[GLLUG] running R on multiple processors with LAM/MPI

Sean O'Malley picasso at madflower.com
Mon Jun 16 13:49:04 EDT 2008

I set up R a while ago, it wasn't that bad. I didnt end up using it, as my
database got too big and slow for the db server machine so i killed the
project until I got a faster machine with bigger/faster disk.. which
hasnt been in the budget yet.. :)

The configuration should be very similar to the directions given for the
Mac as far as configuration. You should be able to loosely follow the
steps and get it running. The big difference will be a difference where
the software is installed. The command line stuff will most likely work
the same. R should work very similar except the GUI which on the mac will
look significantly different since they ported it to cocoa (native mac)

It appears Ubuntu has the packages you need already in their package
format so you shouldnt have to recompile or tweak any software to get it
to work.

and various other related packages

And look at the bright side. once you are done, and have a basic
understanding of this, you can scale it to other computers on your local
network. :)

On Fri, 13 Jun 2008, Qiu Wang wrote:

> Hello, GLLUG-ers
> Hope you are enjoy the nice summer!
> My name is Qiu Wang and I am a student at MSU. I just started using linux
> system last week and I love it. Currently, I am using a open source /GNU
> statistical package R <http://www.r-project.org/about.html> for my project.
> Is anyone using R <http://www.r-project.org/> for parallel computation on a
> linux cluster (e.g., msu's HPC)? I was told that someone once gave a talk
> on R to GLLUG 2-3 yeas ago. Does any one know if that talk touched parallel
> compuation? People at HPC (especically, Ed) are very supportive and will
> update R and install required packages (e.g.,
> Rmpi<http://www.stats.uwo.ca/faculty/yu/Rmpi/>). I really appreciated
> their help and would like to hear some user-side
> expereince from this list.
> Here is another issue the list might be interested with.
> I just got a quad dell inspiron desktop and started to use ubuntu. Oh, it is
> such a chellenge,  huge for me, but I had some fun. One thing I need to
> figure out is to run R on all 4 processors on a single machine. Right now it
> only uses one precessor (R randomly picks one of  the four CPUs to run the
> code. ). It is a waste. I found that running R on a multiprocessor single
> machine has been done on a quad-Mac
> <http://macresearch.org/statistics_supercomputing_on_a_mac_with_r_and_lam_mpi>.
> How can it be done on my Dell quad machine? I have no clue where I should
> start. Any thoughts and ideas will be appreiciated.
> --
> Qiu Wang
> Measurement and Quantitative Methods
> College of Education
> Michigan State University
> ------
-------------- next part --------------
linux-user mailing list
linux-user at egr.msu.edu

More information about the linux-user mailing list