Re: [Rd] Rmpi and OpenMPI ?

From: Dirk Eddelbuettel <edd_at_debian.org>
Date: Wed, 27 Jun 2007 06:20:55 -0500

On 27 June 2007 at 12:21, Markus Schmidberger wrote:
| Hello,
|
| I found your discussion about Rmpi and openMpi at the R-devels archive.
|
| Did you managed to run Rmpi on openMPI (completely)?

The good news is that a few of us have formed a new maintainaer team for OpenMPI in Debian. New and decently maintained packages have been uploaded to the NEW queue. They should reach Debian unstable after the usual inspection delay for 'new' packages (triggered here as we renamed some sub-packages).

As for Rmpi, I have not had time to revisit building it against OpenMPI. It does work ...  

| We try to do this installation at a OS debian with the "openmpi-common"
| package, but it is not working. For example there is no "mpi.h"

... provided you undo some LAM/MPI-ism in Rmpi. That is relatively simple.

Rmpi will need a little bit of work for OpenMPI as Rmpi seems to use some lam-only features.

I hope I will get there, but it will probably be a matter of a few months not days. I need to sort out how to do it in a way to preserve the ability to build with with lam, or fork a new 'ROpenMPI' off it. We'll see. I will certainly report back with progress reports.

Anybody interested in helping port Rmpi to OpenMPI, please contact me off-line.

Regards, Dirk

-- 
Hell, there are no rules here - we're trying to accomplish something. 
                                                  -- Thomas A. Edison

______________________________________________
R-devel_at_r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
Received on Wed 27 Jun 2007 - 11:23:58 GMT

Archive maintained by Robert King, hosted by the discipline of statistics at the University of Newcastle, Australia.
Archive generated by hypermail 2.2.0, at Wed 27 Jun 2007 - 14:35:37 GMT.

Mailing list information is available at https://stat.ethz.ch/mailman/listinfo/r-devel. Please read the posting guide before posting to the list.