Open MPI logo

Open MPI User's Mailing List Archives

  |   Home   |   Support   |   FAQ   |  

This web mail archive is frozen.

This page is part of a frozen web archive of this mailing list.

You can still navigate around this archive, but know that no new mails have been added to it since July of 2016.

Click here to be taken to the new web archives of this list; it includes all the mails that are in this frozen archive plus all new mails that have been sent to the list since it was migrated to the new archives.

From: George Bosilca (bosilca_at_[hidden])
Date: 2006-11-07 23:35:44

The quick answers are yes and nothing. Open MPI seamlessly support
shared memory communications when we detect that 2 processes are on
the same node. In fact Open MPI can use different communication
methods (read networks) between the processes of the same application.

Please read our FAQ for more information on how to specify which
network you want to be used and which not.


On Nov 7, 2006, at 11:07 PM, Chevchenkovic Chevchenkovic wrote:

> Hi,
> Thanks for the reply,
> A few Additional questions,
> 1. Does OpenMPI has the optimisations required to ensure that when
> send/recv is called between 2 ranks on the same node, the shared
> memory kind of methods should be used?
> 2. If a programmer wants to implement such a logic(optimisations
> for local send/recv) , what part of the code should he modify from
> the openMPI stack ?
> regards,
> -Chev
> _______________________________________________
> users mailing list
> users_at_[hidden]