
>>>
>>> Regards
>>> Sreeram Potluri
>>>
>>>
>>>
>>> _______________________________________________
>>> mpich-discuss mailing list
>>> mpich-discuss at mcs.anl.gov
>>> https://lists.mcs.anl.gov/mailman/listinfo/mpich-discuss
>>>
>>
>> --
>> Pavan Balaji
>> http://www.mcs.anl.gov/~balaji
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/mpich-discuss/attachments/20101110/0a12e49f/attachment.htm>


More information about the mpich-discuss mailing list