On Thu, Nov 25, 2004 at 02:11:12PM +0100, Laurent Valdes
wrote:> Hi folks,
>
> is it possible to read more things about R behavior in multiprocessor /
> multihost environment ?
> Is there any distributed computation project associated to it ?
Sure. I'd start with the paper 'Simple Parallel Statistical Computing in
R' by
Tony Rossini, Luke Tierney and Na Li:
http://www.bepress.com/uwbiostat/paper193/
and the references therein. The underlying software (SNOW, Rmpi, Rpvm, ...)
is readily available on CRAN, but you may have to do some work to get the
communications libraries needed (lam, mpi, pvm, ...) built.
One way to get a head start on deployment is to grab a Quantian dvd image:
http://dirk.eddelbuettel.com/quantian
which not only contains R alongside Snow, Rmpi, Rpvm for explicit parallelism
using message parsing, but also provides a ready-to-use openMosix clustering
environment where you can boot nodes 2, 3, ... right over the net off the
first machine booted from the dvd.
Hth, Dirk
PS I amended your subject line as you really asked about multiprocessor and
mulithost rather than multithreading (which is typically inside one cpu, and
which R doesn't do, see http://developer.r-project.org).
--
If your hair is standing up, then you are in extreme danger.
-- http://www.usafa.af.mil/dfp/cockpit-phys/fp1ex3.htm