[R] Running R under Sun Grid Engine with OpenMPI tight integration

Markus Schmidberger schmidb at ibe.med.uni-muenchen.de
Thu Oct 9 18:50:24 CEST 2008


-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi,

actually there is no perfect way to use R one multiprocessors. There are
two options:

* using mpi: This means starting several R sessions and using MPI for
the communication between the R sessions. Task distribution, fail over
solutions, ... have to be implemented by your self. MPI was developed
for multicomputer systems. So it is not very efficient to use it at a
multiprocessor machine. But it works!

* using multiThreads or openMP: There are packages from Luke Tierney
using multi threading or openMP for basic R math functions.
http://www.cs.uiowa.edu/~luke/R/experimental/
This are especially for multiprocessors designed math functions. You can
achieve nice speedups. But unfortunately this is just working for the
basic math functions.

Have a look to the useR2008 conference and Dirks Tutorium. There you can
find a lot of information!

Markus




Chi Chan wrote:
> On Thu, Oct 9, 2008 at 9:02 AM, Dirk Eddelbuettel <edd at debian.org> wrote:
>> You could have a look at the slides from my UseR! 2008 tutorial [1] on
>> 'high-performance computing with R' which covered Open MPI / Rmpi and
>> includes examples.
>>
>> It may just be that you are expecting something that cannot be done.  There
>> is no 'R session on 8 processors'.  R is still single-threaded, and neither
>> Open MPI nor the Sun Grid Engine changes that.  But what you can do is to
>> start R on a node in your cluster and then use Open MPI to distribute load
>> across the 8 processors.  There are examples in my slides.
> 
> So R uses MPI for task distribution only? That sounds like not very
> fine grained parallelism.
> 
> Can I use rsh to start remote R instead of MPI?
> 
> --Chi
> 
> 
>> Hope this helps, Dirk
>>
>> [1] http://dirk.eddelbuettel.com/presentations.html
>>
>> --
>> Three out of two people have difficulties with fractions.
>>
>> ______________________________________________
>> R-help at r-project.org mailing list
>> https://stat.ethz.ch/mailman/listinfo/r-help
>> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
>> and provide commented, minimal, self-contained, reproducible code.
>>
> 
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.


- --
Dipl.-Tech. Math. Markus Schmidberger

Ludwig-Maximilians-Universität München
IBE - Institut für medizinische Informationsverarbeitung,
Biometrie und Epidemiologie
Marchioninistr. 15, D-81377 Muenchen
URL: http://www.ibe.med.uni-muenchen.de
Mail: Markus.Schmidberger [at] ibe.med.uni-muenchen.de
Tel: +49 (089) 7095 - 4599
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.4-svn0 (GNU/Linux)
Comment: Using GnuPG with SUSE - http://enigmail.mozdev.org

iD8DBQFI7jZPwxqyUctZsggRAm9YAJ0VQ3JwzGfYp7ZmwbOJi7HxCTD/rQCglSUP
eKJzO3NelDa55vMD3b+lp3o=
=5CyM
-----END PGP SIGNATURE-----



More information about the R-help mailing list