[BioC] Install Rmpi on Fedora with mpich2 installed.

Li Long lilong at isb-sib.ch
Wed Dec 21 17:06:38 CET 2005

Again, this is from some work a while back.  I haven't looked very closely
at Rmpi recently, and offered the following just for your consideration...

I did modify the configure.in to use environment var MPI_INSTALL so that
I could reference $MPI_INSTALL/lib, $MPI_INSTALL/bin and $MPI_INSTALL/man
in $LD_LIBRARY_PATH, $PATH, $MANPATH, respectively.  For example, if you
use bash in your Linux, you can add lines like
    export MPI_INSTALL=...wherever you install your LAMMPI...
    export PATH=$MPI_INSTALL/bin:$PATH
to your .bashrc file.

This makes sure that your LAMMPI installation preceeds all other MPI
installation(s) and is used consistantly.  When I installed this package
(a long time ago), the configure.in is quite vulnerable if you have more
than one MPI installations around (such as, include mpi.h from one MPI
installation and load .a from another, or something like that).

You should take a closer look at the latest package w/ its doc to decide
how to proceed.

Good luck,


> Thank you very much, Li! With my limited knowledge about MPI and linux, I
> don't know how to set up the MPI_INSTALL, could you please give me more
> details or some reference about this?
> Thank you again!
> Bin
> -----Original Message-----
> From: Li Long [mailto:lilong at isb-sib.ch]
> Sent: Wed 12/21/2005 5:31 AM
> To: Ye, Bin
> Cc: bioconductor at stat.math.ethz.ch
> Subject: Re: [BioC] Install Rmpi on Fedora with mpich2 installed.
> Hi,
> This is probably something you could consider, might be a bit out-dated...
> (I haven't done it recently)
> I think
> (1) the current Rmpi wants LAMMPI, I had a lot of trouble getting it to
>     work with MPICH more than a year ago, and finally gave up on it;
> (2) you could try setting MPI_INSTALL in the environment (I had it) as
> well,
>     the install script might wants it;
> (3) it's not a problem to have both MPICH and LAMMPI on the same system,
> as
>     far as you could keep them separated from each other (you could
> double-
>     check the scripts to make sure)
> Good luck.
> Li

More information about the Bioconductor mailing list