Open MPI logo

Open MPI User's Mailing List Archives

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

Subject: Re: [OMPI users] Is there an mca parameter equivalent to -bind-to-core?
From: Gus Correa (gus_at_[hidden])
Date: 2011-03-24 21:38:37


Jeff Squyres wrote:
> On Mar 23, 2011, at 4:20 PM, Gus Correa wrote:
>
>> However, now when I do "ompi_info -a",
>> the output shows the non-default value 1 twice in a row,
>> then later it shows the default value 0 again!
>
> It's because we wanted to confuse you!
>
> ;-)
>
> Sorry about that; this is a legitimate bug.
> I've fixed it on the trunk and submitted CMRs for both v1.4 and v1.5.
> I believe that the correct value is actually being used, despite what
> your ompi_info is saying.
>
> (I can explain further, if you care)
>

Thank you Jeff, Ralph, Eugene

As far as I can tell by observing [Open]MPI processes
with top (with 'f' & 'J'), and by using the mpiexec --report-bindings
switch, each process is indeed bound to a single core.
This binding happens whether I request it with the
mpiexec --bind-to-core switch,
or if I put the line 'opal_paffinity_alone = 1'
in the openmpi-mca-params.conf file,
exactly as Ralph said it would be.

Hence, the two contradictory values of mpi_paffinity_alone
(1 and 0) reported by 'ompi_info -all', really seem
to have been nothing but a cheap plot by Jeff
to emulate the conspiracy against my sanity
long held by the cosmogenic flipping bits and the flaky logic gates.

Gullible folks like me have always been reluctant
to learn this lesson.
In MPI, as in foreign policy, it seems to be
as that late president used to say:
"Trust, but verify."

Cheers,
Gus Correa