Hi Terry,

        I feel hierarchical collectives are slower compare to tuned one. I had done some benchmark in the past specific to collectives, and this is what i feel based on my observation.

Regards

Neeraj Chourasia (MTS)
Computational Research Laboratories Ltd.
(A wholly Owned Subsidiary of TATA SONS Ltd)
B-101, ICC Trade Towers, Senapati Bapat Road
Pune 411016 (Mah) INDIA
(O) +91-20-6620 9863  (Fax) +91-20-6620 9862
M: +91.9225520634


Terry Dontje <Terry.Dontje@sun.com>
Sent by: users-bounces@open-mpi.org

08/07/2009 04:35 PM
Please respond to
Open MPI Users <users@open-mpi.org>
To
users@open-mpi.org
cc
Subject
Re: [OMPI users] Performance question about OpenMPI and MVAPICH2        on        IB





Craig,

Did your affinity script bind the processes per socket or linearly to cores.  If the former you'll want to look at using rankfiles and place the ranks based on sockets.  TWe have found this especially useful if you are not running fully subscribed on your machines.

Also, if you think the main issue is collectives performance you may want to try using the hierarchical and SM collectives.  However, be forewarned we are right now trying to pound out some errors with these modules.  To enable them you add the following parameters "--mca coll_hierarch_priority 100 --mca coll_sm_priority 100".  We would be very interested in any results you get (failures, improvements, non-improvements).

thanks,

--td

> Message: 4
> Date: Thu, 06 Aug 2009 17:03:08 -0600
> From: Craig Tierney <Craig.Tierney@noaa.gov>
> Subject: Re: [OMPI users] Performance question about OpenMPI and
>                  MVAPICH2 on                 IB
> To: Open MPI Users <users@open-mpi.org>
> Message-ID: <4A7B612C.8070501@noaa.gov>
> Content-Type: text/plain; charset=ISO-8859-1
>
> A followup....
>
> Part of problem was affinity.  I had written a script to do processor
> and memory affinity (which works fine with MVAPICH2).  It is an
> idea that I got from TACC.  However, the script didn't seem to
> work correctly with OpenMPI (or I still have bugs).
>
> Setting --mca mpi_paffinity_alone 1 made things better.  However,
> the performance is still not as good:
>
> Cores   Mvapich2    Openmpi
> ---------------------------
>    8      17.3        17.3
>   16      31.7        31.5
>   32      62.9        62.8
>   64     110.8       108.0
>  128     219.2       201.4
>  256     384.5       342.7
>  512     687.2       537.6
>
> The performance number is GFlops (so larger is better).
>
> The first few numbers show that the executable is the right
> speed.  I verified that IB is being used by using OMB and
> checking latency and bandwidth.  Those numbers are what I
> expect (3GB/s, 1.5mu/s for QDR).
>
> However, the Openmpi version is not scaling as well.  Any
> ideas on why that might be the case?
>
> Thanks,
> Craig

_______________________________________________
users mailing list
users@open-mpi.org
http://www.open-mpi.org/mailman/listinfo.cgi/users

=====-----=====-----===== Notice: The information contained in this e-mail message and/or attachments to it may contain confidential or privileged information. If you are not the intended recipient, any dissemination, use, review, distribution, printing or copying of the information contained in this e-mail message and/or attachments to it are strictly prohibited. If you have received this communication in error, please notify us by reply e-mail or telephone and immediately and permanently delete the message and any attachments. Internet communications cannot be guaranteed to be timely, secure, error or virus-free. The sender does not accept liability for any errors or omissions.Thank you =====-----=====-----=====